version 2
Browse files
README.md
CHANGED
@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
|
|
18 |
|
19 |
This model is a fine-tuned version of [facebook/bart-large](https://huggingface.co/facebook/bart-large) on the None dataset.
|
20 |
It achieves the following results on the evaluation set:
|
21 |
-
- Loss:
|
22 |
-
- Sacrebleu: 66.
|
23 |
|
24 |
## Model description
|
25 |
|
@@ -46,53 +46,16 @@ The following hyperparameters were used during training:
|
|
46 |
- total_train_batch_size: 256
|
47 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
48 |
- lr_scheduler_type: linear
|
49 |
-
- num_epochs:
|
50 |
- mixed_precision_training: Native AMP
|
51 |
|
52 |
### Training results
|
53 |
|
54 |
| Training Loss | Epoch | Step | Validation Loss | Sacrebleu |
|
55 |
|:-------------:|:-----:|:----:|:---------------:|:---------:|
|
56 |
-
| No log | 0.99 | 54 | 0.
|
57 |
-
| No log | 1.99 | 109 | 0.
|
58 |
-
| No log |
|
59 |
-
| No log | 4.0 | 219 | 0.5929 | 66.7857 |
|
60 |
-
| No log | 4.99 | 273 | 0.6427 | 65.8395 |
|
61 |
-
| No log | 5.99 | 328 | 0.6721 | 66.4172 |
|
62 |
-
| No log | 7.0 | 383 | 0.7511 | 66.4660 |
|
63 |
-
| No log | 8.0 | 438 | 0.7662 | 66.6480 |
|
64 |
-
| No log | 8.99 | 492 | 0.7588 | 66.5092 |
|
65 |
-
| No log | 9.99 | 547 | 0.7916 | 66.5144 |
|
66 |
-
| No log | 11.0 | 602 | 0.8172 | 66.6279 |
|
67 |
-
| No log | 12.0 | 657 | 0.8350 | 66.5607 |
|
68 |
-
| No log | 12.99 | 711 | 0.8809 | 66.6095 |
|
69 |
-
| No log | 13.99 | 766 | 0.8843 | 66.4089 |
|
70 |
-
| No log | 15.0 | 821 | 1.0130 | 66.5184 |
|
71 |
-
| No log | 16.0 | 876 | 0.9180 | 66.4269 |
|
72 |
-
| No log | 16.99 | 930 | 0.9794 | 66.5766 |
|
73 |
-
| No log | 17.99 | 985 | 0.9450 | 66.6713 |
|
74 |
-
| No log | 19.0 | 1040 | 0.9880 | 66.7081 |
|
75 |
-
| No log | 20.0 | 1095 | 0.9540 | 66.4440 |
|
76 |
-
| No log | 20.99 | 1149 | 1.0552 | 66.5390 |
|
77 |
-
| No log | 21.99 | 1204 | 0.9806 | 66.5975 |
|
78 |
-
| No log | 23.0 | 1259 | 1.0528 | 66.6404 |
|
79 |
-
| No log | 24.0 | 1314 | 1.0348 | 66.4127 |
|
80 |
-
| No log | 24.99 | 1368 | 1.0758 | 66.6139 |
|
81 |
-
| No log | 25.99 | 1423 | 1.1291 | 66.6778 |
|
82 |
-
| No log | 27.0 | 1478 | 1.1112 | 66.6411 |
|
83 |
-
| No log | 28.0 | 1533 | 1.1305 | 66.5986 |
|
84 |
-
| No log | 28.99 | 1587 | 1.1532 | 66.5047 |
|
85 |
-
| No log | 29.99 | 1642 | 1.1106 | 66.5662 |
|
86 |
-
| No log | 31.0 | 1697 | 1.2084 | 66.6593 |
|
87 |
-
| No log | 32.0 | 1752 | 1.1438 | 66.6117 |
|
88 |
-
| No log | 32.99 | 1806 | 1.1956 | 66.6758 |
|
89 |
-
| No log | 33.99 | 1861 | 1.1630 | 66.7359 |
|
90 |
-
| No log | 35.0 | 1916 | 1.1570 | 66.6989 |
|
91 |
-
| No log | 36.0 | 1971 | 1.1754 | 66.6495 |
|
92 |
-
| No log | 36.99 | 2025 | 1.2456 | 66.7018 |
|
93 |
-
| No log | 37.99 | 2080 | 1.2197 | 66.7990 |
|
94 |
-
| No log | 39.0 | 2135 | 1.1886 | 66.7049 |
|
95 |
-
| No log | 39.45 | 2160 | 1.1977 | 66.7256 |
|
96 |
|
97 |
|
98 |
### Framework versions
|
|
|
18 |
|
19 |
This model is a fine-tuned version of [facebook/bart-large](https://huggingface.co/facebook/bart-large) on the None dataset.
|
20 |
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 0.5669
|
22 |
+
- Sacrebleu: 66.8302
|
23 |
|
24 |
## Model description
|
25 |
|
|
|
46 |
- total_train_batch_size: 256
|
47 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
48 |
- lr_scheduler_type: linear
|
49 |
+
- num_epochs: 3
|
50 |
- mixed_precision_training: Native AMP
|
51 |
|
52 |
### Training results
|
53 |
|
54 |
| Training Loss | Epoch | Step | Validation Loss | Sacrebleu |
|
55 |
|:-------------:|:-----:|:----:|:---------------:|:---------:|
|
56 |
+
| No log | 0.99 | 54 | 0.6545 | 66.3234 |
|
57 |
+
| No log | 1.99 | 109 | 0.5940 | 66.8342 |
|
58 |
+
| No log | 2.96 | 162 | 0.5669 | 66.8302 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
|
61 |
### Framework versions
|