{ "_name_or_path": "OpenDFM/SciDFM-MoE-A5.6B-v1.0", "architectures": [ "SciDFMForCausalLM" ], "auto_map": { "AutoConfig": "configuration_scidfm.SciDFMConfig", "AutoModel": "modeling_scidfm.SciDFMModel", "AutoModelForCausalLM": "modeling_scidfm.SciDFMForCausalLM" }, "attention_dropout": 0.0, "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 3200, "initializer_range": 0.02, "intermediate_size": 8640, "max_position_embeddings": 8192, "model_type": "SciDFM", "num_attention_heads": 32, "num_experts_per_tok": 2, "num_hidden_layers": 26, "num_key_value_heads": 32, "num_local_experts": 8, "output_router_logits": false, "rms_norm_eps": 1e-06, "rope_theta": 10000, "router_aux_loss_coef": 0.02, "router_jitter_noise": 0.0, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.40.1", "use_cache": true, "vocab_size": 32196 }