Hugging Face's logo

daqc
/
kuntur-peru-legal-es-gemma-2b-it

PEFT
Safetensors
trl
sft
Generated from Trainer
Model card Files Community
daqc commited on
Commit
f841f08
·
verified ·
1 Parent(s): 9779168

End of training

Browse files
Files changed (2) hide show
  1. README.md +162 -0
  2. adapter_model.safetensors +1 -1
README.md ADDED
@@ -0,0 +1,162 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: gemma
3
+ library_name: peft
4
+ tags:
5
+ - trl
6
+ - sft
7
+ - generated_from_trainer
8
+ base_model: google/gemma-2b-it
9
+ datasets:
10
+ - generator
11
+ model-index:
12
+ - name: kuntur-peru-legal-es-gemma-2b-it
13
+ results: []
14
+ ---
15
+
16
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
17
+ should probably proofread and complete it, then remove this comment. -->
18
+
19
+ # kuntur-peru-legal-es-gemma-2b-it
20
+
21
+ This model is a fine-tuned version of [google/gemma-2b-it](https://huggingface.co/google/gemma-2b-it) on the generator dataset.
22
+ It achieves the following results on the evaluation set:
23
+ - Loss: 1.1387
24
+
25
+ ## Model description
26
+
27
+ More information needed
28
+
29
+ ## Intended uses & limitations
30
+
31
+ More information needed
32
+
33
+ ## Training and evaluation data
34
+
35
+ More information needed
36
+
37
+ ## Training procedure
38
+
39
+ ### Training hyperparameters
40
+
41
+ The following hyperparameters were used during training:
42
+ - learning_rate: 2.5e-05
43
+ - train_batch_size: 4
44
+ - eval_batch_size: 1
45
+ - seed: 66
46
+ - gradient_accumulation_steps: 4
47
+ - total_train_batch_size: 16
48
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
49
+ - lr_scheduler_type: linear
50
+ - num_epochs: 50
51
+ - mixed_precision_training: Native AMP
52
+
53
+ ### Training results
54
+
55
+ | Training Loss | Epoch | Step | Validation Loss |
56
+ |:-------------:|:-----:|:----:|:---------------:|
57
+ | 3.7041 | 0.51 | 50 | 3.6704 |
58
+ | 2.5585 | 1.02 | 100 | 2.5245 |
59
+ | 1.8723 | 1.53 | 150 | 1.9012 |
60
+ | 1.697 | 2.05 | 200 | 1.6294 |
61
+ | 1.5123 | 2.56 | 250 | 1.5092 |
62
+ | 1.3844 | 3.07 | 300 | 1.4406 |
63
+ | 1.4082 | 3.58 | 350 | 1.3942 |
64
+ | 1.3473 | 4.09 | 400 | 1.3614 |
65
+ | 1.2698 | 4.6 | 450 | 1.3338 |
66
+ | 1.3179 | 5.12 | 500 | 1.3127 |
67
+ | 1.2776 | 5.63 | 550 | 1.2942 |
68
+ | 1.2529 | 6.14 | 600 | 1.2781 |
69
+ | 1.2148 | 6.65 | 650 | 1.2667 |
70
+ | 1.2378 | 7.16 | 700 | 1.2538 |
71
+ | 1.1976 | 7.67 | 750 | 1.2418 |
72
+ | 1.2107 | 8.18 | 800 | 1.2325 |
73
+ | 1.199 | 8.7 | 850 | 1.2216 |
74
+ | 1.1498 | 9.21 | 900 | 1.2149 |
75
+ | 1.1788 | 9.72 | 950 | 1.2059 |
76
+ | 1.0873 | 10.23 | 1000 | 1.1995 |
77
+ | 1.1124 | 10.74 | 1050 | 1.1912 |
78
+ | 1.1161 | 11.25 | 1100 | 1.1858 |
79
+ | 1.1408 | 11.76 | 1150 | 1.1782 |
80
+ | 1.083 | 12.28 | 1200 | 1.1735 |
81
+ | 1.1234 | 12.79 | 1250 | 1.1659 |
82
+ | 1.1065 | 13.3 | 1300 | 1.1609 |
83
+ | 1.112 | 13.81 | 1350 | 1.1555 |
84
+ | 1.0759 | 14.32 | 1400 | 1.1513 |
85
+ | 1.0783 | 14.83 | 1450 | 1.1462 |
86
+ | 1.0466 | 15.35 | 1500 | 1.1455 |
87
+ | 1.0334 | 15.86 | 1550 | 1.1424 |
88
+ | 1.045 | 16.37 | 1600 | 1.1405 |
