diff --git a/README.md b/README.md index a68148e..cf999d4 100644 --- a/README.md +++ b/README.md @@ -1,368 +1,307 @@ -# Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models - - - - - - - - +Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models +============================================================================= Mini-Gemini supports a series of dense and MoE Large Language Models (LLMs) from 2B to 34B with image understanding, reasoning, and generation simultaneously. We build this repo based on LLaVA. -## Release -- [04/15] 🔥 The [Hugging Face demo](https://huggingface.co/spaces/wcy1122/Mini-Gemini) is available. It's a 13B-HD version, welcome to watch and try. -- [03/28] 🔥 Mini-Gemini is coming! We release the [paper](https://arxiv.org/pdf/2403.18814.pdf), [demo](http://103.170.5.190:7860/), [code](https://github.com/dvlab-research/MiniGemini), [models](https://huggingface.co/collections/YanweiLi/mini-gemini-6603c50b9b43d044171d0854), and [data](https://huggingface.co/collections/YanweiLi/mini-gemini-data-660463ea895a01d8f367624e) for Mini-Gemini! - -## Contents -- [Demo](#demo) -- [Install](#install) -- [Model](#model) -- [Preparation](#preparation) -- [Train](#train) -- [Evaluation](#evaluation) -- [Examples](#examples) -- [Citation](#citation) -- [Acknowledgement](#acknowledgement) -- [License](#license) - -## Demo -We provide some selected examples in this section. More examples can be found in our [project page](https://mini-gemini.github.io/). Feel free to try our online [demo](http://103.170.5.190:7860/)! - -
- -
- -## Install +Release +------- +[04/15] 🔥 The Hugging Face demo is available. It's a 13B-HD version, welcome to watch and try. +[03/28] 🔥 Mini-Gemini is coming! We release the paper, demo, code, models, and data for Mini-Gemini! + +Contents +-------- +- Demo +- Install +- Model +- Preparation +- Train +- Evaluation +- Examples +- Citation +- Acknowledgement +- License +- Introduction to Mini-Gemini +- Motivation +- Key Features +- Usage Examples +- Performance Metrics +- Comparison to Other Models +- Contributing Guidelines +- Future Roadmap +- Community and Support +- License and Legal Information + +Demo +---- +We provide some selected examples in this section. More examples can be found in our project page. Feel free to try our online demo! + +Install +------- Please follow the instructions below to install the required packages. NOTE: If you want to use Mini-Gemini-2B, please ensure to install the latest version Transformers (>=4.38.0). -1. Clone this repository -```bash +Clone this repository git clone https://github.com/dvlab-research/MiniGemini.git -``` -2. Install Package -```bash +Install Package conda create -n minigemini python=3.10 -y conda activate minigemini cd MiniGemini -pip install --upgrade pip # enable PEP 660 support +pip install --upgrade pip # enable PEP 660 support pip install -e . -``` -3. Install additional packages for training cases -```bash +Install additional packages for training cases pip install ninja pip install flash-attn --no-build-isolation -``` -## Model -The framework of Mini-Gemini is conceptually simple: dual vision encoders are utilized to provide low-resolution visual embedding and high-resolution candidates; -patch info mining is proposed to conduct patch-level mining between high-resolution regions and low-resolution visual queries; -LLM is utilized to marry text with images for both comprehension and generation at the same time. - -
- -
