Upload 5 files
Browse files- config.json +706 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -0
- vocab.txt +0 -0
config.json
ADDED
|
@@ -0,0 +1,706 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/mnt/dolphinfs/ssd_pool/docker/user/hadoop-aipnlp/zhangchen76/afo_outputs/sparsebert_hidden_minilm_0.8,0.9S_1/ckpt-last",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"SparseBertHiddenMiniLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"hidden_size": 768,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 3072,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"layer_skip": 1,
|
| 16 |
+
"max_position_embeddings": 512,
|
| 17 |
+
"model_type": "bert",
|
| 18 |
+
"num_attention_heads": 12,
|
| 19 |
+
"num_hidden_layers": 12,
|
| 20 |
+
"num_relation_heads": 32,
|
| 21 |
+
"pad_token_id": 0,
|
| 22 |
+
"position_embedding_type": "absolute",
|
| 23 |
+
"sparsity": "90",
|
| 24 |
+
"sparsity_map": {
|
| 25 |
+
"0": {
|
| 26 |
+
"head": {},
|
| 27 |
+
"hidden": {
|
| 28 |
+
"-1": 0,
|
| 29 |
+
"0": 0,
|
| 30 |
+
"1": 0,
|
| 31 |
+
"10": 0,
|
| 32 |
+
"11": 0,
|
| 33 |
+
"2": 0,
|
| 34 |
+
"3": 0,
|
| 35 |
+
"4": 0,
|
| 36 |
+
"5": 0,
|
| 37 |
+
"6": 0,
|
| 38 |
+
"7": 0,
|
| 39 |
+
"8": 0,
|
| 40 |
+
"9": 0
|
| 41 |
+
},
|
| 42 |
+
"neuron": {}
|
| 43 |
+
},
|
| 44 |
+
"10": {
|
| 45 |
+
"head": {
|
| 46 |
+
"1": 5,
|
| 47 |
+
"11": 1,
|
| 48 |
+
"3": 1
|
| 49 |
+
},
|
| 50 |
+
"hidden": {
|
| 51 |
+
"-1": 38,
|
| 52 |
+
"0": 38,
|
| 53 |
+
"1": 38,
|
| 54 |
+
"10": 38,
|
| 55 |
+
"11": 38,
|
| 56 |
+
"2": 38,
|
| 57 |
+
"3": 38,
|
| 58 |
+
"4": 38,
|
| 59 |
+
"5": 38,
|
| 60 |
+
"6": 38,
|
| 61 |
+
"7": 38,
|
| 62 |
+
"8": 38,
|
| 63 |
+
"9": 38
|
| 64 |
+
},
|
| 65 |
+
"neuron": {
|
| 66 |
+
"0": 190,
|
| 67 |
+
"1": 298,
|
| 68 |
+
"11": 1,
|
| 69 |
+
"2": 1323,
|
| 70 |
+
"3": 4,
|
| 71 |
+
"4": 5,
|
| 72 |
+
"5": 8,
|
| 73 |
+
"6": 4,
|
| 74 |
+
"7": 7,
|
| 75 |
+
"8": 2,
|
| 76 |
+
"9": 1
|
| 77 |
+
}
|
| 78 |
+
},
|
| 79 |
+
"20": {
|
| 80 |
+
"head": {
|
| 81 |
+
"0": 1,
|
| 82 |
+
"1": 7,
|
| 83 |
+
"10": 1,
|
| 84 |
+
"11": 4,
|
| 85 |
+
"3": 2,
|
| 86 |
+
"9": 1
|
| 87 |
+
},
|
| 88 |
+
"hidden": {
|
| 89 |
+
"-1": 84,
|
| 90 |
+
"0": 84,
|
| 91 |
+
"1": 84,
|
| 92 |
+
"10": 84,
|
| 93 |
+
"11": 84,
|
