Collections

Code for "EgoForesight: Effectively Scaling Policy Learning with Noisy Egocentric Actions". GitHub: https://github.com/LinShan-Bin/EgoForesight

DPO and GRPO instruction-following models trained from Qwen2.5-0.5B-Instruct. Weights and browser builds.

DFI students of the Qwen3-14B animal-welfare AuditBench organisms: logit/SFT, filtered/unfiltered, Qwen3-14B + Llama-3.1-8B.