25 lines
476 B
JSON
25 lines
476 B
JSON
|
{
|
||
|
|
"input_config": {
|
||
|
|
"model_type": "gpt2",
|
||
|
|
"architectures": [
|
||
|
|
"GPT2LMHeadModel"
|
||
|
|
],
|
||
|
|
"activation_function": "gelu",
|
||
|
|
"attn_pdrop": 0.1,
|
||
|
|
"embd_pdrop": 0.1,
|
||
|
|
"resid_pdrop": 0.1,
|
||
|
|
"initializer_range": 0.02,
|
||
|
|
"layer_norm_epsilon": 1e-05,
|
||
|
|
"n_embd": 768,
|
||
|
|
"n_inner": 3072,
|
||
|
|
"n_head": 12,
|
||
|
|
"n_layer": 12,
|
||
|
|
"vocab_size": 1,
|
||
|
|
"n_ctx": 512,
|
||
|
|
"n_positions": 512
|
||
|
|
},
|
||
|
|
"num_steps_switch": null,
|
||
|
|
"save_steps": [
|
||
|
|
40000
|
||
|
|
]
|
||
|
|
}
|