Modelpublic

aqbond/deepswe-k8s-sync-S0b16-c256-step140to180-step175

sourceHugging Faceapache-2.0updated 27d agoView on Hugging Face

0likes259downloads

Model Card

deepswe-k8s-sync-S0b16 baseline step 175

FSDP-merged Hugging Face weights after training step 175.

—Experiment: deepswe-k8s-sync-S0b16-c256-step140to180-20260728-231847
—Recipe: S0b16 sync PPO, no trajectory repair
—Checkpoint source: outputs_all/checkpoints/.../global_step_175/actor
—Paired replace repo: aqbond/deepswe-k8s-sync-S0b16-c256-replace-tokenexact-step175