| 94 |
+
"2": 84,
|
| 95 |
+
"3": 84,
|
| 96 |
+
"4": 84,
|
| 97 |
+
"5": 84,
|
| 98 |
+
"6": 84,
|
| 99 |
+
"7": 84,
|
| 100 |
+
"8": 84,
|
| 101 |
+
"9": 84
|
| 102 |
+
},
|
| 103 |
+
"neuron": {
|
| 104 |
+
"0": 597,
|
| 105 |
+
"1": 802,
|
| 106 |
+
"11": 16,
|
| 107 |
+
"2": 2294,
|
| 108 |
+
"3": 48,
|
| 109 |
+
"4": 78,
|
| 110 |
+
"5": 123,
|
| 111 |
+
"6": 22,
|
| 112 |
+
"7": 47,
|
| 113 |
+
"8": 23,
|
| 114 |
+
"9": 5
|
| 115 |
+
}
|
| 116 |
+
},
|
| 117 |
+
"30": {
|
| 118 |
+
"head": {
|
| 119 |
+
"0": 1,
|
| 120 |
+
"1": 8,
|
| 121 |
+
"10": 2,
|
| 122 |
+
"11": 4,
|
| 123 |
+
"2": 1,
|
| 124 |
+
"3": 2,
|
| 125 |
+
"6": 1,
|
| 126 |
+
"7": 1,
|
| 127 |
+
"8": 1,
|
| 128 |
+
"9": 2
|
| 129 |
+
},
|
| 130 |
+
"hidden": {
|
| 131 |
+
"-1": 123,
|
| 132 |
+
"0": 123,
|
| 133 |
+
"1": 123,
|
| 134 |
+
"10": 123,
|
| 135 |
+
"11": 123,
|
| 136 |
+
"2": 123,
|
| 137 |
+
"3": 123,
|
| 138 |
+
"4": 123,
|
| 139 |
+
"5": 123,
|
| 140 |
+
"6": 123,
|
| 141 |
+
"7": 123,
|
| 142 |
+
"8": 123,
|
| 143 |
+
"9": 123
|
| 144 |
+
},
|
| 145 |
+
"neuron": {
|
| 146 |
+
"0": 980,
|
| 147 |
+
"1": 1233,
|
| 148 |
+
"10": 2,
|
| 149 |
+
"11": 116,
|
| 150 |
+
"2": 2512,
|
| 151 |
+
"3": 132,
|
| 152 |
+
"4": 233,
|
| 153 |
+
"5": 386,
|
| 154 |
+
"6": 61,
|
| 155 |
+
"7": 146,
|
| 156 |
+
"8": 73,
|
| 157 |
+
"9": 24
|
| 158 |
+
}
|
| 159 |
+
},
|
| 160 |
+
"40": {
|
| 161 |
+
"head": {
|
| 162 |
+
"0": 3,
|
| 163 |
+
"1": 8,
|
| 164 |
+
"10": 2,
|
| 165 |
+
"11": 5,
|
| 166 |
+
"2": 1,
|
| 167 |
+
"3": 3,
|
| 168 |
+
"4": 2,
|
| 169 |
+
"5": 1,
|
| 170 |
+
"6": 2,
|
| 171 |
+
"7": 1,
|
| 172 |
+
"8": 1,
|
| 173 |
+
"9": 4
|
| 174 |
+
},
|
| 175 |
+
"hidden": {
|
| 176 |
+
"-1": 177,
|
| 177 |
+
"0": 177,
|
| 178 |
+
"1": 177,
|
| 179 |
+
"10": 177,
|
| 180 |
+
"11": 177,
|
| 181 |
+
"2": 177,
|
| 182 |
+
"3": 177,
|
| 183 |
+
"4": 177,
|
| 184 |
+
"5": 177,
|
| 185 |
+
"6": 177,
|
| 186 |
+
"7": 177,
|
| 187 |
+
"8": 177,
|
| 188 |
+
"9": 177
|
| 189 |
+
},
|
| 190 |
+
"neuron": {
|
| 191 |
+
"0": 1436,
|
| 192 |
+
"1": 1664,
|
| 193 |
+
"10": 6,
|
| 194 |
+
"11": 518,
|
| 195 |
+
"2": 2634,
|
| 196 |
+
"3": 306,
|
| 197 |
+
"4": 449,
|
| 198 |
+
"5": 732,
|
| 199 |
+
"6": 138,
|
| 200 |
+
"7": 321,
|
| 201 |
+
"8": 184,
|
| 202 |
