2019-02-08 09:05:52 +00:00
|
|
|
{
|
2020-04-24 16:04:48 +00:00
|
|
|
"afn": "gelu",
|
2020-01-31 23:00:28 +00:00
|
|
|
"architectures": [
|
|
|
|
|
"OpenAIGPTLMHeadModel"
|
|
|
|
|
],
|
2019-02-08 09:05:52 +00:00
|
|
|
"attn_pdrop": 0.1,
|
|
|
|
|
"embd_pdrop": 0.1,
|
|
|
|
|
"initializer_range": 0.02,
|
2019-02-08 20:44:42 +00:00
|
|
|
"layer_norm_epsilon": 1e-05,
|
2020-04-24 16:04:48 +00:00
|
|
|
"model_type": "openai-gpt",
|
2019-02-08 09:05:52 +00:00
|
|
|
"n_ctx": 512,
|
|
|
|
|
"n_embd": 768,
|
|
|
|
|
"n_head": 12,
|
|
|
|
|
"n_layer": 12,
|
|
|
|
|
"n_positions": 512,
|
|
|
|
|
"n_special": 0,
|
2020-04-24 16:04:48 +00:00
|
|
|
"predict_special_tokens": true,
|
2019-02-08 09:05:52 +00:00
|
|
|
"resid_pdrop": 0.1,
|
2020-04-24 16:04:48 +00:00
|
|
|
"summary_activation": null,
|
|
|
|
|
"summary_first_dropout": 0.1,
|
|
|
|
|
"summary_proj_to_labels": true,
|
|
|
|
|
"summary_type": "cls_index",
|
|
|
|
|
"summary_use_proj": true,
|
2020-05-11 21:09:56 +00:00
|
|
|
"task_specific_params": {
|
|
|
|
|
"text-generation": {
|
|
|
|
|
"do_sample": true,
|
|
|
|
|
"max_length": 50
|
|
|
|
|
}
|
|
|
|
|
},
|
2019-02-08 09:05:52 +00:00
|
|
|
"vocab_size": 40478
|
2020-05-11 21:09:56 +00:00
|
|
|
}
|
