|
--- |
|
license: bigscience-bloom-rail-1.0 |
|
tags: |
|
- generated_from_trainer |
|
base_model: bigscience/bloom-560m |
|
model-index: |
|
- name: bloom-560m-finetuned-cdn_law |
|
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. --> |
|
|
|
# Canadian Appellate Judgement Model |
|
|
|
This model is a fine-tuned version of [bigscience/bloom-560m](https://huggingface.co/bigscience/bloom-560m) on Canadian appellate decisions (Ontario Court of Appeal and the British Columbia Court of Appeal) found in the [Pile of Law](https://huggingface.co/datasets/pile-of-law/pile-of-law) dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.0135 |
|
|
|
## Intended uses & limitations |
|
|
|
This model is intended to facilitate research into large language models and legal reasoning. It is not intended for use in any legal domain or to support legal work . |
|
|
|
## Training procedure |
|
|
|
This model was trained using the methodology set out in this [notebook](https://huggingface.co/docs/transformers/training). |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 2e-05 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 3.0 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:-----:|:---------------:| |
|
| 2.1285 | 1.0 | 8298 | 2.0347 | |
|
| 1.7999 | 2.0 | 16596 | 1.9876 | |
|
| 1.6069 | 3.0 | 24894 | 2.0135 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.23.1 |
|
- Pytorch 1.11.0 |
|
- Datasets 2.5.2 |
|
- Tokenizers 0.13.1 |
|
|
