{ "architectures": [ "BlueLMForCausalLM" ], "auto_map": { "AutoConfig": "configuration_bluelm.BlueLMConfig", "AutoModelForCausalLM": "modeling_bluelm.BlueLMForCausalLM" }, "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 11008, "max_position_embeddings": 2048, "model_type": "BlueLM", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 32, "pad_token_id": 3, "pretraining_tp": 1, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 10000.0, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.33.1", "use_cache": true, "use_stable_embedding": true, "vocab_size": 100096 }