+
"9": 91
|
| 203 |
+
}
|
| 204 |
+
},
|
| 205 |
+
"50": {
|
| 206 |
+
"head": {
|
| 207 |
+
"0": 5,
|
| 208 |
+
"1": 8,
|
| 209 |
+
"10": 3,
|
| 210 |
+
"11": 5,
|
| 211 |
+
"2": 2,
|
| 212 |
+
"3": 3,
|
| 213 |
+
"4": 3,
|
| 214 |
+
"5": 2,
|
| 215 |
+
"6": 3,
|
| 216 |
+
"7": 2,
|
| 217 |
+
"8": 2,
|
| 218 |
+
"9": 4
|
| 219 |
+
},
|
| 220 |
+
"hidden": {
|
| 221 |
+
"-1": 223,
|
| 222 |
+
"0": 223,
|
| 223 |
+
"1": 223,
|
| 224 |
+
"10": 223,
|
| 225 |
+
"11": 223,
|
| 226 |
+
"2": 223,
|
| 227 |
+
"3": 223,
|
| 228 |
+
"4": 223,
|
| 229 |
+
"5": 223,
|
| 230 |
+
"6": 223,
|
| 231 |
+
"7": 223,
|
| 232 |
+
"8": 223,
|
| 233 |
+
"9": 223
|
| 234 |
+
},
|
| 235 |
+
"neuron": {
|
| 236 |
+
"0": 1749,
|
| 237 |
+
"1": 1933,
|
| 238 |
+
"10": 21,
|
| 239 |
+
"11": 915,
|
| 240 |
+
"2": 2699,
|
| 241 |
+
"3": 519,
|
| 242 |
+
"4": 664,
|
| 243 |
+
"5": 976,
|
| 244 |
+
"6": 252,
|
| 245 |
+
"7": 469,
|
| 246 |
+
"8": 307,
|
| 247 |
+
"9": 187
|
| 248 |
+
}
|
| 249 |
+
},
|
| 250 |
+
"60": {
|
| 251 |
+
"head": {
|
| 252 |
+
"0": 6,
|
| 253 |
+
"1": 8,
|
| 254 |
+
"10": 5,
|
| 255 |
+
"11": 7,
|
| 256 |
+
"2": 3,
|
| 257 |
+
"3": 4,
|
| 258 |
+
"4": 4,
|
| 259 |
+
"5": 3,
|
| 260 |
+
"6": 5,
|
| 261 |
+
"7": 2,
|
| 262 |
+
"8": 2,
|
| 263 |
+
"9": 4
|
| 264 |
+
},
|
| 265 |
+
"hidden": {
|
| 266 |
+
"-1": 284,
|
| 267 |
+
"0": 284,
|
| 268 |
+
"1": 284,
|
| 269 |
+
"10": 284,
|
| 270 |
+
"11": 284,
|
| 271 |
+
"2": 284,
|
| 272 |
+
"3": 284,
|
| 273 |
+
"4": 284,
|
| 274 |
+
"5": 284,
|
| 275 |
+
"6": 284,
|
| 276 |
+
"7": 284,
|
| 277 |
+
"8": 284,
|
| 278 |
+
"9": 284
|
| 279 |
+
},
|
| 280 |
+
"neuron": {
|
| 281 |
+
"0": 2065,
|
| 282 |
+
"1": 2200,
|
| 283 |
+
"10": 67,
|
| 284 |
+
"11": 1392,
|
| 285 |
+
"2": 2762,
|
| 286 |
+
"3": 817,
|
| 287 |
+
"4": 991,
|
| 288 |
+
"5": 1279,
|
| 289 |
+
"6": 454,
|
| 290 |
+
"7": 695,
|
| 291 |
+
"8": 521,
|
| 292 |
+
"9": 397
|
| 293 |
+
}
|
| 294 |
+
},
|
| 295 |
+
"70": {
|
| 296 |
+
"head": {
|
| 297 |
+
"0": 6,
|
| 298 |
+
"1": 8,
|
| 299 |
+
"10": 6,
|
| 300 |
+
"11": 7,
|
| 301 |
+
"2": 4,
|
| 302 |
+
"3": 6,
|
| 303 |
+
"4": 4,
|
| 304 |
+
"5": 5,
|
| 305 |
+
"6": 6,
|
| 306 |
+
"7": 3,
|
| 307 |
+
"8": 5,
|
| 308 |
+
"9": 5
|
| 309 |
+
},
|
| 310 |
+
"hidden": {
|
| 311 |
+
