| { |
| "_name_or_path": "../miniUSE-tw-v1", |
| "architectures": [ |
| "MiniMindLM" |
| ], |
| "auto_map": { |
| "AutoConfig": "LMConfig.LMConfig", |
| "AutoModelForCausalLM": "model.MiniMindLM" |
| }, |
| "aux_loss_alpha": 0.1, |
| "dim": 768, |
| "dropout": 0.0, |
| "flash_attn": true, |
| "hidden_dim": 2048, |
| "max_seq_len": 8192, |
| "model_type": "minimind", |
| "multiple_of": 64, |
| "n_heads": 8, |
| "n_kv_heads": 2, |
| "n_layers": 16, |
| "n_routed_experts": 4, |
| "n_shared_experts": true, |
| "norm_eps": 1e-05, |
| "norm_topk_prob": true, |
| "num_experts_per_tok": 2, |
| "rope_theta": 1000000.0, |
| "scoring_func": "softmax", |
| "seq_aux": true, |
| "torch_dtype": "float32", |
| "transformers_version": "4.48.0", |
| "use_moe": false, |
| "vocab_size": 6400 |
| } |
|
|
