Spaces:

Arun-Sanjay
/

Red-Button

Sleeping

Red-Button / training /rollout_func.py

Phase 1: Initial scaffold, Claude Code workspace, repo structure per PROJECT.md Section 5

f707fd4 about 1 month ago

239 Bytes

	"""Custom GRPO rollout function driving ShutdownGymClient during training.

	TODO (Phase 13): implement shutdown_rollout_func per PROJECT.md Section 16.3.
	num_generations=4 × gradient_accumulation_steps=4 = 16 concurrent env sessions.
	"""