Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,88 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
datasets:
|
| 4 |
+
- stockmark/ner-wikipedia-dataset
|
| 5 |
+
language:
|
| 6 |
+
- ja
|
| 7 |
+
library_name: gliner
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# vumichien/ner-jp-gliner
|
| 11 |
+
|
| 12 |
+
This model is a fine-tuned version of [deberta-v3-base-japanese](ku-nlp/deberta-v3-base-japanese) on the Japanese Ner Wikipedia dataset.
|
| 13 |
+
It achieves the following results:
|
| 14 |
+
- Precision: 96.07%
|
| 15 |
+
- Recall: 89.16%
|
| 16 |
+
- F1 score: 92.49%
|
| 17 |
+
|
| 18 |
+
## Model description
|
| 19 |
+
|
| 20 |
+
More information needed
|
| 21 |
+
|
| 22 |
+
## Intended uses & limitations
|
| 23 |
+
|
| 24 |
+
More information needed
|
| 25 |
+
|
| 26 |
+
## Training and evaluation data
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Training procedure
|
| 31 |
+
|
| 32 |
+
### Training hyperparameters
|
| 33 |
+
The following hyperparameters were used during training:
|
| 34 |
+
- num_steps: 30000
|
| 35 |
+
- train_batch_size: 8
|
| 36 |
+
- eval_every: 3000
|
| 37 |
+
- warmup_ratio: 0.1
|
| 38 |
+
- scheduler_type: "cosine"
|
| 39 |
+
- loss_alpha: -1
|
| 40 |
+
- loss_gamma: 0
|
| 41 |
+
- label_smoothing: 0
|
| 42 |
+
- loss_reduction: "sum"
|
| 43 |
+
- lr_encoder: 1e-5
|
| 44 |
+
- lr_others: 5e-5
|
| 45 |
+
- weight_decay_encoder: 0.01
|
| 46 |
+
- weight_decay_other: 0.01
|
| 47 |
+
|
| 48 |
+
### Training results
|
| 49 |
+
|
| 50 |
+
| Epoch | Training Loss |
|
| 51 |
+
|:-----:|:-------------:|
|
| 52 |
+
| 1 | 1291.582200 |
|
| 53 |
+
| 2 | 53.290100 |
|
| 54 |
+
| 3 | 44.137400 |
|
| 55 |
+
| 4 | 35.286200 |
|
| 56 |
+
| 5 | 20.865500 |
|
| 57 |
+
| 6 | 15.890000 |
|
| 58 |
+
| 7 | 13.369600 |
|
| 59 |
+
| 8 | 11.599500 |
|
| 60 |
+
| 9 | 9.773400 |
|
| 61 |
+
| 10 | 8.372600 |
|
| 62 |
+
| 11 | 7.256200 |
|
| 63 |
+
| 12 | 6.521800 |
|
| 64 |
+
| 13 | 7.203800 |
|
| 65 |
+
| 14 | 7.032900 |
|
| 66 |
+
| 15 | 6.189700 |
|
| 67 |
+
| 16 | 6.897400 |
|
| 68 |
+
| 17 | 6.031700 |
|
| 69 |
+
| 18 | 5.329600 |
|
| 70 |
+
| 19 | 5.411300 |
|
| 71 |
+
| 20 | 5.253800 |
|
| 72 |
+
| 21 | 4.522000 |
|
| 73 |
+
| 22 | 5.107700 |
|
| 74 |
+
| 23 | 4.163300 |
|
| 75 |
+
| 24 | 4.185400 |
|
| 76 |
+
| 25 | 3.403100 |
|
| 77 |
+
| 26 | 3.272400 |
|
| 78 |
+
| 27 | 2.387800 |
|
| 79 |
+
| 28 | 3.039400 |
|
| 80 |
+
| 29 | 2.383000 |
|
| 81 |
+
| 30 | 1.895300 |
|
| 82 |
+
| 31 | 1.748700 |
|
| 83 |
+
| 32 | 1.864300 |
|
| 84 |
+
| 33 | 2.343000 |
|
| 85 |
+
| 34 | 1.356600 |
|
| 86 |
+
| 35 | 1.182000 |
|
| 87 |
+
| 36 | 0.894700 |
|
| 88 |
+
| 37 | 0.954900 |
|