sharp-9b / decision_config.json
ya9919's picture
S²-MAS research release
6888745
Raw History Blame Contribute Delete
388 Bytes
{
"base_model": "Qwen/Qwen3.5-9B",
"model_name": "Qwen3.5-9B",
"method": "lora",
"max_length": 2048,
"architecture": "last-nonpadding-L2-linear-candidate-softmax-v1",
"prompt_version": "full-state-question-options-focus-v1",
"backbone_dtype": "bfloat16",
"head_dtype": "float32",
"normalized_feature_dtype": "float32",
"attention_implementation": "sdpa"
}