89
+ | 1.016 | 16.88 | 1650 | 1.1393 |
90
+ | 1.0449 | 17.39 | 1700 | 1.1371 |
91
+ | 1.0642 | 17.9 | 1750 | 1.1338 |
92
+ | 1.0276 | 18.41 | 1800 | 1.1340 |
93
+ | 1.0328 | 18.93 | 1850 | 1.1313 |
94
+ | 1.0232 | 19.44 | 1900 | 1.1326 |
95
+ | 1.0588 | 19.95 | 1950 | 1.1284 |
96
+ | 0.9971 | 20.46 | 2000 | 1.1298 |
97
+ | 1.0561 | 20.97 | 2050 | 1.1269 |
98
+ | 1.0714 | 21.48 | 2100 | 1.1279 |
99
+ | 1.0358 | 21.99 | 2150 | 1.1270 |
100
+ | 0.9744 | 22.51 | 2200 | 1.1274 |
101
+ | 1.0019 | 23.02 | 2250 | 1.1275 |
102
+ | 0.9362 | 23.53 | 2300 | 1.1258 |
103
+ | 1.0143 | 24.04 | 2350 | 1.1254 |
104
+ | 1.009 | 24.55 | 2400 | 1.1290 |
105
+ | 0.9969 | 25.06 | 2450 | 1.1253 |
106
+ | 0.8828 | 25.58 | 2500 | 1.1256 |
107
+ | 1.022 | 26.09 | 2550 | 1.1257 |
108
+ | 0.9804 | 26.6 | 2600 | 1.1265 |
109
+ | 0.9851 | 27.11 | 2650 | 1.1276 |
110
+ | 0.9617 | 27.62 | 2700 | 1.1265 |
111
+ | 0.9346 | 28.13 | 2750 | 1.1263 |
112
+ | 0.9552 | 28.64 | 2800 | 1.1258 |
113
+ | 0.9376 | 29.16 | 2850 | 1.1287 |
114
+ | 0.9359 | 29.67 | 2900 | 1.1262 |
115
+ | 0.9447 | 30.18 | 2950 | 1.1271 |
116
+ | 0.9646 | 30.69 | 3000 | 1.1278 |
117
+ | 0.926 | 31.2 | 3050 | 1.1293 |
118
+ | 0.9456 | 31.71 | 3100 | 1.1293 |
119
+ | 0.9223 | 32.23 | 3150 | 1.1296 |
120
+ | 0.9589 | 32.74 | 3200 | 1.1278 |
121
+ | 1.0145 | 33.25 | 3250 | 1.1299 |
122
+ | 0.9315 | 33.76 | 3300 | 1.1292 |
123
+ | 0.8946 | 34.27 | 3350 | 1.1311 |
124
+ | 0.9441 | 34.78 | 3400 | 1.1297 |
125
+ | 0.8996 | 35.29 | 3450 | 1.1317 |
126
+ | 0.9307 | 35.81 | 3500 | 1.1290 |
127
+ | 0.9005 | 36.32 | 3550 | 1.1329 |
128
+ | 0.9167 | 36.83 | 3600 | 1.1303 |
129
+ | 0.9393 | 37.34 | 3650 | 1.1322 |
130
+ | 0.9658 | 37.85 | 3700 | 1.1313 |
131
+ | 0.9375 | 38.36 | 3750 | 1.1341 |
132
+ | 0.9176 | 38.87 | 3800 | 1.1326 |
133
+ | 0.8982 | 39.39 | 3850 | 1.1351 |
134
+ | 0.9685 | 39.9 | 3900 | 1.1326 |
135
+ | 0.9216 | 40.41 | 3950 | 1.1355 |
136
+ | 0.9542 | 40.92 | 4000 | 1.1342 |
137
+ | 0.8739 | 41.43 | 4050 | 1.1371 |
138
+ | 0.9329 | 41.94 | 4100 | 1.1355 |
139
+ | 0.9335 | 42.46 | 4150 | 1.1354 |
140
+ | 0.8851 | 42.97 | 4200 | 1.1363 |
141
+ | 0.9217 | 43.48 | 4250 | 1.1377 |
142
+ | 0.8794 | 43.99 | 4300 | 1.1363 |
143
+ | 0.9104 | 44.5 | 4350 | 1.1371 |
144
+ | 0.8751 | 45.01 | 4400 | 1.1367 |
145
+ | 0.9157 | 45.52 | 4450 | 1.1377 |
146
+ | 0.8277 | 46.04 | 4500 | 1.1374 |
147
+ | 0.8858 | 46.55 | 4550 | 1.1384 |
148
+ | 0.9195 | 47.06 | 4600 | 1.1378 |
149
+ | 0.925 | 47.57 | 4650 | 1.1383 |
150
+ | 0.9007 | 48.08 | 4700 | 1.1384 |
151
+ | 0.9184 | 48.59 | 4750 | 1.1385 |
152
+ | 0.8798 | 49.1 | 4800 | 1.1385 |
153
+ | 0.8596 | 49.62 | 4850 | 1.1387 |
154
+
155
+
156
+ ### Framework versions
157
+
158
+ - PEFT 0.10.0
159
+ - Transformers 4.38.0
160
+ - Pytorch 2.2.2+cu121
161
+ - Datasets 2.18.0
162
+ - Tokenizers 0.15.2
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d307622f9fa62d38aaffa5eeff73b1512e714560201fb9d587698f47dc3648cb
3
  size 7391688
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:721f3d284329e2ef3ab27b10facf0116e2331ee597d4de5cb20e5e1382c2be91
3
  size 7391688