{ "activation_function": "gelu", "architectures": [ "GPTBigCodeForCausalLM" ], "attention_softmax_in_fp32": true, "multi_query": true, "resid_pdrop": 0.1, "attn_pdrop": 0.1, "embd_pdrop": 0.1, "bos_token_id": 0, "eos_token_id": 0, "initializer_range": 0.02, "layer_norm_epsilon": 1e-05, "model_type": "gpt_bigcode", "hidden_size": 2048, "num_attention_heads": 16, "n_inner": 8192, "n_layer": 24, "max_position_embeddings": 8192, "pad_key_length": true, "scale_attn_weights": true, "torch_dtype": "float32", "transformers_version": "4.28.1", "use_cache": true, "vocab_size": 49152 }