| { |
| "format": "cs6013-v2", |
| "base_model": "Qwen3.5-4B", |
| "recipe": { |
| "linear": { |
| "mlp.gate_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "mlp.up_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "mlp.down_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "linear_attn.in_proj_qkv": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "linear_attn.in_proj_z": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "linear_attn.out_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "self_attn.q_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "self_attn.k_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "self_attn.v_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "self_attn.o_proj": { |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| } |
| }, |
| "layer_overrides": {}, |
| "embed": { |
| "bits": 8, |
| "group_size": 256, |
| "symmetric": true, |
| "clip_grid": 0 |
| }, |
| "clip_grid": 20, |
| "calibration": { |
| "n_samples": 256, |
| "max_len": 2048 |
| } |
| }, |
| "layout": { |
| "model.language_model.embed_tokens.weight": { |
| "shape": [ |
| 248320, |
| 2560 |
| ], |
| "bits": 8, |
| "group_size": 256, |
| "symmetric": true |
| }, |
| "model.language_model.layers.1.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.1.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.1.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.0.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.1.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.1.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.10.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.11.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.12.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.13.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.14.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.15.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.16.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.17.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.18.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.19.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.2.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.20.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.21.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.22.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.23.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.24.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.25.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.26.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.27.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.28.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.29.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.3.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.30.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.31.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.4.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.5.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.6.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.7.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.8.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.linear_attn.in_proj_z.weight": { |
| "shape": [ |
| 4096, |
| 2560 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| }, |
| "model.language_model.layers.9.linear_attn.out_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "bits": 6, |
| "group_size": 128, |
| "symmetric": false |
| } |
| }, |
| "dropped": { |
| "mtp.layers.0.mlp.down_proj.weight": { |
| "shape": [ |
| 2560, |
| 9216 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.mlp.gate_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.mlp.up_proj.weight": { |
| "shape": [ |
| 9216, |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.0.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.1.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.10.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.11.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.12.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.13.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.14.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.15.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.16.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.17.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.18.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.19.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.2.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.20.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.21.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.22.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.23.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.3.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.4.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.5.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.6.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.7.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.8.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.attn.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.attn.proj.weight": { |
| "shape": [ |
| 1024, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.attn.qkv.bias": { |
| "shape": [ |
| 3072 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.attn.qkv.weight": { |
| "shape": [ |
| 3072, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.mlp.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.mlp.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.mlp.linear_fc2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.mlp.linear_fc2.weight": { |
| "shape": [ |
| 1024, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.norm1.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.norm1.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.norm2.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.blocks.9.norm2.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.linear_fc1.bias": { |
| "shape": [ |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.linear_fc1.weight": { |
| "shape": [ |
| 4096, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.linear_fc2.bias": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.linear_fc2.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.norm.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.merger.norm.weight": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.patch_embed.proj.bias": { |
| "shape": [ |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.patch_embed.proj.weight": { |
| "shape": [ |
| 1024, |
| 3, |
| 2, |
| 16, |
| 16 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "model.visual.pos_embed.weight": { |
| "shape": [ |
| 2304, |
| 1024 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.fc.weight": { |
| "shape": [ |
| 2560, |
| 5120 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.input_layernorm.weight": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.post_attention_layernorm.weight": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.k_norm.weight": { |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.k_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.o_proj.weight": { |
| "shape": [ |
| 2560, |
| 4096 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.q_norm.weight": { |
| "shape": [ |
| 256 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.q_proj.weight": { |
| "shape": [ |
| 8192, |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.layers.0.self_attn.v_proj.weight": { |
| "shape": [ |
| 1024, |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.norm.weight": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.pre_fc_norm_embedding.weight": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| }, |
| "mtp.pre_fc_norm_hidden.weight": { |
| "shape": [ |
| 2560 |
| ], |
| "dtype": "bfloat16" |
| } |
| } |
| } |
