Note SFT bootstraping for Terminal-Bench 2.0 and SWE-Bench.

Note RL for Terminal-Bench 2.0

Note aggregates high-quality agent trajectories from various environments including web browsing, code generation, household tasks, knowledge base querying, and software engineering. The dataset is collected through methods described in Agent Data Protocol.

Note 80,036 trajectories generated by a software engineering agent based on the SWE-agent framework, using various models as action generators. In these trajectories, the agent attempts to solve GitHub issues from the nebius/SWE-bench-extra and the dev split of princeton-nlp/SWE-bench.