Hugging Face's logo

thrunlab
/
Mistral_Sparse_refined_web_relu_2024-03-01

Text Generation
Transformers
Safetensors
sparse_mistral
Generated from Trainer
custom_code
Model card Files Community
vxbrandon commited on
Commit
4d2538f
•
1 Parent(s): 4aded55

End of training

Browse files
Files changed (4) hide show
  1. README.md +8 -209
  2. adapter_config.json +3 -3
  3. adapter_model.safetensors +2 -2
  4. training_args.bin +2 -2
README.md CHANGED
@@ -3,6 +3,8 @@ license: apache-2.0
3
  base_model: mistralai/Mistral-7B-v0.1
4
  tags:
5
  - generated_from_trainer
 
 
6
  model-index:
7
  - name: Mistral_Sparse_refined_web_relu_2024-03-01
8
  results: []
@@ -13,9 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
13
 
14
  # Mistral_Sparse_refined_web_relu_2024-03-01
15
 
16
- This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
17
- It achieves the following results on the evaluation set:
18
- - Loss: 2.3329
19
 
20
  ## Model description
21
 
@@ -34,214 +34,13 @@ More information needed
34
  ### Training hyperparameters
35
 
36
  The following hyperparameters were used during training:
37
- - learning_rate: 1e-05
38
- - train_batch_size: 1
39
- - eval_batch_size: 1
40
- - seed: 0
41
- - distributed_type: multi-GPU
42
- - num_devices: 4
43
- - gradient_accumulation_steps: 4
44
- - total_train_batch_size: 16
45
- - total_eval_batch_size: 4
46
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
47
  - lr_scheduler_type: linear
48
- - training_steps: 4750
49
-
50
- ### Training results
51
-
52
- | Training Loss | Epoch | Step | Validation Loss |
53
- |:-------------:|:-----:|:----:|:---------------:|
54
- | 8.7893 | 0.0 | 25 | 8.7124 |
55
- | 8.1852 | 0.01 | 50 | 8.1766 |
56
- | 7.7552 | 0.01 | 75 | 7.8055 |
57
- | 7.5347 | 0.02 | 100 | 7.6044 |
58
- | 7.1789 | 0.02 | 125 | 7.1167 |
59
- | 4.7625 | 0.02 | 150 | 4.7156 |
60
- | 3.7491 | 0.03 | 175 | 3.8005 |
61
- | 3.2565 | 0.03 | 200 | 3.4172 |
62
- | 3.0841 | 0.04 | 225 | 3.2310 |
63
- | 2.9901 | 0.04 | 250 | 3.1131 |
64
- | 2.8534 | 0.04 | 275 | 3.0370 |
65
- | 2.8256 | 0.05 | 300 | 2.9857 |
66
- | 2.792 | 0.05 | 325 | 2.9492 |
67
- | 2.6055 | 0.06 | 350 | 2.9145 |
68
- | 2.7474 | 0.06 | 375 | 2.8922 |
69
- | 2.7788 | 0.06 | 400 | 2.8716 |
70
- | 2.731 | 0.07 | 425 | 2.8528 |
71
- | 2.6973 | 0.07 | 450 | 2.8403 |
72
- | 2.5808 | 0.08 | 475 | 2.8249 |
73
- | 2.6415 | 0.08 | 500 | 2.8154 |
74
- | 2.718 | 0.08 | 525 | 2.8072 |
75
- | 2.6225 | 0.09 | 550 | 2.7972 |
76
- | 2.7265 | 0.09 | 575 | 2.7863 |
77
- | 2.5424 | 0.1 | 600 | 2.7803 |
78
- | 2.6148 | 0.1 | 625 | 2.7762 |
79
- | 2.6546 | 0.1 | 650 | 2.7655 |
80
- | 2.6728 | 0.11 | 675 | 2.7575 |
81
- | 2.6039 | 0.11 | 700 | 2.7507 |
82
- | 2.6226 | 0.12 | 725 | 2.7488 |
83
- | 2.5294 | 0.12 | 750 | 2.7449 |
84
- | 2.5327 | 0.12 | 775 | 2.7382 |
85
- | 2.6527 | 0.13 | 800 | 2.7354 |
86
- | 2.6078 | 0.13 | 825 | 2.7304 |
87
- | 2.5924 | 0.14 | 850 | 2.7272 |
88
- | 2.4548 | 0.14 | 875 | 2.7242 |
89
- | 2.5582 | 0.14 | 900 | 2.7247 |
90
- | 2.494 | 0.15 | 925 | 2.7189 |
91
