(Unsloth Dynamic Quants) DeepSeek's first-generation of reasoning models with comparable performance to OpenAI-o1, 671B MoE full model, not dense distilled models.

127 2 months ago

f4d24e9138dd · 148B
{
"stop": [
"<|begin▁of▁sentence|>",
"<|end▁of▁sentence|>",
"<|User|>",
"<|Assistant|>"
]
}