|
--- |
|
license: apache-2.0 |
|
base_model: allenai/longformer-base-4096 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: clinical_longformer_new_tokens |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# clinical_longformer_new_tokens |
|
|
|
This model is a fine-tuned version of [allenai/longformer-base-4096](https://huggingface.co/allenai/longformer-base-4096) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.6416 |
|
|
|
## 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: 2e-05 |
|
- train_batch_size: 2 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 56 |
|
- total_train_batch_size: 112 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 1500 |
|
- num_epochs: 4 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 4.3605 | 0.31 | 50 | 3.7983 | |
|
| 3.9502 | 0.62 | 100 | 3.4157 | |
|
| 3.4719 | 0.93 | 150 | 3.2378 | |
|
| 3.3937 | 1.24 | 200 | 3.0982 | |
|
| 3.3123 | 1.56 | 250 | 3.0227 | |
|
| 3.3164 | 1.87 | 300 | 2.9341 | |
|
| 2.9789 | 2.18 | 350 | 2.8408 | |
|
| 2.9593 | 2.49 | 400 | 2.8201 | |
|
| 2.8724 | 2.8 | 450 | 2.7561 | |
|
| 2.8753 | 3.11 | 500 | 2.7083 | |
|
| 2.8017 | 3.42 | 550 | 2.6593 | |
|
| 2.7496 | 3.73 | 600 | 2.6416 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.34.1 |
|
- Pytorch 2.1.0+cu118 |
|
- Datasets 2.14.6 |
|
- Tokenizers 0.14.1 |
|
|