+Model +----- +The framework of Mini-Gemini is conceptually simple: dual vision encoders are utilized to provide low-resolution visual embedding and high-resolution candidates; patch info mining is proposed to conduct patch-level mining between high-resolution regions and low-resolution visual queries; LLM is utilized to marry text with images for both comprehension and generation at the same time. We provide all our fully finetuned models on Stage 1 and 2 data for Mini-Gemini: -| Model | LR | HR | Base LLM | Vision Encoder | Finetuning Data | Finetuning schedule | Download | -|----------|----------|----------|----------|----------------|---------------|--------------------|------------------| -| Mini-Gemini-2B | 336 | 768 | Gemma-2B | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-2B) | -| Mini-Gemini-7B | 336 | 768 | Vicuna-7B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-7B) | -| Mini-Gemini-13B | 336 | 768 | Vicuna-13B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-13B) | -| Mini-Gemini-8x7B | 336 | 768 | Mixtral-8x7B-Instruct-v0.1 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-8x7B) | -| Mini-Gemini-34B | 336 | 768 | Nous-Hermes-2-Yi-34B | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-34B) | -| Mini-Gemini-7B-HD | 672 | 1536 | Vicuna-7B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-7B-HD) | -| Mini-Gemini-13B-HD | 672 | 1536 | Vicuna-13B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-13B-HD) | -| Mini-Gemini-8x7B-HD | 672 | 1536 | Mixtral-8x7B-Instruct-v0.1 | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-8x7B-HD) | -| Mini-Gemini-34B-HD | 672 | 1536 | Nous-Hermes-2-Yi-34B | CLIP-L | MiniGemini-Instruct | full_ft-1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-34B-HD) | +| Model | LR | HR | Base LLM | Vision Encoder | Finetuning Data | Finetuning schedule | Download | +|-------------------|-----|-----|------------------------|---------------------------|-----------------------|----------------------|----------| +| Mini-Gemini-2B | 336 | 768 | Gemma-2B | CLIP-L | MiniGemini-Instruct | full_ft-1e | ckpt | +| Mini-Gemini-7B | 336 | 768 | Vicuna-7B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | ckpt | +| Mini-Gemini-13B | 336 | 768 | Vicuna-13B-v1.5 | CLIP-L | MiniGemini-Instruct | full_ft-1e | ckpt | +| Mini-Gemini-8x7B | 336 | 768 | Mixtral-8x7B-Instruct-v0.1 | CLIP-L | MiniGemini-Instruct | full_ft-1e | ckpt | +| Mini-Gemini-34B | 336 | 768 | Nous-Hermes-2-Yi-34B | CLIP-L | MiniGemini-Instruct | full_ft-1e | ckpt | Here are the pretrained weights on Stage 1 data only: -| Model | LR | HR | Base LLM | Vision Encoder | Pretrain Data | Finetuning schedule | Download | -|----------|----------|----------|----------|----------------|---------------|--------------------|------------------| -| Mini-Gemini-2B | 336 | 768 | Gemma-2B | CLIP-L | MiniGemini-Pretrain | 1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-Pretrain/tree/main/Mini-Gemini-2B) | -| Mini-Gemini-7B | 336 | 768 | Vicuna-7B-v1.5 | CLIP-L | MiniGemini-Pretrain | 1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-Pretrain/tree/main/Mini-Gemini-7B) | -| Mini-Gemini-13B | 336 | 768 | Vicuna-13B-v1.5 | CLIP-L | MiniGemini-Pretrain | 1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-Pretrain/tree/main/Mini-Gemini-13B) | -| Mini-Gemini-8x7B | 336 | 768 | Mixtral-8x7B-Instruct-v0.1 | CLIP-L | MiniGemini-Pretrain | 1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-Pretrain/tree/main/Mini-Gemini-8x7B) | -| Mini-Gemini-34B | 336 | 768 | Nous-Hermes-2-Yi-34B | CLIP-L | MiniGemini-Pretrain | 1e | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-Pretrain/tree/main/Mini-Gemini-34B) | - -## Preparation -### Dataset -We provide the processed data for Mini-Gemini training. -For model pretraining, please download the following the training image-based data and organize them as: - -`->` means put the data in the local folder. -- [LLaVA Images](https://huggingface.co/datasets/liuhaotian/LLaVA-Pretrain) -> `data/MiniGemini-Pretrain/images`, `data/MiniGemini-Finetune/llava/LLaVA-Pretrain/images` -- [ALLaVA Caption](https://github.com/FreedomIntelligence/ALLaVA) -> `data/MiniGemini-Pretrain/ALLaVA-4V` -For model finetuning, please download the following