Pinned Loading

1. multimodal-diffusion-steering multimodal-diffusion-steering Public

   VLM-steered diffusion policy for robot manipulation: Classifier-Free Guidance + Gemini best-of-N trajectory reranking for legible/safe pick-and-place (PyBullet, Franka Panda). Thesis code.

   Python

2. gemini-vlm-goal-inference gemini-vlm-goal-inference Public

   Do frontier VLMs (Gemini/GPT/Claude) infer a robot’s goal from partial motion? Goal-inference benchmark for robot-motion legibility - MS thesis code.

   HTML

3. openvla-oft-reproduction openvla-oft-reproduction Public

   Reproduction of OpenVLA-OFT on the LIBERO manipulation benchmark - closed-loop VLA rollouts and success evaluation (my runs)

   Python

4. prismatic-vlm-legibility prismatic-vlm-legibility Public

   VLM evaluation pipeline for robot-motion legibility, built on TRI-ML Prismatic-VLMs - my pipeline, prompts, and analysis

   Python

5. dpo-llm-alignment dpo-llm-alignment Public

   LLM alignment via SFT + Direct Preference Optimization on Anthropic HH-RLHF - reproduction and my experiments

   Python

6. ddpo-diffusion-rl ddpo-diffusion-rl Public

   Fine-tuning Stable Diffusion v1.5 with Denoising Diffusion Policy Optimization (DDPO) + LoRA for reward-driven generation (aesthetic, compressibility) - reproduction + my inference toolkit

   Python