Hugging Face's logo

vuiseng9
/
bert-base-squadv1-block-pruning-hybrid-filled-lt-nncf-57.92sparse-lt

Transformers
PyTorch
ONNX
bert
Inference Endpoints
Model card Files Community
bert-base-squadv1-block-pruning-hybrid-filled-lt-nncf-57.92sparse-lt / XP_linear_layer_sparsity_20M_params_57.92_sparsity.md
Chua, Vui Seng
Add collaterals
41d4286
preview code
|
raw history blame contribute delete
No virus
10.6 kB
| | layer_id | layer_type | param_type | shape | nparam | nnz | sparsity |
|----:|:---------------------------------------------------------|:-------------|:-------------|:-----------|---------:|-------:|-----------:|
| 5 | nncf_module.bert.encoder.layer.0.attention.self.query | NNCFLinear | weight | [320, 768] | 245760 | 93583 | 0.61921 |
| 7 | nncf_module.bert.encoder.layer.0.attention.self.key | NNCFLinear | weight | [320, 768] | 245760 | 98270 | 0.600138 |
| 9 | nncf_module.bert.encoder.layer.0.attention.self.value | NNCFLinear | weight | [320, 768] | 245760 | 113605 | 0.53774 |
| 11 | nncf_module.bert.encoder.layer.0.attention.output.dense | NNCFLinear | weight | [768, 320] | 245760 | 117208 | 0.523079 |
| 15 | nncf_module.bert.encoder.layer.0.intermediate.dense | NNCFLinear | weight | [185, 768] | 142080 | 97073 | 0.316772 |
| 17 | nncf_module.bert.encoder.layer.0.output.dense | NNCFLinear | weight | [768, 185] | 142080 | 94692 | 0.33353 |
| 21 | nncf_module.bert.encoder.layer.1.attention.self.query | NNCFLinear | weight | [320, 768] | 245760 | 118436 | 0.518083 |
| 23 | nncf_module.bert.encoder.layer.1.attention.self.key | NNCFLinear | weight | [320, 768] | 245760 | 118116 | 0.519385 |
| 25 | nncf_module.bert.encoder.layer.1.attention.self.value | NNCFLinear | weight | [320, 768] | 245760 | 107511 | 0.562537 |
| 27 | nncf_module.bert.encoder.layer.1.attention.output.dense | NNCFLinear | weight | [768, 320] | 245760 | 111189 | 0.547571 |
| 31 | nncf_module.bert.encoder.layer.1.intermediate.dense | NNCFLinear | weight | [315, 768] | 241920 | 148783 | 0.384991 |
| 33 | nncf_module.bert.encoder.layer.1.output.dense | NNCFLinear | weight | [768, 315] | 241920 | 143166 | 0.408209 |
| 37 | nncf_module.bert.encoder.layer.2.attention.self.query | NNCFLinear | weight | [576, 768] | 442368 | 162735 | 0.632128 |
| 39 | nncf_module.bert.encoder.layer.2.attention.self.key | NNCFLinear | weight | [576, 768] | 442368 | 164795 | 0.627471 |
| 41 | nncf_module.bert.encoder.layer.2.attention.self.value | NNCFLinear | weight | [576, 768] | 442368 | 135670 | 0.69331 |
| 43 | nncf_module.bert.encoder.layer.2.attention.output.dense | NNCFLinear | weight | [768, 576] | 442368 | 138445 | 0.687037 |
| 47 | nncf_module.bert.encoder.layer.2.intermediate.dense | NNCFLinear | weight | [339, 768] | 260352 | 154035 | 0.408359 |
| 49 | nncf_module.bert.encoder.layer.2.output.dense | NNCFLinear | weight | [768, 339] | 260352 | 150816 | 0.420723 |
| 53 | nncf_module.bert.encoder.layer.3.attention.self.query | NNCFLinear | weight | [576, 768] | 442368 | 170623 | 0.614296 |
| 55 | nncf_module.bert.encoder.layer.3.attention.self.key | NNCFLinear | weight | [576, 768] | 442368 | 178401 | 0.596714 |
| 57 | nncf_module.bert.encoder.layer.3.attention.self.value | NNCFLinear | weight | [576, 768] | 442368 | 171905 | 0.611398 |
