{ "_name_or_path": "/CodeFuse-13B", "architectures": [ "GPTNeoXForCausalLM" ], "attention_dropout": 0.0, "bos_token_id": 100256, "classifier_dropout": 0.1, "eos_token_id": 100256, "hidden_act": "gelu", "hidden_dropout": 0.0, "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 20480, "layer_norm_eps": 1e-05, "max_position_embeddings": 4096, "model_type": "gpt_neox", "num_attention_heads": 40, "num_hidden_layers": 40, "rope_scaling": null, "rotary_emb_base": 10000, "rotary_pct": 1.0, "tie_word_embeddings": false, "torch_dtype": "float16", "transformers_version": "4.31.0", "use_cache": false, "use_parallel_residual": true, "vocab_size": 100831 }