Metadata-Version: 2.3
Name: text2dataset
Version: 0.1.4
Summary: Easily turn large English text datasets into Japanese text datasets using open LLMs
Project-URL: Repository, https://github.com/llm-jp/text2dataset
Author-email: speed1313 <speedtry13@icloud.com>
License: MIT License
        
        Copyright (c) 2024 speed
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
License-File: LICENSE
Requires-Python: >=3.12.1
Requires-Dist: click>=8.1.7
Requires-Dist: datasets>=3.0.0
Requires-Dist: vllm>=0.6.1
Requires-Dist: wandb>=0.18.0
Description-Content-Type: text/markdown

# text2dataset
Easily turn large English text datasets into Japanese text datasets using open LLMs.

A tool for converting a datasets.Dataset by translating the data in the "txt" column using Open LLM like gemma2 with vLLM, and adding a new "txt_ja" column (translated text in Japanese).
This tool is inspired by [img2dataset](https://github.com/rom1504/img2dataset).

## Features
- Save the intermediate results in shards:
  - By setting the `number_sample_per_shard` parameter, the dataset can be saved in shards as specified by the number of samples per shard.
- Resume from checkpoint:
  - By setting the `resume_from_checkpoint` parameter, the translation can be resumed from where it left off.
- Logging with wandb:
  - By setting the `use_wandb` parameter, the metrics such as examples_per_sec and count can be logged to wandb.
- Push to Hugging Face Hub:
  - By setting the `push_to_hub` parameter, the translated dataset can be pushed to the Hugging Face Hub.


## Usage

```bash
$ python src/text2dataset/main.py \
    --model_id "google/gemma-2-9b-it" \
    --batch_size 16384 \
    --input_format parquet \
    --input_path "/path/to/input" \
    --source_column "caption" \
    --target_column "caption_ja" \
    --push_to_hub False \
    --push_to_hub_path "/path/to/hub" \
    --output_dir "/path/to/output" \
    --output_format parquet \
    --gpu_id 0 \
    --number_sample_per_shard 10000 \
    --use_wandb True
```

### Example
```python
>>> from datasets import load_dataset
>>> load_dataset("parquet", data_files="/path/to/input", split="train")
DatasetDict({
    train: Dataset({
        features: ['__key__', '__url__', 'jpg', 'json', 'txt'],
        num_rows: 1000
    })
})
>>> load_dataset("parquet", data_files="/path/to/output")
DatasetDict({
    train: Dataset({
        features: ['__key__', '__url__', 'jpg', 'json', 'txt', 'txt_ja'],
        num_rows: 1000
    })
})
```

## Areas for Improvement
- Data Paarallel Inference:
  - Currently, only one model is used for inference. This can be improved by using DataParallel. If you know how to do this with vLLM, please let me know or Pull Request.




## References
- https://github.com/vllm-project/vllm
- https://github.com/rom1504/img2dataset