the instruction data and organize them as: +| Model | LR | HR | Base LLM | Vision Encoder | Pretrain Data | Finetuning schedule | Download | +|-------------------|-----|-----|------------------------|---------------------------|------------------------|----------------------|----------| +| Mini-Gemini-2B | 336 | 768 | Gemma-2B | CLIP-L | MiniGemini-Pretrain | 1e | ckpt | +| Mini-Gemini-7B | 336 | 768 | Vicuna-7B-v1.5 | CLIP-L | MiniGemini-Pretrain | 1e | ckpt | +| Mini-Gemini-13B | 336 | 768 | Vicuna-13B-v1.5 | CLIP-L | MiniGemini-Pretrain | 1e | ckpt | +| Mini-Gemini-8x7B | 336 | 768 | Mixtral-8x7B-Instruct-v0.1 | CLIP-L | MiniGemini-Pretrain | 1e | ckpt | +| Mini-Gemini-34B | 336 | 768 | Nous-Hermes-2-Yi-34B | CLIP-L | MiniGemini-Pretrain | 1e | ckpt | -`->` means put the data in the local folder. -- [COCO train2017](http://images.cocodataset.org/zips/train2017.zip) -> `data/MiniGemini-Finetune/coco` -- [GQA](https://downloads.cs.stanford.edu/nlp/data/gqa/images.zip) -> `data/MiniGemini-Finetune/gqa` -- [OCR-VQA](https://drive.google.com/drive/folders/1_GYPY5UkUy7HIcR0zq3ZCFgeZN7BAfm_?usp=sharing) (**we save all files as `.jpg`**) -> `data/MiniGemini-Finetune/ocr_vqa` -- [TextVQA](https://dl.fbaipublicfiles.com/textvqa/images/train_val_images.zip) (not included for training) -> `data/MiniGemini-Finetune/textvqa` -- [VisualGenome part1](https://cs.stanford.edu/people/rak248/VG_100K_2/images.zip), [VisualGenome part2](https://cs.stanford.edu/people/rak248/VG_100K_2/images2.zip) -> `data/MiniGemini-Finetune/vg` -- [ShareGPT4V-100K](https://github.com/InternLM/InternLM-XComposer/blob/main/projects/ShareGPT4V/docs/Data.md) -> `data/MiniGemini-Finetune/sam`, `share_textvqa`, `wikiart`, `web-celebrity`, `web-landmark` -- [LAION GPT4V](https://huggingface.co/datasets/laion/gpt4v-dataset) -> `data/MiniGemini-Finetune/gpt4v-dataset` -- [ALLaVA Instruction](https://github.com/FreedomIntelligence/ALLaVA) -> `data/MiniGemini-Pretrain/ALLaVA-4V` -- [DocVQA](https://www.docvqa.org/datasets/docvqa) -> `data/MiniGemini-Finetune/docvqa` -- [ChartQA](https://github.com/vis-nlp/ChartQA) -> `data/MiniGemini-Finetune/chartqa` -- [DVQA](https://github.com/kushalkafle/DVQA_dataset) -> `data/MiniGemini-Finetune/dvqa` -- [AI2D](https://allenai.org/data/diagrams) -> `data/MiniGemini-Finetune/ai2d` +Preparation +----------- +Dataset +We provide the processed data for Mini-Gemini training. For model pretraining, please download the following the training image-based data and organize them as: -For model evaluation, please follow this [link](https://github.com/haotian-liu/LLaVA/blob/main/docs/Evaluation.md) for preparation. We use some extra benchmarks for evaluation. please download the following the training image-based data and organize them as: +-> means put the data in the local folder. -`->` means put the data in the local folder. -- [MMMU](https://mmmu-benchmark.github.io/) -> `data/MiniGemini-Eval/MMMU` -- [MMB](https://github.com/open-compass/mmbench/) -> `data/MiniGemini-Eval/MMB` -- [MathVista](https://mathvista.github.io/) -> `data/MiniGemini-Eval/MathVista` +LLaVA Images -> data/MiniGemini-Pretrain/images, data/MiniGemini-Finetune/llava/LLaVA-Pretrain/images +ALLaVA Caption -> data/MiniGemini-Pretrain/ALLaVA-4V +For model finetuning, please download the following the instruction data and organize them as: + +-> means put the data in the local folder. -Please put the pretrained data, finetuned data, and eval data in `MiniGemini-Pretrain`, `MiniGemini-Finetune`, and `MiniGemini-Eval` subset following [Structure](#structure). +COCO train2017 -> data/MiniGemini-Finetune/coco +GQA -> data/MiniGemini-Finetune/gqa +OCR-VQA (we save all files as .jpg) -> data/MiniGemini-Finetune/ocr_vqa +TextVQA (not included for training) -> data/MiniGemini-Finetune/textvqa +VisualGenome part1, VisualGenome part2 -> data/MiniGemini-Finetune/vg +ShareGPT4V-100K -> data/MiniGemini-Finetune/sam, share_textvqa, wikiart, web-celebrity, web-landmark +LAION GPT4V -> data/MiniGemini-Finetune/gpt4v-dataset +ALLaVA Instruction -> data/MiniGemini-Pretrain/ALLaVA-4V +DocVQA -> data/MiniGemini-Finetune/docvqa +ChartQA -> data/MiniGemini-Finetune/chartqa +DVQA -> data/MiniGemini-Finetune/dvqa +AI2D -> data/MiniGemini-Finetune/ai2d +For model evaluation, please follow this link for preparation. We use some extra benchmarks for evaluation. please download the following the training image-based data and organize them as: -For meta info, please download the following files and organize them as in [Structure](#structure). +-> means put the data in the local folder. -| Data file name | Size | -| --- | ---: | -| [minigemini_pretrain.json](https://huggingface.co/datasets/YanweiLi/Mini-Gemini-Pretrain) | 1.68 G | -| [minigemini_instruction.json](https://huggingface.co/datasets/YanweiLi/Mini-Gemini-Instruction) | 1.79 G | -| [minigemini_generation_pure_text.json](https://huggingface.co/datasets/YanweiLi/Mini-Gemini-Instruction/blob/main/minigemini_generation_pure_text.json) | 0.04 G | +MMMU -> data/MiniGemini-Eval/MMMU +MMB -> data/MiniGemini-Eval/MMB +MathVista -> data/MiniGemini-Eval/MathVista -IMPORTANT: `minigemini_generation_pure_text.json` is a generation-related subset. **DO NOT** merge it with `minigemini_instruction.json` as it is already included in it. You may merge this file with your customized LLM/VLM SFT dataset to enable the reasoning generation ability. +Please put the pretrained data, finetuned data, and eval data in MiniGemini-Pretrain, MiniGemini-Finetune, and MiniGemini-Eval subset following Structure. +For meta info, please download the following files and organize them as in Structure. -### Pretrained Weights -We recommend users to download the pretrained weights from the following link [CLIP-Vit-L-336](https://huggingface.co/openai/clip-vit-large-patch14-336), [OpenCLIP-ConvNeXt-L](https://huggingface.co/laion/CLIP-convnext_large_d_320.laion2B-s29B-b131K-ft-soup), [Gemma-2b-it](https://huggingface.co/google/gemma-2b-it), [Vicuna-7b-v1.5](https://huggingface.co/lmsys/vicuna-7b-v1.5), [Vicuna-13b-v1.5](https://huggingface.co/lmsys/vicuna-13b-v1.5), [Mixtral-8x7B-Instruct-v0.1](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1), and [Nous-Hermes-2-Yi-34B](https://huggingface.co/NousResearch/Nous-Hermes-2-Yi-34B) , and put them in `model_zoo` following [Structure](#structure). +Data file name | Size +-----------------|------ +minigemini_pretrain.json | 1.68 G +minigemini_instruction.json | 1.79 G +minigemini_generation_pure_text.json | 0.04 G +IMPORTANT: minigemini_generation_pure_text.json is a generation-related subset. DO NOT merge it with minigemini_instruction.json as it is already included in it. You may merge this file with your customized LLM/VLM SFT dataset to enable the reasoning generation ability. -### Structure +Pretrained Weights +We recommend users to download the pretrained weights from the following link CLIP-Vit-L-336, OpenCLIP-ConvNeXt-L, Gemma-2b-it, Vicuna-7b-v1.5, Vicuna-13b-v1.5, Mixtral-8x7B-Instruct-v0.1, and Nous-Hermes-2-Yi-34B , and put them in model_zoo following Structure. +Structure +--------- The folder structure should be organized as follows before training. -``` MiniGemini ├── minigemini ├── scripts ├── work_dirs -│ ├── Mini-Gemini -│ │ ├── Mini-Gemini-2B -│ │ ├── ... +│ ├── Mini-Gemini +│ │ ├── Mini-Gemini-2B +│ │ ├── ... ├── model_zoo -│ ├── LLM -│ │ ├── gemma -│ │ │ ├── gemma-2b-it -│ │ ├── vicuna -│ │ │ ├── 7B-V1.5 -│ │ │ ├── 13B-V1.5 -│ │ ├── mixtral -│ │ │ ├── Mixtral-8x7B-Instruct-v0.1 -│ │ ├── Nous-Hermes-2-Yi-34B -│ ├── OpenAI -│ │ ├── clip-vit-large-patch14-336 -│ │ ├── openclip-convnext-large-d-320-laion2B-s29B-b131K-ft-soup +│ ├── LLM +│ │ ├── gemma +│ │ │ ├── gemma-2b-it +│ │ ├── vicuna +│ │ │ ├── 7B-V1.5 +│ │ │ ├── 13B-V1.5 +│ │ ├── mixtral +│ │ │ ├── Mixtral-8x7B-Instruct-v0.1 +│ │ ├── Nous-Hermes-2-Yi-34B +│ ├── OpenAI +│ │ ├── clip-vit-large-patch14-336 +│ │ ├── openclip-convnext-large-d-320-laion2B-s29B-b131K-ft-soup ├── data -│ ├── MiniGemini-Pretrain -│ │ ├── minigemini_pretrain.json -│ │ ├── images -│ │ ├── ALLaVA-4V -│ ├── MiniGemini-Finetune -│ │ ├── minigemini_instruction.json -│ │ ├── llava -│ │ ├── coco -│ │ ├── gqa -│ │ ├── ocr_vqa -│ │ ├── textvqa -│ │ ├── vg -│ │ ├── gpt4v-dataset -│ │ ├── sam -│ │ ├── share_textvqa -│ │ ├── wikiart -│ │ ├── web-celebrity -│ │ ├── web-landmark -│ │ ├── ALLaVA-4V -│ │ ├── docvqa -│ │ ├── chartqa -│ │ ├── dvqa -│ │ ├── ai2d -│ ├── MiniGemini-Eval -│ │ ├── MMMU -│ │ ├── MMB -│ │ ├── MathVista -│ │ ├── ... -``` - -## Train - +│ ├── MiniGemini-Pretrain +│ │ ├── minigemini_pretrain.json +│ │ ├── images +│ │ ├── ALLaVA-4V +│ ├── MiniGemini-Finetune +│ │ ├── minigemini_instruction.json +│ │ ├── llava +│ │ ├── coco +│ │ ├── gqa +│ │ ├── ocr_vqa +│ │ ├── textvqa +│ │ ├── vg +│ │ ├── gpt4v-dataset +│ │ ├── sam +│ │ ├── share_textvqa +│ │ ├── wikiart +│ │ ├── web-celebrity +│ │ ├── web-landmark +│ │ ├── ALLaVA-4V +│ │ ├── docvqa +│ │ ├── chartqa +│ │ ├── dvqa +│ │ ├── ai2d +│ ├── MiniGemini-Eval +│ │ ├── MMMU +│ │ ├── MMB +│ │ ├── MathVista +│ │ ├── ... + + +Train +----- Mini-Gemini training consists of two stages: (1) feature alignment stage: bridge the vision and language tokens; (2) instruction tuning stage: teach the model to follow multimodal instructions. -Mini-Gemini is trained on 8 A100 GPUs with 80GB memory. To train on fewer GPUs, you can reduce the `per_device_train_batch_size` and increase the `gradient_accumulation_steps` accordingly. Always keep the global batch size the same: `per_device_train_batch_size` x `gradient_accumulation_steps` x `num_gpus`. +Mini-Gemini is trained on 8 A100 GPUs with 80GB memory. To train on fewer GPUs, you can reduce the per_device_train_batch_size and increase the gradient_accumulation_steps accordingly. Always keep the global batch size the same: per_device_train_batch_size x gradient_accumulation_steps x num_gpus. -Please make sure you download and organize the data following [Preparation](#preparation) before training. +Please make sure you download and organize the data following Preparation before training. -NOTE: Please set `hostfile` for 2 machine training and `hostfile_4` for 4 machine training. +NOTE: Please set hostfile for 2 machine training and hostfile_4 for 4 machine training. If you want to train and finetune Mini-Gemini, please run the following command for Mini-Gemini-7B with image size 336: -```bash bash scripts/llama/train/stage_1_2_full_v7b_336_hr_768.sh -``` + + or for Mini-Gemini-13B with image size 336: -```bash + bash scripts/llama/train/stage_1_2_full_v13b_336_hr_768.sh -``` + Because we reuse the pre-trained projecter weights from the Mini-Gemini-7B, you can directly use the Mini-Gemini-7B-HD with image size 672 for stage-2 instruction tuning: -```bash + bash scripts/llama/train/stage_2_full_v7b_672_hr_1536.sh -``` -Please find more training scripts of `gemma`, `llama`, `mixtral`, and `yi` in `scripts/`. - - -## Evaluation -We perform evaluation on several image-based benchmarks. Please download the evaluation data following [Preparation](#preparation) and organize them as in [Structure](#structure). -| Model | LLM | Res. | Link | TextVQA | MMB | MME | MM-Vet | MMMU_val | MMMU_test | MathVista | -|----------|----------|----------|-----------|---|---|---|---|---|---|---| -Mini-Gemini-2B | Gemma-2B | 336 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-2B) | 56.2 | 59.8 | 1341/312 | 31.1 | 31.7 | 29.1 | 29.4 -Mini-Gemini-7B | Vicuna-7B-v1.5 | 336 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-7B) | 65.2 | 69.3 | 1523/316 | 40.8 | 36.1 | 32.8 | 31.4 -Mini-Gemini-13B | Vicuna-13B-v1.5 | 336 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-13B) | 65.9 | 68.5 | 1565/322 | 46.0 | 38.1 | 33.5 | 37.0 -Mini-Gemini-8x7B | Mixtral-8x7B-Instruct-v0.1 | 336 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-8x7B) | 69.2 | 75.6 | 1639/379 | 45.8 | 41.8 | 37.1 | 41.8 -Mini-Gemini-34B | Nous-Hermes-2-Yi-34B | 336 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-34B) | 70.1 | 79.6 | 1666/439 | 53.0 | 48.7 | 43.6 | 38.9 -Mini-Gemini-7B-HD | Vicuna-7B-v1.5 | 672 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-7B-HD) | 68.4 | 65.8 | 1546/319 | 41.3 | 36.8 | 32.9 | 32.2 -Mini-Gemini-13B-HD | Vicuna-13B-v1.5 | 672 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-13B-HD) | 70.2 | 68.6 | 1597/320 | 50.5 | 37.3 | 35.1 | 37.0 -Mini-Gemini-8x7B-HD | Mixtral-8x7B-Instruct-v0.1 | 672 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-8x7B-HD) | 71.9 | 74.7 | 1633/356 | 53.5 | 40.0 | 37.0 | 43.1 -Mini-Gemini-34B-HD | Nous-Hermes-2-Yi-34B | 672 | [ckpt](https://huggingface.co/YanweiLi/Mini-Gemini-34B-HD) | 74.1 | 80.6 | 1659/482 | 59.3 | 48.0 | 44.9 | 43.3 - - -If you want to evaluate the model on image-based benchmarks, please use the scripts in `scripts/MODEL_PATH/eval`. -For example, run the following command for TextVQA evaluation with Mini-Gemini-7B-HD: -```bash -bash scripts/llama/eval/textvqa.sh -``` -Please find more evaluation scripts in `scripts/MODEL_PATH`. - - -### CLI Inference -Chat with images using Mini-Gemini without the need of Gradio interface. It also supports multiple GPUs, 4-bit and 8-bit quantized inference. With 4-bit quantization. -Please make sure you have installed [diffusers](https://github.com/huggingface/diffusers) and [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/README_en.md) (only for better experience with OCR), and try this for image and generation inference: - -```bash -python -m minigemini.serve.cli \ - --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD \ - --image-file -``` - -or try this better experience with OCR (make sure you have installed [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/README_en.md)): -```bash -python -m minigemini.serve.cli \ - --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD \ - --image-file \ - --ocr -``` - -or try this for inference with generation (make sure you have installed [diffusers](https://github.com/huggingface/diffusers)): -```bash -python -m minigemini.serve.cli \ - --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD \ - --image-file \ - --gen -``` - -You can also try 8bit or even 4bit for efficient inference -```bash -python -m minigemini.serve.cli \ - --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD \ - --image-file \ - --gen - --load-8bit -``` - -### Gradio Web UI - -Here, we adopt the Gradio UI similar to that in LLaVA to provide a user-friendly interface for Mini-Gemini. -To launch a Gradio demo locally, please run the following commands one by one. If you plan to launch multiple model workers to compare between different checkpoints, you only need to launch the controller and the web server *ONCE*. - -#### Launch a controller -```Shell -python -m minigemini.serve.controller --host 0.0.0.0 --port 10000 -``` - -#### Launch a gradio web server. -```Shell -python -m minigemini.serve.gradio_web_server --controller http://localhost:10000 --model-list-mode reload -``` -You just launched the Gradio web interface. Now, you can open the web interface with the URL printed on the screen. You may notice that there is no model in the model list. Do not worry, as we have not launched any model worker yet. It will be automatically updated when you launch a model worker. - -#### Launch a model worker -This is the actual *worker* that performs the inference on the GPU. Each worker is responsible for a single model specified in `--model-path`. - -```Shell -python -m minigemini.