"-1": 346,
|
| 312 |
+
"0": 346,
|
| 313 |
+
"1": 346,
|
| 314 |
+
"10": 346,
|
| 315 |
+
"11": 346,
|
| 316 |
+
"2": 346,
|
| 317 |
+
"3": 346,
|
| 318 |
+
"4": 346,
|
| 319 |
+
"5": 346,
|
| 320 |
+
"6": 346,
|
| 321 |
+
"7": 346,
|
| 322 |
+
"8": 346,
|
| 323 |
+
"9": 346
|
| 324 |
+
},
|
| 325 |
+
"neuron": {
|
| 326 |
+
"0": 2266,
|
| 327 |
+
"1": 2390,
|
| 328 |
+
"10": 206,
|
| 329 |
+
"11": 1745,
|
| 330 |
+
"2": 2820,
|
| 331 |
+
"3": 1173,
|
| 332 |
+
"4": 1308,
|
| 333 |
+
"5": 1537,
|
| 334 |
+
"6": 694,
|
| 335 |
+
"7": 951,
|
| 336 |
+
"8": 791,
|
| 337 |
+
"9": 708
|
| 338 |
+
}
|
| 339 |
+
},
|
| 340 |
+
"80": {
|
| 341 |
+
"head": {
|
| 342 |
+
"0": 6,
|
| 343 |
+
"1": 9,
|
| 344 |
+
"10": 8,
|
| 345 |
+
"11": 8,
|
| 346 |
+
"2": 6,
|
| 347 |
+
"3": 6,
|
| 348 |
+
"4": 6,
|
| 349 |
+
"5": 9,
|
| 350 |
+
"6": 7,
|
| 351 |
+
"7": 4,
|
| 352 |
+
"8": 5,
|
| 353 |
+
"9": 5
|
| 354 |
+
},
|
| 355 |
+
"hidden": {
|
| 356 |
+
"-1": 422,
|
| 357 |
+
"0": 422,
|
| 358 |
+
"1": 422,
|
| 359 |
+
"10": 422,
|
| 360 |
+
"11": 422,
|
| 361 |
+
"2": 422,
|
| 362 |
+
"3": 422,
|
| 363 |
+
"4": 422,
|
| 364 |
+
"5": 422,
|
| 365 |
+
"6": 422,
|
| 366 |
+
"7": 422,
|
| 367 |
+
"8": 422,
|
| 368 |
+
"9": 422
|
| 369 |
+
},
|
| 370 |
+
"neuron": {
|
| 371 |
+
"0": 2450,
|
| 372 |
+
"1": 2541,
|
| 373 |
+
"10": 574,
|
| 374 |
+
"11": 2105,
|
| 375 |
+
"2": 2863,
|
| 376 |
+
"3": 1559,
|
| 377 |
+
"4": 1676,
|
| 378 |
+
"5": 1862,
|
| 379 |
+
"6": 1071,
|
| 380 |
+
"7": 1294,
|
| 381 |
+
"8": 1154,
|
| 382 |
+
"9": 1126
|
| 383 |
+
}
|
| 384 |
+
},
|
| 385 |
+
"85": {
|
| 386 |
+
"head": {
|
| 387 |
+
"0": 6,
|
| 388 |
+
"1": 9,
|
| 389 |
+
"10": 10,
|
| 390 |
+
"11": 8,
|
| 391 |
+
"2": 6,
|
| 392 |
+
"3": 6,
|
| 393 |
+
"4": 7,
|
| 394 |
+
"5": 9,
|
| 395 |
+
"6": 8,
|
| 396 |
+
"7": 7,
|
| 397 |
+
"8": 5,
|
| 398 |
+
"9": 7
|
| 399 |
+
},
|
| 400 |
+
"hidden": {
|
| 401 |
+
"-1": 468,
|
| 402 |
+
"0": 468,
|
| 403 |
+
"1": 468,
|
| 404 |
+
"10": 468,
|
| 405 |
+
"11": 468,
|
| 406 |
+
"2": 468,
|
| 407 |
+
"3": 468,
|
| 408 |
+
"4": 468,
|
| 409 |
+
"5": 468,
|
| 410 |
+
"6": 468,
|
| 411 |
+
"7": 468,
|
| 412 |
+
"8": 468,
|
| 413 |
+
"9": 468
|
| 414 |
+
},
|
| 415 |
+
"neuron": {
|
| 416 |
+
"0": 2554,
|
| 417 |
+
"1": 2635,
|