- | 2.6211 | 0.15 | 950 | 2.7179 |
92
- | 2.5384 | 0.16 | 975 | 2.7137 |
93
- | 2.5729 | 0.16 | 1000 | 2.7129 |
94
- | 2.6008 | 0.16 | 1025 | 2.7106 |
95
- | 2.63 | 0.17 | 1050 | 2.7072 |
96
- | 2.6351 | 0.17 | 1075 | 2.7037 |
97
- | 2.554 | 0.18 | 1100 | 2.7032 |
98
- | 2.5746 | 0.18 | 1125 | 2.7021 |
99
- | 2.5616 | 0.18 | 1150 | 2.6981 |
100
- | 2.5972 | 0.19 | 1175 | 2.6957 |
101
- | 2.4717 | 0.19 | 1200 | 2.6958 |
102
- | 2.5365 | 0.2 | 1225 | 2.6934 |
103
- | 2.5518 | 0.2 | 1250 | 2.6907 |
104
- | 2.5734 | 0.2 | 1275 | 2.6896 |
105
- | 2.5785 | 0.21 | 1300 | 2.6866 |
106
- | 2.5566 | 0.21 | 1325 | 2.6825 |
107
- | 2.441 | 0.22 | 1350 | 2.6804 |
108
- | 2.5228 | 0.22 | 1375 | 2.6802 |
109
- | 2.5348 | 0.22 | 1400 | 2.6798 |
110
- | 2.4799 | 0.23 | 1425 | 2.6781 |
111
- | 2.4588 | 0.23 | 1450 | 2.6793 |
112
- | 2.4106 | 0.24 | 1475 | 2.6758 |
113
- | 2.4957 | 0.24 | 1500 | 2.6790 |
114
- | 2.4221 | 0.24 | 1525 | 2.6765 |
115
- | 2.4654 | 0.25 | 1550 | 2.6730 |
116
- | 2.5873 | 0.25 | 1575 | 2.6755 |
117
- | 2.5377 | 0.26 | 1600 | 2.6697 |
118
- | 2.5468 | 0.26 | 1625 | 2.6713 |
119
- | 2.5882 | 0.26 | 1650 | 2.6702 |
120
- | 2.4971 | 0.27 | 1675 | 2.6666 |
121
- | 2.5402 | 0.27 | 1700 | 2.6676 |
122
- | 2.5384 | 0.28 | 1725 | 2.6654 |
123
- | 2.5233 | 0.28 | 1750 | 2.6661 |
124
- | 2.5066 | 0.28 | 1775 | 2.6622 |
125
- | 2.4284 | 0.29 | 1800 | 2.6623 |
126
- | 2.5244 | 0.29 | 1825 | 2.6618 |
127
- | 2.5192 | 0.3 | 1850 | 2.6600 |
128
- | 2.5019 | 0.3 | 1875 | 2.6606 |
129
- | 2.5243 | 0.3 | 1900 | 2.6596 |
130
- | 2.5779 | 0.31 | 1925 | 2.6597 |
131
- | 2.5309 | 0.31 | 1950 | 2.6591 |
132
- | 2.5095 | 0.32 | 1975 | 2.6592 |
133
- | 2.4623 | 0.32 | 2000 | 2.6600 |
134
- | 2.5021 | 0.32 | 2025 | 2.6577 |
135
- | 2.4794 | 0.33 | 2050 | 2.6531 |
136
- | 2.4743 | 0.33 | 2075 | 2.6536 |
137
- | 2.5537 | 0.34 | 2100 | 2.6539 |
138
- | 2.5996 | 0.34 | 2125 | 2.6539 |
139
- | 2.5221 | 0.34 | 2150 | 2.6537 |
140
- | 2.4528 | 0.35 | 2175 | 2.6497 |
141
- | 2.5035 | 0.35 | 2200 | 2.6475 |
142
- | 2.4639 | 0.36 | 2225 | 2.6482 |
143
- | 2.457 | 0.36 | 2250 | 2.6482 |
144
- | 2.5714 | 0.36 | 2275 | 2.6467 |
145
- | 2.5397 | 0.37 | 2300 | 2.6461 |
146
- | 2.4343 | 0.37 | 2325 | 2.6459 |
147
- | 2.4829 | 0.38 | 2350 | 2.6468 |
148
- | 2.5359 | 0.38 | 2375 | 2.6457 |
149
- | 2.5172 | 0.38 | 2400 | 2.6463 |
150
- | 2.3848 | 0.39 | 2425 | 2.6472 |
151
- | 2.3813 | 0.39 | 2450 | 2.6469 |
152
- | 2.4746 | 0.4 | 2475 | 2.6450 |
153
- | 2.5501 | 0.4 | 2500 | 2.6440 |
154
- | 2.5325 | 0.4 | 2525 | 2.6437 |
155
- | 2.3975 | 0.41 | 2550 | 2.6410 |
156
- | 2.4247 | 0.41 | 2575 | 2.6402 |
157
- | 2.5072 | 0.42 | 2600 | 2.6428 |
158
- | 2.5073 | 0.42 | 2625 | 2.6420 |
159
- | 2.5514 | 0.42 | 2650 | 2.6393 |
160
- | 2.5216 | 0.43 | 2675 | 2.6389 |
161
- | 2.5287 | 0.43 | 2700 | 2.6398 |
162
- | 2.5277 | 0.44 | 2725 | 2.6406 |
163
- | 2.4907 | 0.44 | 2750 | 2.6388 |
164
- | 2.5073 | 0.44 | 2775 | 2.6387 |
165
- | 2.5895 | 0.45 | 2800 | 2.6403 |
166
- | 2.4773 | 0.45 | 2825 | 2.6396 |
167
- | 2.5233 | 0.46 | 2850 | 2.6389 |
