"summary_type": "cls_index" not supported

[tomyoung903]

Hello,

Thank you for releasing the code.

Running the code as instructed loads config information:

{
"activation_function": "gelu_new",
"architectures": [
"GPT2LMHeadModel"
],
"attn_pdrop": 0.1,
"bos_token_id": 50256,
"embd_pdrop": 0.1,
"eos_token_id": 50256,
"finetuning_task": null,
"initializer_range": 0.02,
"layer_norm_epsilon": 1e-05,
"model_type": "gpt2",
"n_ctx": 1024,
"n_embd": 768,
"n_head": 12,
"n_layer": 12,
"n_positions": 1024,
"num_labels": 1,
"output_attentions": false,
"output_hidden_states": false,
"resid_pdrop": 0.1,
"summary_activation": null,
"summary_first_dropout": 0.1,
"summary_proj_to_labels": true,
"summary_type": "cls_index",
"summary_use_proj": true,
"task_specific_params": {
"text-generation": {
"do_sample": true,
"max_length": 50
}
},
"torchscript": false,
"vocab_size": 50257
}

But "summary_type": "cls_index" is not supported in "modeling_utils.py", throwing an error.

Did I miss anything? Should "summary_type": "cls_index" be "token_ids" instead?

[kiseliu]

I think it's better to replace "token_ids" to "cls_index" in modeling_utils.py.