| 59 | nncf_module.bert.encoder.layer.3.attention.output.dense | NNCFLinear | weight | [768, 576] | 442368 | 169172 | 0.617576 |
| 63 | nncf_module.bert.encoder.layer.3.intermediate.dense | NNCFLinear | weight | [368, 768] | 282624 | 163163 | 0.422685 |
| 65 | nncf_module.bert.encoder.layer.3.output.dense | NNCFLinear | weight | [768, 368] | 282624 | 157506 | 0.442701 |
| 69 | nncf_module.bert.encoder.layer.4.attention.self.query | NNCFLinear | weight | [576, 768] | 442368 | 175772 | 0.602657 |
| 71 | nncf_module.bert.encoder.layer.4.attention.self.key | NNCFLinear | weight | [576, 768] | 442368 | 177087 | 0.599684 |
| 73 | nncf_module.bert.encoder.layer.4.attention.self.value | NNCFLinear | weight | [576, 768] | 442368 | 163996 | 0.629277 |
| 75 | nncf_module.bert.encoder.layer.4.attention.output.dense | NNCFLinear | weight | [768, 576] | 442368 | 159335 | 0.639813 |
| 79 | nncf_module.bert.encoder.layer.4.intermediate.dense | NNCFLinear | weight | [386, 768] | 296448 | 167726 | 0.434214 |
| 81 | nncf_module.bert.encoder.layer.4.output.dense | NNCFLinear | weight | [768, 386] | 296448 | 159865 | 0.460732 |
| 85 | nncf_module.bert.encoder.layer.5.attention.self.query | NNCFLinear | weight | [384, 768] | 294912 | 114186 | 0.612813 |
| 87 | nncf_module.bert.encoder.layer.5.attention.self.key | NNCFLinear | weight | [384, 768] | 294912 | 132782 | 0.549757 |
| 89 | nncf_module.bert.encoder.layer.5.attention.self.value | NNCFLinear | weight | [384, 768] | 294912 | 134830 | 0.542813 |
| 91 | nncf_module.bert.encoder.layer.5.attention.output.dense | NNCFLinear | weight | [768, 384] | 294912 | 131941 | 0.552609 |
| 95 | nncf_module.bert.encoder.layer.5.intermediate.dense | NNCFLinear | weight | [336, 768] | 258048 | 153916 | 0.403537 |
| 97 | nncf_module.bert.encoder.layer.5.output.dense | NNCFLinear | weight | [768, 336] | 258048 | 145794 | 0.435012 |
| 101 | nncf_module.bert.encoder.layer.6.attention.self.query | NNCFLinear | weight | [448, 768] | 344064 | 131878 | 0.616705 |
| 103 | nncf_module.bert.encoder.layer.6.attention.self.key | NNCFLinear | weight | [448, 768] | 344064 | 144502 | 0.580014 |
| 105 | nncf_module.bert.encoder.layer.6.attention.self.value | NNCFLinear | weight | [448, 768] | 344064 | 130911 | 0.619516 |
| 107 | nncf_module.bert.encoder.layer.6.attention.output.dense | NNCFLinear | weight | [768, 448] | 344064 | 125928 | 0.633998 |
| 111 | nncf_module.bert.encoder.layer.6.intermediate.dense | NNCFLinear | weight | [280, 768] | 215040 | 135283 | 0.370894 |
| 113 | nncf_module.bert.encoder.layer.6.output.dense | NNCFLinear | weight | [768, 280] | 215040 | 131619 | 0.387932 |
| 117 | nncf_module.bert.encoder.layer.7.attention.self.query | NNCFLinear | weight | [448, 768] | 344064 | 132120 | 0.616002 |
| 119 | nncf_module.bert.encoder.layer.7.attention.self.key | NNCFLinear | weight | [448, 768] | 344064 | 152223 | 0.557574 |
| 121 | nncf_module.bert.encoder.layer.7.attention.self.value | NNCFLinear | weight | [448, 768] | 344064 | 141066 | 0.590001 |
| 123 | nncf_module.bert.encoder.layer.7.attention.output.dense | NNCFLinear | weight | [768, 448] | 344064 | 135662 | 0.605707 |
