{ "activation_function": "gelu", "architectures": [ "GPTBigCodeLMHeadModel" ], "attention_softmax_in_fp32": true, "attention_window_size": 8192, "attn_pdrop": 0.1, "bos_token_id": 0, "embd_pdrop": 0.1, "eos_token_id": 0, "head_groups": 2, "initializer_range": 0.02, "layer_norm_epsilon": 1e-05, "model_type": "gpt_bigcode", "n_embd": 3072, "n_head": 24, "n_inner": 12288, "n_layer": 30, "n_positions": 16384, "resid_pdrop": 0.1, "rotary_embedding_scale": -13.81551, "scale_attention_softmax_in_fp32": true, "scale_attn_weights": true, "summary_activation": null, "summary_first_dropout": 0.1, "summary_proj_to_labels": true, "summary_type": "cls_index", "summary_use_proj": true, "transformers_version": "4.34.1", "use_cache": true, "use_position_embeddings": false, "use_rotary_embeddings": true, "vocab_size": 49152 }