gs-GreBerta
This model is a fine-tuned version of bowphs/GreBerta on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4270
- Top1: 36.9478
- Top5: 62.2490
- Top10: 68.0723
- Top20: 71.4859
- Bertscore F1 Top1: 85.8014
- Bertscore F1 Top1 Mean: 85.8014
- Bertscore F1 Top5: 91.7554
- Bertscore F1 Top5 Mean: 81.4919
- Bertscore F1 Top10: 93.3481
- Bertscore F1 Top10 Mean: 79.7881
- Bertscore F1 Top20: 94.3306
- Bertscore F1 Top20 Mean: 78.4723
- Cos Sim Top1 Max: 68.1975
- Cos Sim Top1 Mean: 68.1975
- Cos Sim Top5 Max: 79.9159
- Cos Sim Top5 Mean: 61.7656
- Cos Sim Top10 Max: 82.8232
- Cos Sim Top10 Mean: 59.8611
- Cos Sim Top20 Max: 85.6527
- Cos Sim Top20 Mean: 58.0537
- Cluster Inclusion Rate: 67.4699
- Mean Inclusion Margin: 0.1014
- Mean Gold Centroid Cosine Sim: 73.0709
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:
- learning_rate: 4.7499747713783996e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Top1 | Top5 | Top10 | Top20 | Bertscore F1 Top1 | Bertscore F1 Top1 Mean | Bertscore F1 Top5 | Bertscore F1 Top5 Mean | Bertscore F1 Top10 | Bertscore F1 Top10 Mean | Bertscore F1 Top20 | Bertscore F1 Top20 Mean | Cos Sim Top1 Max | Cos Sim Top1 Mean | Cos Sim Top5 Max | Cos Sim Top5 Mean | Cos Sim Top10 Max | Cos Sim Top10 Mean | Cos Sim Top20 Max | Cos Sim Top20 Mean | Cluster Inclusion Rate | Mean Inclusion Margin | Mean Gold Centroid Cosine Sim |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1.8347 | 1.0 | 882 | 1.6459 | 37.9518 | 63.4538 | 68.8755 | 72.4900 | 85.9921 | 85.9921 | 91.9632 | 81.5230 | 93.4430 | 79.7370 | 94.3738 | 78.4921 | 69.6010 | 69.6010 | 80.9902 | 62.9953 | 83.7362 | 60.8302 | 85.8610 | 59.0321 | 64.0562 | 0.0928 | 73.5296 |
| 1.6137 | 2.0 | 1764 | 1.4800 | 34.9398 | 62.4498 | 69.4779 | 72.0884 | 85.4250 | 85.4250 | 91.9193 | 81.4959 | 93.5057 | 79.7619 | 94.4510 | 78.4752 | 67.6464 | 67.6464 | 80.0759 | 61.9107 | 83.2952 | 59.9759 | 85.9502 | 58.2344 | 66.6667 | 0.0934 | 73.1168 |
| 1.5620 | 3.0 | 2646 | 1.4228 | 36.9478 | 62.2490 | 68.0723 | 71.4859 | 85.8014 | 85.8014 | 91.7554 | 81.4919 | 93.3481 | 79.7881 | 94.3306 | 78.4723 | 68.1975 | 68.1975 | 79.9159 | 61.7656 | 82.8232 | 59.8611 | 85.6527 | 58.0537 | 67.4699 | 0.1014 | 73.0709 |
Framework versions
- Transformers 5.17.0
- Pytorch 2.14.0+cu126
- Datasets 3.6.0
- Tokenizers 0.23.2