| { | |
| "activation_function": "gelu_new", | |
| "attention_probs_dropout_prob": 0.0, | |
| "attn_pdrop": 0.0, | |
| "bos_token_id": 98, | |
| "embd_pdrop": 0.0, | |
| "eos_token_id": 98, | |
| "gradient_checkpointing": false, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.0, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 37, | |
| "layer_norm_epsilon": 1e-05, | |
| "model_type": "gptj", | |
| "n_ctx": 512, | |
| "n_embd": 32, | |
| "n_head": 4, | |
| "n_inner": null, | |
| "n_layer": 5, | |
| "n_positions": 512, | |
| "pad_token_id": 98, | |
| "resid_pdrop": 0.0, | |
| "rotary_dim": 4, | |
| "scale_attn_weights": true, | |
| "transformers_version": "4.11.0.dev0", | |
| "type_vocab_size": 16, | |
| "use_cache": true, | |
| "vocab_size": 1000 | |
| } | |