train_hellaswag_101112_1760638082 / adapter_config.json
rbelanec's picture
Training in progress, step 8979
b860ca2 verified
raw
history blame contribute delete
483 Bytes
{
"auto_mapping": null,
"base_model_name_or_path": "meta-llama/Meta-Llama-3-8B-Instruct",
"encoder_dropout": 0.1,
"encoder_hidden_size": 768,
"encoder_num_layers": 2,
"encoder_reparameterization_type": "MLP",
"inference_mode": true,
"modules_to_save": null,
"num_attention_heads": 32,
"num_layers": 32,
"num_transformer_submodules": 1,
"num_virtual_tokens": 100,
"peft_type": "P_TUNING",
"revision": null,
"task_type": "CAUSAL_LM",
"token_dim": 4096
}