Fine-tuned google/gemma-2-9b-it on princeton-nlp/gemma2-ultrafeedback-armorm with the SimPO objective.

1,003 3 months ago

1f0c17ce1cdb · 118B
{
"num_ctx": 4096,
"num_predict": 4096,
"repeat_penalty": 1,
"stop": [
"<start_of_turn>",
"<end_of_turn>"
]
}