| { | |
| "architectures": [ | |
| "MiniMoE" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "mini_moe.MiniMoEConfig", | |
| "AutoModelForCausalLM": "mini_moe.MiniMoE" | |
| }, | |
| "dim": 1024, | |
| "drop_rate": 0.0, | |
| "dtype": "float32", | |
| "model_type": "mini-moe", | |
| "num_attention_kv_heads": 8, | |
| "num_attention_q_heads": 16, | |
| "num_expert": 8, | |
| "num_layers": 12, | |
| "qkv_bias": false, | |
| "rope_base": 10000, | |
| "top_k": 4, | |
| "transformers_version": "4.56.2", | |
| "use_aux_loss": false, | |
| "vocab_size": 32000 | |
| } | |