{ "architectures": [ "PhiForCausalLM" ], "attention_probs_dropout_prob": 0, "bos_token_id": 50256, "eos_token_id": 50256, "hidden_act": "gelu_new", "hidden_dropout_prob": 0, "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 16384, "layer_norm_eps": 1e-05, "max_position_embeddings": 8192, "model_type": "phi", "num_attention_heads": 16, "num_hidden_layers": 28, "rotary_emb_base": 10000, "partial_rotary_factor": 0.5, "tie_word_embeddings": false, "torch_dtype": "float16", "transformers_version": "4.19.0.dev0", "use_cache": true, "vocab_size": 50400, "rope_scaling": { "type": "yarn", "factor": 8.0, "original_max_position_embeddings": 2048 } }