Qwen2.5-32B-Instruct finetune surpassing R1-Distill from Scratch
tools
52 Pulls Updated 5 weeks ago
997a823b673a · 151B
{
"min_p": 0.1,
"num_ctx": 42768,
"repeat_penalty": 1.1,
"stop": [
"<|im_end|>",
"<|endoftext|>"
],
"temperature": 0.6,
"top_k": 40,
"top_p": 0.95
}