monte-inc/qwen2.5-1.5b-cloudsync-support-resume-01M2HP4M1G955GXWV3XBFR5YT6-train110

Resume set — CloudSync Pro support agent, GRPO step 110

Optimizer state and weights for resuming training, not for inference. NeMo-RL v0.7.0 checkpoint layout, written by run 01M2HP4M1G955GXWV3XBFR5YT6 (50 GRPO steps from the SFT checkpoint; 96.07% on the dev exam).

For serving, use the model instead: monte-inc/qwen2.5-1.5b-cloudsync-support, folder 01M2HP4M1G955GXWV3XBFR5YT6/train-110/ — which also carries the full model card: task, results, training recipe and limitations.