fix(v4): increase T4 SFT to 50 eps, add action trace to eval output

fix(v4): reduce MAX_COMP_LEN 384→192, fix eval hanging with for_inference

feat(training): add clean v4 notebook with all reward fixes

fix(notebook): rewrite run_grpo_episode with runbook-compliance reward

fix(grpo): fresh reward design — runbook-compliance scoring

fix(grpo): eliminate zero-advantage collapse from stale rewards

fix: format runbook as readable key-value pairs for model

fix: increase runbook truncation limit 800->2000 chars

fix(grpo): harden rollout reward injection on cloud runtime

fix(pm_ops_trainer): use direct closure capture instead of _self_ref

fix(pm_ops_trainer): use direct closure capture instead of _self_ref

fix(pm_ops_trainer): capture rewards via rollout_func wrapper, not inputs

fix(train_v3): use PMOpsGRPOTrainer to fix reward kwarg plumbing

Merge remote-tracking branch 'hf/main' into train/v1

fix(training): fix GRPO reward bugs + add v3 notebook with SFT warmup

fix(reward): harsh -1.0 for zero JSON output, -0.30 for zero task completion

fix(rollout): safe model unwrapping for accelerate/PEFT/DDP in _generate_no_vllm

fix(rollout): replace generate_rollout_completions (vLLM-only) with direct model.generate()

feat(training): PMOpsGRPOTrainer — override _calculate_rewards to inject rollout rewards directly

fix(train_v2): drop Unsloth, use bitsandbytes+PEFT to fix dep conflicts

fix(training): reward signal flow + new v2 notebook

modifying the training script to be more memory efficient: Unsloth, also reduced the number of steps to 15 instead of 40

writing the GRPO script to train qwen1.7 on A100 GRPO