Hugging Face's logo
Jim Lai's picture

Jim Lai

grimjim
66
whoashish115's profile picture kramello18's profile picture Marcus-Arcadius's profile picture
·

AI & ML interests

Experimenting primarily with 7B-12B parameter text completion models. Not all models are intended for direct end use, but aim for research and/or educational purposes. Recent Contributions: stabilized refusal direction ablation via Gram-Schmidt orthonormalization and norm-preserving interventions; confirmed reasoning transfer via model merger.

Recent Activity

repliedto BoldingBuilds's post 6 days ago
Abliterated thinking models have a habit: they decide what to say, then keep arguing with themselves until the token budget runs out. Qwen3.8-27B Abliterated ThinkFix is an uncensored Qwen3.8-27B with one output row edited so the model closes its thinking when it's ready to answer. On 40 sensitive prompts the edit never saw, 8k output cap: 33 replies finished cleanly instead of 22, 5 hit the limit instead of 17, 61 minutes for the set instead of 75. Agent use was checked four ways and the edit changes nothing there. MTP draft head kept. Q4_K_M to Q8_0, each tested after the edit. The card lists what the edit was fit on and what it does not do. https://huggingface.co/BoldingBuilds/Qwen3.8-27B-Abliterated-ThinkFix-GGUF
updated a collection about 1 month ago
Tools
posted an update about 2 months ago
I think it's clear in retrospect that "frankenmerges", which repeated blocks of layers, amounted to a crude approximation of looped transformers architecture, hence them able to work at all instead of just breaking. They lucked out due to much of the signal passing through residual streams being preserved and only modulated along the way. That said, not all models are suited for this. Models which feature ever-increasing magnitudes as inference progressess through layers risk exploding precision limits.
View all activity

Organizations

Blog-explorers's profile picture Social Post Explorers's profile picture Hugging Face Discord Community's profile picture Debased AI's profile picture Anthracite Core's profile picture Anthracite's profile picture SillyTilly's profile picture smE's profile picture Hugging Face Context Course's profile picture GrimJim's Merge Workshop's profile picture Tokenizers and Chat Templates Only's profile picture MCP-1st-Birthday's profile picture GrimJim's Experimental Workshop's profile picture Build Small Hackathon's profile picture ICML 2026 Agent Reproductions's profile picture