| { | |
| "architectures": [ | |
| "TCMoEForCausalLM" | |
| ], | |
| "bos_token_id": 0, | |
| "eos_token_id": 0, | |
| "hidden_size": 1024, | |
| "initializer_range": 0.006, | |
| "intermediate_size": 2816, | |
| "max_position_embeddings": 2048, | |
| "model_type": "tcmoe", | |
| "moe_topk": 2, | |
| "norm_eps": 1e-05, | |
| "num_attention_heads": 16, | |
| "num_experts": 8, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 2, | |
| "num_null_experts": 2, | |
| "rope_pct": 1.0, | |
| "rope_theta": 10000.0, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.49.0", | |
| "use_cache": true, | |
| "vocab_size": 50432 | |
| } | |