| 418 |
+
"10": 936,
|
| 419 |
+
"11": 2278,
|
| 420 |
+
"2": 2889,
|
| 421 |
+
"3": 1780,
|
| 422 |
+
"4": 1836,
|
| 423 |
+
"5": 2037,
|
| 424 |
+
"6": 1292,
|
| 425 |
+
"7": 1489,
|
| 426 |
+
"8": 1360,
|
| 427 |
+
"9": 1401
|
| 428 |
+
}
|
| 429 |
+
},
|
| 430 |
+
"90": {
|
| 431 |
+
"head": {
|
| 432 |
+
"0": 6,
|
| 433 |
+
"1": 9,
|
| 434 |
+
"10": 11,
|
| 435 |
+
"11": 8,
|
| 436 |
+
"2": 7,
|
| 437 |
+
"3": 7,
|
| 438 |
+
"4": 9,
|
| 439 |
+
"5": 9,
|
| 440 |
+
"6": 9,
|
| 441 |
+
"7": 9,
|
| 442 |
+
"8": 6,
|
| 443 |
+
"9": 8
|
| 444 |
+
},
|
| 445 |
+
"hidden": {
|
| 446 |
+
"-1": 522,
|
| 447 |
+
"0": 522,
|
| 448 |
+
"1": 522,
|
| 449 |
+
"10": 522,
|
| 450 |
+
"11": 522,
|
| 451 |
+
"2": 522,
|
| 452 |
+
"3": 522,
|
| 453 |
+
"4": 522,
|
| 454 |
+
"5": 522,
|
| 455 |
+
"6": 522,
|
| 456 |
+
"7": 522,
|
| 457 |
+
"8": 522,
|
| 458 |
+
"9": 522
|
| 459 |
+
},
|
| 460 |
+
"neuron": {
|
| 461 |
+
"0": 2653,
|
| 462 |
+
"1": 2713,
|
| 463 |
+
"10": 1390,
|
| 464 |
+
"11": 2433,
|
| 465 |
+
"2": 2913,
|
| 466 |
+
"3": 2013,
|
| 467 |
+
"4": 2064,
|
| 468 |
+
"5": 2243,
|
| 469 |
+
"6": 1569,
|
| 470 |
+
"7": 1741,
|
| 471 |
+
"8": 1648,
|
| 472 |
+
"9": 1688
|
| 473 |
+
}
|
| 474 |
+
},
|
| 475 |
+
"95": {
|
| 476 |
+
"head": {
|
| 477 |
+
"0": 8,
|
| 478 |
+
"1": 10,
|
| 479 |
+
"10": 11,
|
| 480 |
+
"11": 9,
|
| 481 |
+
"2": 9,
|
| 482 |
+
"3": 9,
|
| 483 |
+
"4": 9,
|
| 484 |
+
"5": 9,
|
| 485 |
+
"6": 11,
|
| 486 |
+
"7": 9,
|
| 487 |
+
"8": 9,
|
| 488 |
+
"9": 9
|
| 489 |
+
},
|
| 490 |
+
"hidden": {
|
| 491 |
+
"-1": 599,
|
| 492 |
+
"0": 599,
|
| 493 |
+
"1": 599,
|
| 494 |
+
"10": 599,
|
| 495 |
+
"11": 599,
|
| 496 |
+
"2": 599,
|
| 497 |
+
"3": 599,
|
| 498 |
+
"4": 599,
|
| 499 |
+
"5": 599,
|
| 500 |
+
"6": 599,
|
| 501 |
+
"7": 599,
|
| 502 |
+
"8": 599,
|
| 503 |
+
"9": 599
|
| 504 |
+
},
|
| 505 |
+
"neuron": {
|
| 506 |
+
"0": 2753,
|
| 507 |
+
"1": 2799,
|
| 508 |
+
"10": 2001,
|
| 509 |
+
"11": 2654,
|
| 510 |
+
"2": 2948,
|
| 511 |
+
"3": 2337,
|
| 512 |
+
"4": 2381,
|
| 513 |
+
"5": 2519,
|
| 514 |
+
"6": 1992,
|
| 515 |
+
"7": 2148,
|
| 516 |
+
"8": 2061,
|
| 517 |
+
"9": 2161
|
| 518 |
+
}
|
| 519 |
+
},
|
| 520 |
+
"96": {
|
| 521 |
+
"head": {
|
| 522 |
+
"0": 9,
|
| 523 |
+
"1": 10,
|
| 524 |
+
