Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
datasets:
|
| 4 |
+
- vumichien/meals-data-gliner
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
library_name: gliner
|
| 8 |
+
---
|
| 9 |
+
# vumichien/ner-jp-gliner
|
| 10 |
+
|
| 11 |
+
This model is a fine-tuned version of [deberta-v3-base-small](microsoft/deberta-v3-small) on the meals synthetic dataset that generated by Mistral 8B.
|
| 12 |
+
It achieves the following results:
|
| 13 |
+
- Precision: 84.79%
|
| 14 |
+
- Recall: 75.04%
|
| 15 |
+
- F1 score: 79.62%
|
| 16 |
+
|
| 17 |
+
## Model description
|
| 18 |
+
|
| 19 |
+
More information needed
|
| 20 |
+
|
| 21 |
+
## Intended uses & limitations
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Training and evaluation data
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training procedure
|
| 30 |
+
|
| 31 |
+
### Training hyperparameters
|
| 32 |
+
The following hyperparameters were used during training:
|
| 33 |
+
- num_steps: 30000
|
| 34 |
+
- train_batch_size: 8
|
| 35 |
+
- eval_every: 3000
|
| 36 |
+
- warmup_ratio: 0.1
|
| 37 |
+
- scheduler_type: "cosine"
|
| 38 |
+
- loss_alpha: -1
|
| 39 |
+
- loss_gamma: 0
|
| 40 |
+
- label_smoothing: 0
|
| 41 |
+
- loss_reduction: "sum"
|
| 42 |
+
- lr_encoder: 1e-5
|
| 43 |
+
- lr_others: 5e-5
|
| 44 |
+
- weight_decay_encoder: 0.01
|
| 45 |
+
- weight_decay_other: 0.01
|
| 46 |
+
|
| 47 |
+
### Training results
|
| 48 |
+
|
| 49 |
+
| Epoch | Training Loss |
|
| 50 |
+
|:-----:|:-------------:|
|
| 51 |
+
| 1 | No log |
|
| 52 |
+
| 2 | No log |
|
| 53 |
+
| 3 | 75.730700 |
|
| 54 |
+
| 4 | 75.730700 |
|
| 55 |
+
| 5 | 75.730700 |
|
| 56 |
+
| 6 | 61.001500 |
|
| 57 |
+
| 7 | 61.001500 |
|
| 58 |
+
| 8 | 54.493900 |
|
| 59 |
+
| 9 | 54.493900 |
|
| 60 |
+
| 10 | 54.493900 |
|
| 61 |
+
| 11 | 48.829000 |
|
| 62 |
+
| 12 | 48.829000 |
|
| 63 |
+
| 13 | 44.892000 |
|
| 64 |
+
| 14 | 44.892000 |
|
| 65 |
+
| 15 | 44.892000 |
|
| 66 |
+
| 16 | 41.297600 |
|
| 67 |
+
| 17 | 41.297600 |
|
| 68 |
+
| 18 | 41.297600 |
|
| 69 |
+
| 19 | 37.768500 |
|
| 70 |
+
| 20 | 37.768500 |
|
| 71 |
+
| 21 | 35.017400 |
|
| 72 |
+
| 22 | 35.017400 |
|
| 73 |
+
| 23 | 35.017400 |
|
| 74 |
+
| 24 | 32.340500 |
|
| 75 |
+
| 25 | 32.340500 |
|
| 76 |
+
| 26 | 29.995400 |
|
| 77 |
+
| 27 | 29.995400 |
|
| 78 |
+
| 28 | 29.995400 |
|
| 79 |
+
| 29 | 28.467700 |
|
| 80 |
+
| 30 | 28.467700 |
|
| 81 |
+
| 31 | 28.467700 |
|
| 82 |
+
| 32 | 26.469200 |
|
| 83 |
+
| 33 | 26.469200 |
|
| 84 |
+
| 34 | 25.156200 |
|
| 85 |
+
| 35 | 25.156200 |
|
| 86 |
+
| 36 | 25.156200 |
|
| 87 |
+
| 37 | 24.252900 |
|
| 88 |
+
| 38 | 24.252900 |
|
| 89 |
+
| 39 | 23.941300 |
|
| 90 |
+
| 40 | 23.941300 |
|
| 91 |
+
| 41 | 23.941300 |
|
| 92 |
+
| 42 | 22.776800 |
|
| 93 |
+
| 43 | 22.776800 |
|
| 94 |
+
| 44 | 22.776800 |
|
| 95 |
+
| 45 | 23.013400 |
|
| 96 |
+
| 46 | 23.013400 |
|
| 97 |
+
| 47 | 22.030100 |
|
| 98 |
+
| 48 | 22.030100 |
|
| 99 |
+
| 49 | 22.030100 |
|
| 100 |
+
| 50 | 21.937700 |
|
| 101 |
+
| 51 | 21.948900 |
|