Replace model_config.yaml (full 98-field dataclass dump) with a minimal config.json (only the fields this model actually sets)
a5d4f68 verified | { | |
| "hidden_size": 896, | |
| "num_hidden_layers": 16, | |
| "num_attention_heads": 14, | |
| "num_key_value_heads": 2, | |
| "intermediate_size": 256, | |
| "shared_intermediate_size": 3072, | |
| "vocab_size": 32000, | |
| "max_position_embeddings": 2048, | |
| "attention_type": "gqa", | |
| "mlp_type": "tr_hash_engine", | |
| "num_experts": 4, | |
| "top_k": 2, | |
| "top_k_primary_weight": 0.5, | |
| "routing_strategy": "token_id_multi_hash", | |
| "route_hash_count": 2, | |
| "shared_expert": true, | |
| "shared_output_scale": 1.0, | |
| "routed_output_scale": 2.0, | |
| "use_qk_norm": true, | |
| "tie_word_embeddings": true | |
| } | |