"10": 11,
|
| 525 |
+
"11": 10,
|
| 526 |
+
"2": 9,
|
| 527 |
+
"3": 9,
|
| 528 |
+
"4": 9,
|
| 529 |
+
"5": 10,
|
| 530 |
+
"6": 11,
|
| 531 |
+
"7": 9,
|
| 532 |
+
"8": 9,
|
| 533 |
+
"9": 9
|
| 534 |
+
},
|
| 535 |
+
"hidden": {
|
| 536 |
+
"-1": 614,
|
| 537 |
+
"0": 614,
|
| 538 |
+
"1": 614,
|
| 539 |
+
"10": 614,
|
| 540 |
+
"11": 614,
|
| 541 |
+
"2": 614,
|
| 542 |
+
"3": 614,
|
| 543 |
+
"4": 614,
|
| 544 |
+
"5": 614,
|
| 545 |
+
"6": 614,
|
| 546 |
+
"7": 614,
|
| 547 |
+
"8": 614,
|
| 548 |
+
"9": 614
|
| 549 |
+
},
|
| 550 |
+
"neuron": {
|
| 551 |
+
"0": 2768,
|
| 552 |
+
"1": 2822,
|
| 553 |
+
"10": 2133,
|
| 554 |
+
"11": 2689,
|
| 555 |
+
"2": 2955,
|
| 556 |
+
"3": 2392,
|
| 557 |
+
"4": 2439,
|
| 558 |
+
"5": 2579,
|
| 559 |
+
"6": 2077,
|
| 560 |
+
"7": 2223,
|
| 561 |
+
"8": 2148,
|
| 562 |
+
"9": 2266
|
| 563 |
+
}
|
| 564 |
+
},
|
| 565 |
+
"97": {
|
| 566 |
+
"head": {
|
| 567 |
+
"0": 10,
|
| 568 |
+
"1": 10,
|
| 569 |
+
"10": 11,
|
| 570 |
+
"11": 10,
|
| 571 |
+
"2": 9,
|
| 572 |
+
"3": 9,
|
| 573 |
+
"4": 9,
|
| 574 |
+
"5": 10,
|
| 575 |
+
"6": 11,
|
| 576 |
+
"7": 10,
|
| 577 |
+
"8": 11,
|
| 578 |
+
"9": 10
|
| 579 |
+
},
|
| 580 |
+
"hidden": {
|
| 581 |
+
"-1": 637,
|
| 582 |
+
"0": 637,
|
| 583 |
+
"1": 637,
|
| 584 |
+
"10": 637,
|
| 585 |
+
"11": 637,
|
| 586 |
+
"2": 637,
|
| 587 |
+
"3": 637,
|
| 588 |
+
"4": 637,
|
| 589 |
+
"5": 637,
|
| 590 |
+
"6": 637,
|
| 591 |
+
"7": 637,
|
| 592 |
+
"8": 637,
|
| 593 |
+
"9": 637
|
| 594 |
+
},
|
| 595 |
+
"neuron": {
|
| 596 |
+
"0": 2799,
|
| 597 |
+
"1": 2846,
|
| 598 |
+
"10": 2294,
|
| 599 |
+
"11": 2746,
|
| 600 |
+
"2": 2970,
|
| 601 |
+
"3": 2487,
|
| 602 |
+
"4": 2537,
|
| 603 |
+
"5": 2656,
|
| 604 |
+
"6": 2207,
|
| 605 |
+
"7": 2355,
|
| 606 |
+
"8": 2291,
|
| 607 |
+
"9": 2409
|
| 608 |
+
}
|
| 609 |
+
},
|
| 610 |
+
"98": {
|
| 611 |
+
"head": {
|
| 612 |
+
"0": 11,
|
| 613 |
+
"1": 10,
|
| 614 |
+
"10": 11,
|
| 615 |
+
"11": 10,
|
| 616 |
+
"2": 10,
|
| 617 |
+
"3": 10,
|
| 618 |
+
"4": 10,
|
| 619 |
+
"5": 10,
|
| 620 |
+
"6": 11,
|
| 621 |
+
"7": 10,
|
| 622 |
+
"8": 11,
|
| 623 |
+
"9": 10
|
| 624 |
+
},
|
| 625 |
+
"hidden": {
|
| 626 |
+
"-1": 660,
|
| 627 |
+
"0": 660,
|
| 628 |
+
"1": 660,
|
| 629 |
+
"10": 660,
|
| 630 |
+
"11": 660,
