Spaces:

Humanlearning
/

Cyber_analyst-round1

Sleeping

Cyber_analyst-round1 / training /configs

11.4 kB

Ctrl+K

1 contributor

feat: introduce GRPO GPU fallback support, enhance training script with warmstart tagging, and add learning rate parameter for improved training flexibility

1b6d30b 21 days ago

reward_ablations
feat: introduce reward ablation configurations for enhanced training flexibility, implement YAML loading with extends support, and add reward variant tracking in training scripts 21 days ago
grpo_small.yaml

5.21 kB
feat: update README with GPU-utilization tuning instructions, enhance modal training script with run name parameter, and modify GRPO configuration for trace logging and vLLM settings 22 days ago
sft_warmstart_fast.yaml

3.72 kB
feat: introduce GRPO GPU fallback support, enhance training script with warmstart tagging, and add learning rate parameter for improved training flexibility 21 days ago