Ctrl+K
- baseline
- grpo
- internvl3-8b-instruct-lora_epoch10_5e-6
- llava-ov-lora
- ood
- qwen2.5vl-7b-lora_epoch10_2e-5
- qwen2.5vl-7b-qvq_thinking_full_v2
- qwen2.5vl-7b-thinking_lora_v2
- qwen2.5vl-7b-thinking_v2_full_cls_comet_grpo
- qwen2.5vl-7b-thinking_v2_full_cls_grpo
- qwen2.5vl-7b-thinking_v2_full_comet_grpo_continue
- qwen2.5vl-7b-thinking_v2_full_comet_grpo_continue_e8
- selective_loss
- 4.94 kB