|
--- |
|
license: cc-by-nc-4.0 |
|
datasets: |
|
- vumichien/meals-data-gliner |
|
language: |
|
- en |
|
library_name: gliner |
|
--- |
|
# vumichien/ner-jp-gliner |
|
|
|
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. |
|
It achieves the following results: |
|
- Precision: 84.79% |
|
- Recall: 75.04% |
|
- F1 score: 79.62% |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
The following hyperparameters were used during training: |
|
- num_steps: 30000 |
|
- train_batch_size: 8 |
|
- eval_every: 3000 |
|
- warmup_ratio: 0.1 |
|
- scheduler_type: "cosine" |
|
- loss_alpha: -1 |
|
- loss_gamma: 0 |
|
- label_smoothing: 0 |
|
- loss_reduction: "sum" |
|
- lr_encoder: 1e-5 |
|
- lr_others: 5e-5 |
|
- weight_decay_encoder: 0.01 |
|
- weight_decay_other: 0.01 |
|
|
|
### Training results |
|
|
|
| Epoch | Training Loss | |
|
|:-----:|:-------------:| |
|
| 1 | No log | |
|
| 2 | 2008.786600 | |
|
| 3 | 2008.786600 | |
|
| 4 | 117.661100 | |
|
| 5 | 84.863400 | |
|
| 6 | 84.863400 | |
|
| 7 | 66.872200 | |
|
| 8 | 66.872200 | |
|
| 9 | 58.574600 | |
|
| 10 | 53.905900 | |
|
| 11 | 53.905900 | |
|
| 12 | 48.563900 | |
|
| 13 | 48.563900 | |
|
| 14 | 43.970700 | |
|
| 15 | 38.940100 | |
|
| 16 | 38.940100 | |
|
| 17 | 35.543100 | |
|
| 18 | 35.543100 | |
|
| 19 | 33.050500 | |
|
| 20 | 30.091100 | |
|
| 21 | 30.091100 | |
|
| 22 | 27.275200 | |
|
| 23 | 27.275200 | |
|
| 24 | 25.327500 | |
|
| 25 | 23.171200 | |
|
| 26 | 23.171200 | |
|
| 27 | 20.940300 | |
|
| 28 | 19.034100 | |
|
| 29 | 19.034100 | |
|
| 30 | 17.366400 | |
|
| 31 | 17.366400 | |
|
| 32 | 16.570800 | |
|
| 33 | 15.673200 | |
|
| 34 | 15.673200 | |
|
| 35 | 14.457500 | |
|
| 36 | 14.457500 | |
|
| 37 | 13.064500 | |
|
| 38 | 12.786100 | |
|
| 39 | 12.786100 | |
|
| 40 | 11.934400 | |
|
| 41 | 11.934400 | |
|
| 42 | 11.225800 | |
|
| 43 | 10.106500 | |
|
| 44 | 10.106500 | |
|
| 45 | 9.200000 | |
|
| 46 | 9.200000 | |
|
| 47 | 9.449100 | |
|
| 48 | 8.979400 | |
|
| 49 | 8.979400 | |
|
| 50 | 7.840100 | |
|
| 51 | 7.949600 | |
|
| 52 | 7.949600 | |
|
| 53 | 7.233800 | |
|
| 54 | 7.233800 | |
|
| 55 | 7.383200 | |
|
| 56 | 6.114800 | |
|
| 57 | 6.114800 | |
|
| 58 | 6.421800 | |
|
| 59 | 6.421800 | |
|
| 60 | 6.191000 | |
|
| 61 | 5.932200 | |
|
| 62 | 5.932200 | |
|
| 63 | 5.706100 | |
|
| 64 | 5.706100 | |
|
| 65 | 5.567800 | |
|
| 66 | 5.104100 | |
|
| 67 | 5.104100 | |
|
| 68 | 5.407800 | |
|
| 69 | 5.407800 | |
|
| 70 | 5.607500 | |
|
| 71 | 4.967500 | |
|
| 72 | 4.967500 | |
|
| 73 | 5.362100 | |
|
| 74 | 5.362100 | |
|
| 75 | 5.425800 | |
|
| 76 | 5.283100 | |
|
| 77 | 5.283100 | |
|
| 78 | 4.250000 | |
|
| 79 | 4.330900 | |
|
| 80 | 4.330900 | |
|
| 81 | 4.088400 | |
|
| 82 | 4.088400 | |
|
| 83 | 4.512400 | |
|
| 84 | 4.513500 | |
|
| 85 | 4.513500 | |
|
| 86 | 4.327000 | |
|
| 87 | 4.327000 | |
|
| 88 | 5.152200 | |
|
| 89 | 3.776100 | |
|
| 90 | 3.776100 | |
|
| 91 | 3.762500 | |
|
| 92 | 3.762500 | |
|
| 93 | 4.054900 | |
|
| 94 | 3.579700 | |
|
| 95 | 3.579700 | |
|
| 96 | 3.391500 | |
|
| 97 | 3.391500 | |
|
| 98 | 4.863200 | |
|
|