168
- | 2.4071 | 0.46 | 2875 | 2.6391 |
169
- | 2.4482 | 0.46 | 2900 | 2.6370 |
170
- | 2.448 | 0.47 | 2925 | 2.6401 |
171
- | 2.53 | 0.47 | 2950 | 2.6376 |
172
- | 2.4738 | 0.48 | 2975 | 2.6371 |
173
- | 2.527 | 0.48 | 3000 | 2.6397 |
174
- | 2.6096 | 0.48 | 3025 | 2.6365 |
175
- | 2.5199 | 0.49 | 3050 | 2.6336 |
176
- | 2.5394 | 0.49 | 3075 | 2.6369 |
177
- | 2.5508 | 0.5 | 3100 | 2.6347 |
178
- | 2.4674 | 0.5 | 3125 | 2.6350 |
179
- | 2.5266 | 0.5 | 3150 | 2.6322 |
180
- | 2.4413 | 0.51 | 3175 | 2.6321 |
181
- | 2.5058 | 0.51 | 3200 | 2.6328 |
182
- | 2.4239 | 0.52 | 3225 | 2.6321 |
183
- | 2.4125 | 0.52 | 3250 | 2.6324 |
184
- | 2.5025 | 0.52 | 3275 | 2.6314 |
185
- | 2.4606 | 0.53 | 3300 | 2.6307 |
186
- | 2.3506 | 0.53 | 3325 | 2.6285 |
187
- | 2.4761 | 0.54 | 3350 | 2.6282 |
188
- | 2.3911 | 0.54 | 3375 | 2.6316 |
189
- | 2.4107 | 0.54 | 3400 | 2.6287 |
190
- | 2.4333 | 0.55 | 3425 | 2.6285 |
191
- | 2.3588 | 0.55 | 3450 | 2.6285 |
192
- | 2.4693 | 0.56 | 3475 | 2.6294 |
193
- | 2.5011 | 0.56 | 3500 | 2.6305 |
194
- | 2.4835 | 0.56 | 3525 | 2.6284 |
195
- | 2.4571 | 0.57 | 3550 | 2.6303 |
196
- | 2.4615 | 0.57 | 3575 | 2.6306 |
197
- | 2.6016 | 0.58 | 3600 | 2.6292 |
198
- | 2.4092 | 0.58 | 3625 | 2.6283 |
199
- | 2.6206 | 0.58 | 3650 | 2.6306 |
200
- | 2.4158 | 0.59 | 3675 | 2.6294 |
201
- | 2.5485 | 0.59 | 3700 | 2.6307 |
202
- | 2.4632 | 0.6 | 3725 | 2.6298 |
203
- | 2.3924 | 0.6 | 3750 | 2.6312 |
204
- | 2.3574 | 0.6 | 3775 | 2.6293 |
205
- | 2.5634 | 0.61 | 3800 | 2.6278 |
206
- | 2.3534 | 0.61 | 3825 | 2.6280 |
207
- | 2.4695 | 0.62 | 3850 | 2.6271 |
208
- | 2.4102 | 0.62 | 3875 | 2.6299 |
209
- | 2.4975 | 0.62 | 3900 | 2.6270 |
210
- | 2.43 | 0.63 | 3925 | 2.6282 |
211
- | 2.4575 | 0.63 | 3950 | 2.6269 |
212
- | 2.4499 | 0.64 | 3975 | 2.6265 |
213
- | 2.5443 | 0.64 | 4000 | 2.6269 |
214
- | 2.4959 | 0.64 | 4025 | 2.6264 |
215
- | 2.4556 | 0.65 | 4050 | 2.6264 |
216
- | 2.392 | 0.65 | 4075 | 2.6288 |
217
- | 2.4302 | 0.66 | 4100 | 2.6270 |
218
- | 2.4959 | 0.66 | 4125 | 2.6250 |
219
- | 2.4299 | 0.66 | 4150 | 2.6219 |
220
- | 2.3562 | 0.67 | 4175 | 2.6214 |
221
- | 2.506 | 0.67 | 4200 | 2.6239 |
222
- | 2.4648 | 0.68 | 4225 | 2.6220 |
223
- | 2.4507 | 0.68 | 4250 | 2.6230 |
224
- | 2.512 | 0.68 | 4275 | 2.6233 |
225
- | 2.4447 | 0.69 | 4300 | 2.6223 |
226
- | 2.4966 | 0.69 | 4325 | 2.6233 |
227
- | 2.4802 | 0.7 | 4350 | 2.6223 |
228
- | 2.5423 | 0.7 | 4375 | 2.6211 |
229
- | 2.4769 | 0.7 | 4400 | 2.6219 |
230
- | 2.4902 | 0.71 | 4425 | 2.6215 |
231
- | 2.5369 | 0.71 | 4450 | 2.6231 |
232
- | 2.4059 | 0.72 | 4475 | 2.6215 |
233
- | 2.4218 | 0.72 | 4500 | 2.6203 |
234
- | 2.5241 | 0.72 | 4525 | 2.6187 |
235
- | 2.4926 | 0.73 | 4550 | 2.6165 |
236
- | 2.3994 | 0.73 | 4575 | 2.6169 |
237
- | 2.4045 | 0.74 | 4600 | 2.6113 |
238
- | 2.367 | 0.74 | 4625 | 2.6147 |
239
- | 2.5633 | 0.74 | 4650 | 2.6170 |
240
- | 2.4582 | 0.75 | 4675 | 2.6160 |
241
- | 2.4708 | 0.75 | 4700 | 2.6157 |
242
- | 2.4963 | 0.76 | 4725 | 2.6136 |
243
- | 2.4042 | 0.76 | 4750 | 2.6117 |
244
-
245
 