|
| 631 |
+
"2": 660,
|
| 632 |
+
"3": 660,
|
| 633 |
+
"4": 660,
|
| 634 |
+
"5": 660,
|
| 635 |
+
"6": 660,
|
| 636 |
+
"7": 660,
|
| 637 |
+
"8": 660,
|
| 638 |
+
"9": 660
|
| 639 |
+
},
|
| 640 |
+
"neuron": {
|
| 641 |
+
"0": 2826,
|
| 642 |
+
"1": 2870,
|
| 643 |
+
"10": 2455,
|
| 644 |
+
"11": 2794,
|
| 645 |
+
"2": 2984,
|
| 646 |
+
"3": 2588,
|
| 647 |
+
"4": 2634,
|
| 648 |
+
"5": 2726,
|
| 649 |
+
"6": 2350,
|
| 650 |
+
"7": 2501,
|
| 651 |
+
"8": 2440,
|
| 652 |
+
"9": 2535
|
| 653 |
+
}
|
| 654 |
+
},
|
| 655 |
+
"99": {
|
| 656 |
+
"head": {
|
| 657 |
+
"0": 11,
|
| 658 |
+
"1": 10,
|
| 659 |
+
"10": 11,
|
| 660 |
+
"11": 11,
|
| 661 |
+
"2": 11,
|
| 662 |
+
"3": 11,
|
| 663 |
+
"4": 11,
|
| 664 |
+
"5": 11,
|
| 665 |
+
"6": 11,
|
| 666 |
+
"7": 11,
|
| 667 |
+
"8": 11,
|
| 668 |
+
"9": 10
|
| 669 |
+
},
|
| 670 |
+
"hidden": {
|
| 671 |
+
"-1": 691,
|
| 672 |
+
"0": 691,
|
| 673 |
+
"1": 691,
|
| 674 |
+
"10": 691,
|
| 675 |
+
"11": 691,
|
| 676 |
+
"2": 691,
|
| 677 |
+
"3": 691,
|
| 678 |
+
"4": 691,
|
| 679 |
+
"5": 691,
|
| 680 |
+
"6": 691,
|
| 681 |
+
"7": 691,
|
| 682 |
+
"8": 691,
|
| 683 |
+
"9": 691
|
| 684 |
+
},
|
| 685 |
+
"neuron": {
|
| 686 |
+
"0": 2861,
|
| 687 |
+
"1": 2915,
|
| 688 |
+
"10": 2683,
|
| 689 |
+
"11": 2862,
|
| 690 |
+
"2": 2998,
|
| 691 |
+
"3": 2709,
|
| 692 |
+
"4": 2746,
|
| 693 |
+
"5": 2845,
|
| 694 |
+
"6": 2551,
|
| 695 |
+
"7": 2678,
|
| 696 |
+
"8": 2635,
|
| 697 |
+
"9": 2695
|
| 698 |
+
}
|
| 699 |
+
}
|
| 700 |
+
},
|
| 701 |
+
"torch_dtype": "float32",
|
| 702 |
+
"transformers_version": "4.9.2",
|
| 703 |
+
"type_vocab_size": 2,
|
| 704 |
+
"use_cache": true,
|
| 705 |
+
"vocab_size": 30522
|
| 706 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e12ad6d7600e11f76aae2c8d3e4ec577c55f74fac5a9ba60cfcc5a418874c9ca
|
| 3 |
+
size 75559930
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"do_lower_case": true, "do_basic_tokenize": true, "never_split": null, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "/mnt/dolphinfs/ssd_pool/docker/user/hadoop-aipnlp/zhangchen76/afo_outputs/sparsebert_hidden_minilm_0.8,0.9S_1/ckpt-last", "use_fast": true, "tokenizer_file": null, "tokenizer_class": "BertTokenizer"}
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|