license: apache-2.0
pipeline_tag: visual-question-answering
😈Imp
The Imp project aims to provide a family of a strong multimodal small
language models (MSLMs). Our imp-v0-3b
is a strong MSLM with only 3B parameters, which is build upon a small yet powerful SLM Phi-2 (2.7B) and a powerful visual encoder SigLIP (0.4B), and trained on the LLaVA-v1.5 training set.
As shown in the Table below, imp-v0-3b
significantly outperforms the counterparts of similar model sizes, and even achieves slightly better performance than the strong LLaVA-7B model on various multimodal benchmarks.
We release our model weights and provide an example below to run our model . Detailed technical report and corresponding training/evaluation code will be released soon on our GitHub repo. We will persistently improve our model and release the next versions to further improve model performance :)
How to use
You can use the following code for model inference. We minimize the required dependency libraries that only the transformers
and torch
packages are used. The format of text instructions is similar to LLaVA.
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from PIL import Image
torch.set_default_device("cuda")
#Create model
model = AutoModelForCausalLM.from_pretrained(
"MILVLG/imp-v0-3b",
torch_dtype=torch.float16,
device_map="auto",
trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("MILVLG/imp-v0-3b", trust_remote_code=True)
#Set inputs
text = "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: <image>\nWhat are the colors of the bus in the image? ASSISTANT:"
image = Image.open("images/bus.jpg")
input_ids = tokenizer(text, return_tensors='pt').input_ids
image_tensor = model.image_preprocess(image)
#Generate the answer
output_ids = model.generate(
input_ids,
max_new_tokens=100,
images=image_tensor,
use_cache=True)[0]
print(tokenizer.decode(output_ids[input_ids.shape[1]:], skip_special_tokens=True).strip())
Model evaluation
We perform evaluation on 9 commonly-used benchmarks to validate the effectiveness of our model, including 5 academic VQA benchmarks and 4 popular MLLM benchmarks.
Models | Size | VQAv2 | GQA | VisWiz | SQA (IMG) | TextVQA | POPE | MME | MMB | MM-Vet |
---|---|---|---|---|---|---|---|---|---|---|
LLaVA-v1.5-lora | 7B | 79.10 | 63.00 | 47.80 | 68.40 | 58.20 | 86.40 | 1476.9 | 66.10 | 30.2 |
TinyGPT-V | 3B | - | 33.60 | 24.80 | - | - | - | - | - | - |
LLaVA-Phi | 3B | 71.40 | - | 35.90 | 68.40 | 48.60 | 85.00 | 1335.1 | 59.80 | 28.9 |
MobileVLM | 3B | - | 59.00 | - | 61.00 | 47.50 | 84.90 | 1288.9 | 59.60 | - |
MC-LLaVA-3b | 3B | 64.24 | 49.6 | 24.88 | - | 38.59 | 80.59 | - | - | - |
Imp-v0 (ours) | 3B | 79.45 | 58.55 | 50.09 | 69.96 | 59.38 | 88.02 | 1434 | 66.49 | 33.1 |
License
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
About us
This project is maintained by the MILVLG@Hangzhou Dianzi University (HDU) led by Prof. Zhou Yu and Jun Yu, and is mainly developed by Zhenwei Shao and Xuecheng Ouyang. We hope our model may serve as a strong baseline to inspire future research on MSLM, as well as its derivative applications on mobile devices and robots.