| { | |
| "_name_or_path": "Qwen/Qwen2.5-7B-Instruct", | |
| "architectures": [ | |
| "Qwen2ForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 151643, | |
| "eos_token_id": 151645, | |
| "hidden_act": "silu", | |
| "hidden_size": 3584, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 18944, | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 28, | |
| "model_type": "qwen2", | |
| "num_attention_heads": 28, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 4, | |
| "quant_config": { | |
| "model.layers.0.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.0.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.1.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.10.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.11.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.12.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.13.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.14.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.15.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.16.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.17.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.18.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.19.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.2.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.20.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.21.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.22.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.23.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.24.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.25.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.26.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.27.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.3.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.4.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.5.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.6.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.7.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.8.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.mlp.down_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 18944, | |
| "in_features": 18944, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.mlp.gate_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.mlp.up_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 18944, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.self_attn.k_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.self_attn.o_proj": { | |
| "bias": null, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.self_attn.q_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 3584, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| }, | |
| "model.layers.9.self_attn.v_proj": { | |
| "bias": true, | |
| "enable_norm": true, | |
| "enable_perm": true, | |
| "group_num": 1, | |
| "group_size": 3584, | |
| "in_features": 3584, | |
| "indices_as_float": false, | |
| "is_indice_packed": true, | |
| "num_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "num_res_centroids": [ | |
| -1, | |
| 65536 | |
| ], | |
| "out_features": 512, | |
| "outlier_size": 0, | |
| "vector_lens": [ | |
| -1, | |
| 8 | |
| ] | |
| } | |
| }, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 1000000.0, | |
| "sliding_window": null, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.45.1", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 152064 | |
| } | |