|
--- |
|
license: mit |
|
datasets: |
|
- Vi-VLM/Vista |
|
language: |
|
- vi |
|
library_name: adapter-transformers |
|
pipeline_tag: text-classification |
|
--- |
|
|
|
# Introducing MoE-LLaVA-Qwen1.5-1.8B×4-Top2 for Vietnamese |
|
|
|
We are excited to present MoE-LLaVA-Qwen1.5-1.8B×4-Top2, tailored for the Vietnamese language. This model is part of our ongoing efforts to develop Vision Language Models (VLM) for Vietnamese, a domain that is currently limited and predominantly features larger models (~7B parameters). Our model activates approximately 2.2B parameters per call, significantly reducing the memory footprint, and it can be quantized for local execution. |
|
|
|
## Training Dataset |
|
|
|
Our model is trained on the comprehensive [Vi-VLM/Vista dataset](https://huggingface.co/datasets/Vi-VLM/Vista), which includes around 700,000 Vietnamese vision-language samples curated by Gemini Pro. We employed various prompt engineering techniques, including: |
|
|
|
- **Few-shot Learning** |
|
- **Caption-based Prompting** |
|
- **Image-based Prompting** |
|
|
|
For the COCO dataset, we utilized Llava-style prompts to generate data. For the ShareGPT4V dataset, translation prompts were applied. |
|
|
|
### Techniques Used |
|
|
|
- **Caption-based Prompting:** Utilizes accurate captions and bounding boxes from the original dataset. |
|
- **Image-based Prompting:** Leverages images to generate captions and conversations. |
|
|
|
## Bias, Risks, and Limitations |
|
|
|
The dataset may contain biases originating from its sources. Users should remain aware of these potential biases when utilizing the dataset. |
|
|
|
## More Information |
|
|
|
This dataset represents the first stage of a two-stage development process for a larger model. Stay tuned for future developments by subscribing to our updates. |