Instructions to use nguyenthanhdo/llama-3.2-3b-math with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nguyenthanhdo/llama-3.2-3b-math with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nguyenthanhdo/llama-3.2-3b-math")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nguyenthanhdo/llama-3.2-3b-math")
model = AutoModelForCausalLM.from_pretrained("nguyenthanhdo/llama-3.2-3b-math")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use nguyenthanhdo/llama-3.2-3b-math with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nguyenthanhdo/llama-3.2-3b-math"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nguyenthanhdo/llama-3.2-3b-math",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nguyenthanhdo/llama-3.2-3b-math

SGLang

How to use nguyenthanhdo/llama-3.2-3b-math with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nguyenthanhdo/llama-3.2-3b-math" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nguyenthanhdo/llama-3.2-3b-math",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nguyenthanhdo/llama-3.2-3b-math" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nguyenthanhdo/llama-3.2-3b-math",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nguyenthanhdo/llama-3.2-3b-math with Docker Model Runner:
```
docker model run hf.co/nguyenthanhdo/llama-3.2-3b-math
```

See axolotl config

axolotl version: 0.6.0

base_model: /workspace/models/llama-3.2-3b-wizard
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

chat_template: llama3
datasets:
  - path: json
    data_files:
      - /workspace/data/numina-20k/train.jsonl
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content

val_set_size: 0.0
output_dir: /workspace/models/llama-3.2-3b-math

sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 2
num_epochs: 3
optimizer: paged_adamw_8bit
lr_scheduler: constant
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 2
xformers_attention:
flash_attention: true

warmup_steps: 100
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
save_total_limit: 10
debug:
deepspeed: /workspace/axolotl-0.6.0/deepspeed_configs/zero3_bf16.json
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

workspace/models/llama-3.2-3b-math

This model was trained from scratch on the json dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 2
eval_batch_size: 2
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 8
total_train_batch_size: 32
total_eval_batch_size: 4
optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: constant
lr_scheduler_warmup_steps: 100
num_epochs: 3

Training results

Framework versions

Transformers 4.46.3
Pytorch 2.5.1+cu121
Datasets 3.1.0
Tokenizers 0.20.3

Downloads last month: -

Safetensors

Model size

3B params

Tensor type

BF16