Upload folder using huggingface_hub

9576446 25 days ago

661 Bytes

license: apache-2.0
base_model: laion/r2egym-nl2bash-stack-bugsseq-fixthink-again
tags:
  - reinforcement-learning
  - code
  - nemotron
  - terminal
  - terminus-structured
  - rloo-n

RL Nemotron Easy (Step 63) - Terminus Structured

RL-trained Qwen3-8B with terminus-structured agent on nemotron_easy_2000_tasks dataset.

Base model: laion/r2egym-nl2bash-stack-bugsseq-fixthink-again (8B SFT)
Training: 63 RL steps on nemotron_easy_2000_tasks (data processing / terminal tasks)
Agent: terminus-structured (bash, view, edit, create, search, submit tools)
Algorithm: RLOO-N, lr=1e-5
SWEBench-100 n=1: 33% (partial eval)