Pinned Loading

1. can-i-finetune-this can-i-finetune-this Public

   Estimate whether a Hugging Face model fits and fine-tunes on your local GPU.

   Python 793 107

2. pairjudge pairjudge Public

   Pairwise LLM judges (A/B/tie): budget-aware multi-turn packing, position-bias correction, pseudo-label distillation. Generalized from the 4th-place (gold) solution to Kaggle LMSYS Chatbot Arena.

   Python 169 12

3. tracedistill tracedistill Public

   Distill teacher chains-of-thought into a LoRA adapter via a strict boxed-answer format contract + two-phase Train→Nudge (silver-medal NVIDIA Nemotron reasoning recipe, as a tested library).

   Python 8 1

4. RepoGuardBench RepoGuardBench Public

   Benchmarking repository-borne prompt injection attacks and lightweight defenses for local coding agents. DL4C @ ICML 2026.

   Python 4

5. delta-mfp-local-agents delta-mfp-local-agents Public

   Delta-MFP: counterfactual-replay failure diagnosis for local tool-use agents (FAGEN @ ICML 2026)

   Python 3

6. huggingface/peft huggingface/peft Public

   🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

   Python 21.5k 2.4k