| { | |
| "vocab_size": 152064, | |
| "pad_id": 151645, | |
| "eos_id": -1, | |
| "dim": 256, | |
| "n_layers": 18, | |
| "n_heads": 12, | |
| "n_kv_heads": 6, | |
| "use_kan": true, | |
| "train_softmax_temp": true, | |
| "use_softmax_temp_proj": true, | |
| "softmax_bias": false, | |
| "multiple_of": 256, | |
| "ffn_dim_multiplier": null, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 500000, | |
| "use_scaled_rope": false, | |
| "max_batch_size": 100, | |
| "max_seq_len": 128, | |
| "num_experts": 14, | |
| "num_experts_per_tok": 4, | |
| "model_type": "KANaMoEv2" | |
| } |