|
--- |
|
language: |
|
- ru |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- murakami |
|
model-index: |
|
- name: murakami_rugpt3small |
|
results: [] |
|
--- |
|
|
|
# murakami_rugpt3small |
|
|
|
## Model description |
|
|
|
Fine-tuned from [sberbank-ai/rugpt3small_based_on_gpt2](https://huggingface.co/sberbank-ai/rugpt3small_based_on_gpt2) |
|
|
|
## Intended uses & limitations |
|
|
|
Generate novels in the style of Haruki Murakami, in Russian. |
|
|
|
Just my exercise to learn the Huggingface ecosystem, really. |
|
|
|
## Training and evaluation data |
|
|
|
Fine-tuned on [murakami](https://huggingface.co/datasets/vldsavelyev/murakami) dataset, |
|
which was built from Russian translations of novels by Haruki Murakami. |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 5e-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 |
|
- gradient_checkpointing: True |
|
- num_epochs: 3.0 |
|
- mixed_precision_training: Native AMP (fp16=True) |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.27.3 |
|
- Pytorch 2.0.0+cu117 |
|
- Datasets 2.10.1 |
|
- Tokenizers 0.13.2 |
|
|