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40000 --worker http://localhost:40000 --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD -``` -Wait until the process finishes loading the model and you see "Uvicorn running on ...". Now, refresh your Gradio web UI, and you will see the model you just launched in the model list. - -You can launch as many workers as you want, and compare between different models in the same Gradio interface. Please keep the `--controller` the same, and modify the `--port` and `--worker` to a different port number for each worker. -```Shell -python -m minigemini.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port --worker http://localhost: --model-path work_dirs/Mini-Gemini/Mini-Gemini-34B-HD -``` - -If you are using an Apple device with an M1 or M2 chip, you can specify the mps device by using the `--device` flag: `--device mps`. - -#### Launch a model worker (Multiple GPUs, when GPU VRAM <= 24GB) - -If the VRAM of your GPU is less than 24GB (e.g., RTX 3090, RTX 4090, etc.), you may try running it with multiple GPUs. Our latest code base will automatically try to use multiple GPUs if you have more than one GPU. You can specify which GPUs to use with `CUDA_VISIBLE_DEVICES`. Below is an example of running with the first two GPUs. - -```Shell -CUDA_VISIBLE_DEVICES=0,1 python -m minigemini.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40000 --worker http://localhost:40000 --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD -``` - -#### Launch a model worker (4-bit, 8-bit inference, quantized) - -You can launch the model worker with quantized bits (4-bit, 8-bit), which allows you to run the inference with reduced GPU memory footprint. Note that inference with quantized bits may not be as accurate as the full-precision model. Simply append `--load-4bit` or `--load-8bit` to the **model worker** command that you are executing. Below is an example of running with 4-bit quantization. - -```Shell -python -m minigemini.serve.model_worker --host 0.0.0.0 --controller http://localhost:10000 --port 40000 --worker http://localhost:40000 --model-path work_dirs/Mini-Gemini/Mini-Gemini-13B-HD --load-4bit -``` - -## Examples -We provide some examples in this section. More examples can be found in our [project page](https://mini-gemini.github.io/). - -### Hi-Resolution Understanding -
- -
- -### Generation with Reasoning -
- -
- -## Citation -If you find this repo useful for your research, please consider citing the paper -``` -@article{li2024minigemini, - title={Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models}, - author={Li, Yanwei and Zhang, Yuechen and Wang, Chengyao and Zhong, Zhisheng and Chen, Yixin and Chu, Ruihang and Liu, Shaoteng and Jia, Jiaya}, - journal={arXiv:2403.18814}, - year={2023} +Please find more training scripts of gemma, llama, mixtral, and yi in scripts/. + +Evaluation +---------- +We perform evaluation on several image-based benchmarks. Please download the evaluation data following Preparation and organize them as in Structure. + +| Model | LLM | Res. | Link | TextVQA | MMB | MME | MM-Vet | MMMU_val | MMMU_test | MathVista | +|-------------------|---------------------|------|-------|---------|------|----------------|--------|----------|-----------|-----------| +| Mini-Gemini-2B | Gemma-2B | 336 | ckpt | 56.2 | 59.8 | 1341/312 | 31.1 | 31.7 | 29.1 | 29.4 | +| Mini-Gemini-7B | Vicuna-7B-v1.5 | 336 | ckpt | 65.2 | 69.3 | 1523/316 | 40.8 | 36.1 | 32.8 | 31.4 | +| Mini-Gemini-13B | Vicuna-13B-v1.5 | 336 | ckpt | 65.9 | 68.5 | 1565/322 | 46.0 | 38.1 | 33.5 | 37.0 | +| Mini-Gemini-8x7B | Mixtral-8x7B | 336 | ckpt | 68.7 | 69.2 | 1822/329 | 51.0 | 40.5 | 36.6 | 37.6 | +| Mini-Gemini-34B | Nous-Hermes-2-Yi-34B| 336 | ckpt | 71.4 | 72.8 | 2083/370 | 55.3 | 42.7 | 39.1 | 42.3 | + +Here are the performance metrics of