---
license: mit
language:
- en
library_name: transformers
pipeline_tag: image-to-text
tags:
- video-to-text
- video-captioning
- image-to-text
- image-captioning
- visual-question-answering
- blip-2
---
# Model Card for EILEV BLIP-2-OPT-2.7B

![Teaser](teaser.png)

[Salesforce/blip2-opt-2.7b](https://huggingface.co/Salesforce/blip2-opt-2.7b) trained using [EILEV](https://github.com/yukw777/EILEV), a novel training method that can elicit in-context learning in vision-language models (VLMs) for egocentric videos without requiring massive, naturalistic egocentric video datasets.

## Model Details

### Model Description

EILEV BLIP-2-OPT-2.7B is a VLM optimized for egocentric video. It can perform in-context learning over videos and texts. It was trained on Ego4D.

### Model Sources

- **Repository:** https://github.com/yukw777/EILEV
- **Paper:** https://arxiv.org/abs/2311.17041
- **Demo:** https://2e09-141-212-106-177.ngrok-free.app

## Bias, Risks, and Limitations

EILEV BLIP-2-OPT-2.7B uses off-the-shelf OPT as the language model. It inherits the same risks and limitations as mentioned in Meta's model card.

> Like other large language models for which the diversity (or lack thereof) of training
> data induces downstream impact on the quality of our model, OPT-175B has limitations in terms
> of bias and safety. OPT-175B can also have quality issues in terms of generation diversity and
> hallucination. In general, OPT-175B is not immune from the plethora of issues that plague modern
> large language models.
> 

EILEV BLIP-2-OPT-2.7B has not been tested in real world applications. It should not be directly deployed in any applications. Researchers should first carefully assess the safety and fairness of the model in relation to the specific context they’re being deployed within.

## How to Get Started with the Model

Please check out the official repository: https://github.com/yukw777/EILEV