| { | |
| "architectures": [ | |
| "BailingMoeForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_bailing_moe.BailingMoeConfig", | |
| "AutoModel": "modeling_bailing_moe.BailingMoeModel", | |
| "AutoModelForCausalLM": "modeling_bailing_moe.BailingMoeForCausalLM", | |
| "AutoModelForTokenClassification": "modeling_bailing_moe.BailingMoeForTokenClassification" | |
| }, | |
| "eos_token_id": 126081, | |
| "pad_token_id": 126081, | |
| "first_k_dense_replace": 0, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.006, | |
| "intermediate_size": 5632, | |
| "max_position_embeddings": 32768, | |
| "model_type": "bailing_moe", | |
| "moe_intermediate_size": 1408, | |
| "num_experts": 64, | |
| "num_shared_experts": 2, | |
| "norm_topk_prob": true, | |
| "num_attention_heads": 16, | |
| "num_experts_per_tok": 6, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 4, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 600000, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.36.0", | |
| "use_cache": true, | |
| "use_bias": false, | |
| "use_qkv_bias": false, | |
| "vocab_size": 126464, | |
| "embedding_dropout": 0.0, | |
| "norm_head": true, | |
| "norm_softmax": false, | |
| "output_dropout": 0.0, | |
| "output_router_logits": false | |
| } | |