tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
1937d55 verified Download config.json from huggingface-mic/MiniMax-M2.5-mic: direct link, hf CLI and curl.
- Browser
- Download file 1.29 kB
-
https://huggingface.co/huggingface-mic/MiniMax-M2.5-mic/resolve/main/config.json
- Command line
-
hf download hf://huggingface-mic/MiniMax-M2.5-mic/config.json
-
curl -L -o config.json https://huggingface.co/huggingface-mic/MiniMax-M2.5-mic/resolve/main/config.json
1.29 kB
| { | |
| "architectures": [ | |
| "MiniMaxM2ForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "attn_type_list": [ | |
| 1, | |
| 1 | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_minimax_m2.MiniMaxM2Config", | |
| "AutoModelForCausalLM": "modeling_minimax_m2.MiniMaxM2ForCausalLM" | |
| }, | |
| "bos_token_id": 200034, | |
| "dtype": "float32", | |
| "eos_token_id": 200020, | |
| "head_dim": 16, | |
| "hidden_act": "silu", | |
| "hidden_size": 64, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 128, | |
| "max_position_embeddings": 196608, | |
| "model_type": "minimax_m2", | |
| "mtp_transformer_layers": 1, | |
| "num_attention_heads": 4, | |
| "num_experts_per_tok": 2, | |
| "num_hidden_layers": 2, | |
| "num_key_value_heads": 2, | |
| "num_local_experts": 4, | |
| "num_mtp_modules": 3, | |
| "output_router_logits": false, | |
| "pad_token_id": null, | |
| "qk_norm_type": "per_layer", | |
| "rms_norm_eps": 1e-06, | |
| "rope_parameters": { | |
| "partial_rotary_factor": 0.5, | |
| "rope_theta": 5000000, | |
| "rope_type": "default" | |
| }, | |
| "rotary_dim": 64, | |
| "router_aux_loss_coef": 0.001, | |
| "router_jitter_noise": 0.0, | |
| "scoring_func": "sigmoid", | |
| "shared_intermediate_size": 0, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "5.22.0.dev0", | |
| "use_cache": true, | |
| "use_mtp": true, | |
| "use_qk_norm": true, | |
| "use_routing_bias": true, | |
| "vocab_size": 200064 | |
| } | |