cot-self-instruct
llama-3.1-instruct_v8_self_gen_v2_cotllama-3.1_v6_500-instruct_self_gen_cot-hfqwen2.5-7b-instruct-new-math-1k-grpo-with-length-0.1-cot-prompt-v6-self-correctweb-self-cot-sciworld_Llama-3.1-8B-Instruct-100stepweb-self-cot-sciworld_Llama-3.1-8B-Instruct-50stepweb-self-cot-sciworld_Llama-3.1-8B-Instruct-200stepweb-self-cot-sciworld_Llama-3.2-3B-Instruct-100step