Instructions to use MM-MVR/UniViTAR-0.3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MM-MVR/UniViTAR-0.3B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="MM-MVR/UniViTAR-0.3B")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("MM-MVR/UniViTAR-0.3B", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Request for Batch Processing Support in Inference
#1
by jamessyx - opened
Excellent work on UniViTAR!
I have a question regarding the inference capabilities. From the provided code, it appears that UniViTAR currently only supports inference on single (packed) samples and doesn't support batch processing for multiple samples.
Issue:
When working with larger batch sizes (e.g., batch_size=1000), putting all samples into a single sequence seems impractical and inefficient. Could you please provide a batch processing inference script that can handle multiple samples efficiently? This would be extremely helpful for scenarios where we need to process large numbers of samples.