This model was trained 2x faster with Unsloth and Huggingface's TRL library. This is an experiment on fixing models with incorrect behaviors.

7B

26 Pulls Updated 3 months ago

4ddf52c3a1ac · 146B
{ "stop": [ "<|start_header_id|>", "<|end_header_id|>", "<|eot_id|>", "<|reserved_special_token" ], "temperature": 0.6 }