| 127 | nncf_module.bert.encoder.layer.7.intermediate.dense | NNCFLinear | weight | [211, 768] | 162048 | 109590 | 0.323719 |
| 129 | nncf_module.bert.encoder.layer.7.output.dense | NNCFLinear | weight | [768, 211] | 162048 | 107335 | 0.337635 |
| 133 | nncf_module.bert.encoder.layer.8.attention.self.query | NNCFLinear | weight | [448, 768] | 344064 | 129148 | 0.62464 |
| 135 | nncf_module.bert.encoder.layer.8.attention.self.key | NNCFLinear | weight | [448, 768] | 344064 | 130060 | 0.621989 |
| 137 | nncf_module.bert.encoder.layer.8.attention.self.value | NNCFLinear | weight | [448, 768] | 344064 | 108162 | 0.685634 |
| 139 | nncf_module.bert.encoder.layer.8.attention.output.dense | NNCFLinear | weight | [768, 448] | 344064 | 103447 | 0.699338 |
| 143 | nncf_module.bert.encoder.layer.8.intermediate.dense | NNCFLinear | weight | [108, 768] | 82944 | 63275 | 0.237136 |
| 145 | nncf_module.bert.encoder.layer.8.output.dense | NNCFLinear | weight | [768, 108] | 82944 | 62725 | 0.243767 |
| 149 | nncf_module.bert.encoder.layer.9.attention.self.query | NNCFLinear | weight | [320, 768] | 245760 | 107145 | 0.564026 |
| 151 | nncf_module.bert.encoder.layer.9.attention.self.key | NNCFLinear | weight | [320, 768] | 245760 | 101811 | 0.58573 |
| 153 | nncf_module.bert.encoder.layer.9.attention.self.value | NNCFLinear | weight | [320, 768] | 245760 | 52182 | 0.787671 |
| 155 | nncf_module.bert.encoder.layer.9.attention.output.dense | NNCFLinear | weight | [768, 320] | 245760 | 53210 | 0.783488 |
| 159 | nncf_module.bert.encoder.layer.9.intermediate.dense | NNCFLinear | weight | [53, 768] | 40704 | 33461 | 0.177943 |
| 161 | nncf_module.bert.encoder.layer.9.output.dense | NNCFLinear | weight | [768, 53] | 40704 | 32551 | 0.2003 |
| 165 | nncf_module.bert.encoder.layer.10.attention.self.query | NNCFLinear | weight | [384, 768] | 294912 | 112430 | 0.618768 |
| 167 | nncf_module.bert.encoder.layer.10.attention.self.key | NNCFLinear | weight | [384, 768] | 294912 | 109594 | 0.628384 |
| 169 | nncf_module.bert.encoder.layer.10.attention.self.value | NNCFLinear | weight | [384, 768] | 294912 | 61774 | 0.790534 |
| 171 | nncf_module.bert.encoder.layer.10.attention.output.dense | NNCFLinear | weight | [768, 384] | 294912 | 64183 | 0.782366 |
| 175 | nncf_module.bert.encoder.layer.10.intermediate.dense | NNCFLinear | weight | [86, 768] | 66048 | 50455 | 0.236086 |
| 177 | nncf_module.bert.encoder.layer.10.output.dense | NNCFLinear | weight | [768, 86] | 66048 | 49741 | 0.246896 |
| 181 | nncf_module.bert.encoder.layer.11.attention.self.query | NNCFLinear | weight | [384, 768] | 294912 | 88129 | 0.701168 |
| 183 | nncf_module.bert.encoder.layer.11.attention.self.key | NNCFLinear | weight | [384, 768] | 294912 | 85288 | 0.710802 |
| 185 | nncf_module.bert.encoder.layer.11.attention.self.value | NNCFLinear | weight | [384, 768] | 294912 | 47258 | 0.839756 |
| 187 | nncf_module.bert.encoder.layer.11.attention.output.dense | NNCFLinear | weight | [768, 384] | 294912 | 49311 | 0.832794 |
| 191 | nncf_module.bert.encoder.layer.11.intermediate.dense | NNCFLinear | weight | [105, 768] | 80640 | 62254 | 0.228001 |
| 193 | nncf_module.bert.encoder.layer.11.output.dense | NNCFLinear | weight | [768, 105] | 80640 | 61669 | 0.235255 |