trillionlabs/SimScholar-RL
Viewer • Updated • 13k • 39
Simulated Scholar Search (S3) — Models & Datasets
Datasets and model checkpoints for training and studying scientific-literature search agents in the S3 local environment.
Note 13,000 synthetic single-hop literature-search QA items for agentic reinforcement learning.
trillionlabs/SimScholar-SFT
Viewer • Updated • 14.6k • 45Note 14,633 labeled single- and two-hop ReAct trajectories with complete tool-use conversations.
trillionlabs/sim-scholar-qwen3-4b
Text Generation • 4B • Updated • 43Note Qwen3-4B research checkpoint trained with agentic RL in the S3 literature-search environment.