File size: 4,199 Bytes
064e744 fedff3e 064e744 fd01149 a359739 543bb42 fd01149 b83932c 3420145 b83932c fd01149 6846de5 fd01149 6846de5 fd01149 6846de5 fd01149 6846de5 fd01149 20abc8b b08ccca fd01149 543bb42 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 |
---
license: mit
datasets:
- liuhaotian/LLaVA-Pretrain
- liuhaotian/LLaVA-Instruct-150K
language:
- en
- zh
library_name: transformers
---
# WORK IN PROGRESS
We present TinyLLaVA, a small vision-language chatbot (1.4B) that reaches comparable performances with contemporary vision language models on common benchmarks, using less parameters.
TinyLLaVA was trained by finetuning [TinyLlama](https://huggingface.co/PY007/TinyLlama-1.1B-Chat-v0.3) on the [LLaVA-1.5](https://github.com/haotian-liu/LLaVA) dataset, following the training recipe of [LLaVA-1.5](https://github.com/haotian-liu/LLaVA). For more details, please refer to the [LLaVA-1.5 paper](https://arxiv.org/abs/2310.03744).
## Model Performance
We have evaluated TinyLLaVA on [GQA](https://cs.stanford.edu/people/dorarad/gqa/about.html), [VizWiz](https://www.vizwiz.com/), [VQAv2](https://visualqa.org/), [TextVQA](https://textvqa.org/) and [SQA](https://github.com/lupantech/ScienceQA).
| Model | VQAv2 | GQA | SQA | TextVQA | VizWiz |
| -------------------- | :------------: | :------------: | :------------: | :------------: | :------------: |
| TinyLLaVA-v1-1.4B | 73.41 | 57.54 | 59.40 | 46.37 | 49.56 |
| BLIP-2 | 41.00 | 41.00 | 61.00 | 42.50 | 19.60 |
| LLaVA-v1.5-7B | 78.50 | 62.00 | 66.80 | 61.3 | 50 |
| LLaVA-v1.5-13B | 80.00 | 63.30 | 71.60 | 61.3 | 53.6 |
| Qwen-VL-7B | 78.80 | 59.30 | 67.10 | 63.8 | 35.2 |
| Qwen-VL-13B | 78.20 | 57.50 | 68.20 | 61.5 | 38.9 |
More evaluations are ongoing.
## Model Preparations
### Transformers Version
Make sure to have `transformers >= 4.35.3`.
### Prompt Template
The model supports multi-image and multi-prompt generation. When using the model, make sure to follow the correct prompt template (`USER: <image>xxx\nASSISTANT:`), where `<image>` token is a place-holding special token for image embeddings.
## Model Inference from `pipeline` and `transformers`
### Using `pipeline`:
Below we used [`"bczhou/tiny-llava-v1-hf"`](https://huggingface.co/bczhou/tiny-llava-v1-hf) checkpoint.
```python
from transformers import pipeline
from PIL import Image
import requests
model_id = "bczhou/tiny-llava-v1-hf"
pipe = pipeline("image-to-text", model=model_id)
url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/ai2d-demo.jpg"
image = Image.open(requests.get(url, stream=True).raw)
prompt = "USER: <image>\nWhat does the label 15 represent? (1) lava (2) core (3) tunnel (4) ash cloud\nASSISTANT:"
outputs = pipe(image, prompt=prompt, generate_kwargs={"max_new_tokens": 200})
print(outputs[0])
>>> {"generated_text': 'USER: \nWhat does the label 15 represent? (1) lava (2) core (3) tunnel (4) ash cloud\nASSISTANT: The label 15 represents lava, which is a type of volcanic rock."}
```
### Using pure `transformers`:
Below is an example script to run generation in `float16` precision on a GPU device:
```python
import requests
from PIL import Image
import torch
from transformers import AutoProcessor, LlavaForConditionalGeneration
model_id = "bczhou/tiny-llava-v1-hf"
prompt = "USER: <image>\nWhat are these?\nASSISTANT:"
image_file = "http://images.cocodataset.org/val2017/000000039769.jpg"
model = LlavaForConditionalGeneration.from_pretrained(
model_id,
torch_dtype=torch.float16,
low_cpu_mem_usage=True,
).to(0)
processor = AutoProcessor.from_pretrained(model_id)
raw_image = Image.open(requests.get(image_file, stream=True).raw)
inputs = processor(prompt, raw_image, return_tensors='pt').to(0, torch.float16)
output = model.generate(**inputs, max_new_tokens=200, do_sample=False)
print(processor.decode(output[0][2:], skip_special_tokens=True))
```
## Contact
This model was trained by [Baichuan Zhou](https://baichuanzhou.github.io/), from Beihang Univerisity, under the supervision of [Prof. Lei Huang](https://huangleibuaa.github.io/).
|