home / grpo-gsm8k-test / summaries

summaries: 26

This data as json

id task_name model_tag total_examples correct accuracy no_answer_count stop_reason_counts duration_human pass_k temperature top_p max_tokens error model
26 hendrycks_math(0) grpo-gsm8k-test-30ep-success_global_step_650 5000 2552 0.5104 640 {"stop:-": 4428, "length:-": 570, "stop:Problem:": 2} 2m 35s 1 0.0 0.95 1024   /mnt/data8tb/Documents/project/rlvr_winter/verl-my-rlvr/finished-models/grpo-gsm8k-test-30ep-success/global_step_650/actor/huggingface
Powered by Datasette · Queries took 1.019ms