Text Generation
MLX
Safetensors
qwen4_exp
apple-silicon
quantized
mixed-precision
axquant
axq
development
qwen4-exp
MXFP4
mtp
vision
conversational
4-bit precision
Instructions to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP") prompt = "Write a story about Einstein" messages = [{"role": "user", "content": prompt}] prompt = tokenizer.apply_chat_template( messages, add_generation_prompt=True ) text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP"
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "mlx-lm": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP" } ] } } }Run Pi
# Start Pi in your project directory: pi
- MLX LM
How to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Interactive chat REPL mlx_lm.chat --model "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP"
Run an OpenAI-compatible server
# Install MLX LM uv tool install mlx-lm # Start the server mlx_lm.server --model "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP" # Calling the OpenAI-compatible server with curl curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP", "messages": [ {"role": "user", "content": "Hello"} ] }' - Hermes Agent
How to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with Hermes Agent:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP"
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP"
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP" \ --custom-provider-id mlx-lm \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
Download axquant_quantizer_execution.json from AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP: direct link, hf CLI and curl.
- Browser
- Download file 191 kB
-
https://huggingface.co/AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP/resolve/main/axquant_quantizer_execution.json
- Command line
-
hf download hf://AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP/axquant_quantizer_execution.json
-
curl -L -o axquant_quantizer_execution.json https://huggingface.co/AutomatosX/AX-Qwen3.8-Flash-Next-MLX-AXQ-MXFP4-MTP/resolve/main/axquant_quantizer_execution.json
191 kB
| { | |
| "created_at": "2026-10-04T19:16:32.264697Z", | |
| "plan_sha256": "bc260598c0db87fe643fe31d1936aa03b7f5c2529f2bc8437a32c554f16958c8", | |
| "records": [ | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.0.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.key_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_0", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_1", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_2", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_3", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_4", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_5", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_6", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_7", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_8", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_9", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_10", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_11", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_12", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_13", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_14", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_15", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_16", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_17", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_18", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_19", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_20", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_21", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_22", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_23", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_24", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_25", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_26", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_27", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_28", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_29", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_30", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_31", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_32", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_33", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_34", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_35", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_36", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_37", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_38", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_39", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_40", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_41", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_42", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_43", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_44", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_45", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_46", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_47", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_48", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_49", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_50", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_51", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_52", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_53", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_54", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_55", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_56", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_57", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_58", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_59", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_60", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_61", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_62", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_63", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_64", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_65", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_66", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_67", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_68", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_69", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_70", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_71", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_72", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_73", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_74", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_75", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_76", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_77", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_78", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_79", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_80", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_81", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_82", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_83", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_84", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_85", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_86", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_87", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_88", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_89", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_90", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_91", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_92", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_93", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_94", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_95", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_96", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_97", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_100", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_101", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_98", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_99", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_102", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_103", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_104", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_105", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_106", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_107", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_108", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_109", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_110", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_111", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_112", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_113", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_114", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_115", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_116", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_117", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_118", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_119", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_120", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_121", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_122", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_123", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_124", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_125", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_126", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.ple_embedding.ngram_embedding.shard_127", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.1.ple.value_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.10.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.11.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.12.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.13.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.14.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.15.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.16.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.17.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.18.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.19.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.2.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.20.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.21.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.22.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.23.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.24.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.25.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.26.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.27.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.28.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.29.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.3.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.30.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.31.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.32.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.33.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.34.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.35.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.36.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.37.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.38.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.39.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.4.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.40.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.41.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.42.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.43.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.44.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.45.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.46.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.47.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.5.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.6.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.self_attn.indexer.index_qk_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.self_attn.k_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.self_attn.o_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.self_attn.q_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.7.self_attn.v_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.8.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.linear_attn.in_proj_a", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.linear_attn.in_proj_b", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.linear_attn.in_proj_qkv", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.linear_attn.in_proj_z", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.linear_attn.out_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.experts.gate_up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.embed_tokens", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.experts.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.shared_expert.down_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.shared_expert.gate_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 4, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.shared_expert.up_proj", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 32, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "model.language_model.layers.9.mlp.shared_expert_gate", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| }, | |
| { | |
| "bits": 8, | |
| "fallback": false, | |
| "group_size": 64, | |
| "metadata": {}, | |
| "method": "affine", | |
| "module_path": "lm_head", | |
| "note": "mlx-vlm affine packing", | |
| "success": true | |
| } | |
| ], | |
| "schema_version": "axquant.quantizer-execution.v2" | |
| } | |