246
  ### Framework versions
247
 
 
3
  base_model: mistralai/Mistral-7B-v0.1
4
  tags:
5
  - generated_from_trainer
6
+ datasets:
7
+ - openwebtext-10k
8
  model-index:
9
  - name: Mistral_Sparse_refined_web_relu_2024-03-01
10
  results: []
 
15
 
16
  # Mistral_Sparse_refined_web_relu_2024-03-01
17
 
18
+ This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the openwebtext-10k dataset.
 
 
19
 
20
  ## Model description
21
 
 
34
  ### Training hyperparameters
35
 
36
  The following hyperparameters were used during training:
37
+ - learning_rate: 5e-05
38
+ - train_batch_size: 32
39
+ - eval_batch_size: 4
40
+ - seed: 42
 
 
 
 
 
41
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
42
  - lr_scheduler_type: linear
43
+ - num_epochs: 3.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44
 
45
  ### Framework versions
46
 
adapter_config.json CHANGED
@@ -20,10 +20,10 @@
20
  "revision": null,
21
  "target_modules": [
22
  "gate_proj",
23
- "down_proj",
24
  "up_proj",
25
- "v_proj",
26
- "q_proj"
27
  ],
28
  "task_type": "CAUSAL_LM"
29
  }
 
20
  "revision": null,
21
  "target_modules": [
22
  "gate_proj",
23
+ "q_proj",
24
  "up_proj",
25
+ "down_proj",
26
+ "v_proj"
27
  ],
28
  "task_type": "CAUSAL_LM"
29
  }
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:02ee57ad39ebd6e0d18b13d220fedc406e7ab90f87599ee0dc5a94b56c51ee2b
3
- size 281061608
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f7c17cd4d2e899317c82d864f0d675ee0cf465eadfd0f84985fbb4002ea155c
3
+ size 562079784
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:98be37ba85ca295fc4954a0c0bb2c47b3b9d9339a8475034bb88ecc1078d67f8
3
- size 6520
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8280bcd9953b85bcb0bf58f5a8f5f65e53af679718bfde76d3361dfe649d25c4
3
+ size 4728