Pinned Loading

1. CONSTRAINT-DECOMPOSITION-for-Multi-Objective-RLHF CONSTRAINT-DECOMPOSITION-for-Multi-Objective-RLHF Public

   Early-stage research exploring decomposed reward modeling for complex instruction-following in large language models.

   Python 1

2. ai-physicist-central-llm ai-physicist-central-llm Public

   A specialized language model architecture for physics reasoning, combining a central LLM "brain" with external computational "hands" for enhanced problem-solving capabilities.

   Python 1

3. RL-for-LLM-training-Variance-Stabilized-Dropout-Implementation RL-for-LLM-training-Variance-Stabilized-Dropout-Implementation Public

   RL Stabilized Dropout

   Python 1

4. Reinforcement-learning_ML_Profiler Reinforcement-learning_ML_Profiler Public

   This repo implements a realistic ML engineering task. Think of it like a mini-version of what you'd build at an ML company to profile model behavior during fine-tuning experiments.

   Python 1

5. decomposed-reward-rlhf decomposed-reward-rlhf Public

   Technical report

   Python 1

6. DataLedger DataLedger Public

   DataLedger defines a cryptographic provenance standard for AI training datasets. A dataset publisher signs a manifest document with an ed25519 key. A training pipeline records which manifests it co…

   Python 1