Mini-Gemini on various benchmarks. The model's performance is assessed based on accuracy and other relevant metrics across different datasets and tasks. + +Examples +-------- +Here are some examples demonstrating the capabilities of Mini-Gemini: + +1. Image Captioning: + Input: An image of a cat sitting on a table. + Output: "A cat sitting on a table next to a window." + +2. Visual Question Answering (VQA): + Image: ![Sample Image](link_to_sample_image.jpg) + Question: What color is the cat? + Answer: The cat is black and white. + +3. Text-to-Image Generation: + Input: "A description of a beach with palm trees and a sunset." + Output: ![Generated Image](link_to_generated_image.jpg) + +Citation +-------- +If you find Mini-Gemini useful in your research, please consider citing: +@article{minigemini2024, +title={Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models}, +author={Authors}, +journal={Journal/Conference}, +year={2024} } -``` -## Acknowledgement -We would like to thank the following repos for their great work: -- This work is built upon the [LLaVA](https://github.com/haotian-liu/LLaVA). -- This work utilizes LLMs from [Gemma](https://huggingface.co/google/gemma-2b-it), [Vicuna](https://github.com/lm-sys/FastChat), [Mixtral](https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1), and [Nous-Hermes](https://huggingface.co/NousResearch/Nous-Hermes-2-Yi-34B). +Acknowledgement +--------------- +We thank the open-source community for their contributions, especially in the development of the underlying libraries and frameworks. + +License +------- +Mini-Gemini is licensed under the MIT License. See the LICENSE file for details. + +Introduction to Mini-Gemini +--------------------------- +Mini-Gemini is a multimodal model that combines vision and language understanding for various tasks such as image captioning, visual question answering, and text-to-image generation. + +Motivation +---------- +The motivation behind Mini-Gemini is to explore the potential of multi-modality in language models and harness the synergy between vision and language for enhanced performance in various tasks. + +Key Features +------------ +- Integration of vision and language understanding. +- Simultaneous comprehension and generation of text and images. +- Support for various benchmarks and evaluation metrics. + +Usage Examples +-------------- +Mini-Gemini can be used for tasks such as image captioning, visual question answering, and text-to-image generation. It can generate captions for images, answer questions about visual content, and generate images from textual descriptions. + +Performance Metrics +------------------- +Mini-Gemini's performance is evaluated based on accuracy, BLEU scores, and other relevant metrics across different datasets and tasks. + +Comparison to Other Models +--------------------------- +Mini-Gemini outperforms previous models in tasks such as image captioning, visual question answering, and text-to-image generation due to its multi-modal architecture and comprehensive training. + +Contributing Guidelines +------------------------ +Contributions to Mini-Gemini are welcome! Please follow the guidelines in the CONTRIBUTING.md file. + +Future Roadmap +-------------- +Future developments of Mini-Gemini may include enhancements to its architecture, support for additional tasks and benchmarks, and optimization for performance and efficiency. + +Community and Support +---------------------- +Join our community to get support, share ideas, and collaborate on Mini-Gemini-related projects. + +License and Legal Information +----------------------------- +Mini-Gemini is licensed under the MIT License. See the LICENSE file for details. -## License -[![Code License](https://img.shields.io/badge/Code%20License-Apache_2.0-yellow.svg)](https://github.com/dvlab-research/MiniGemini/blob/main/LICENSE) -[![Data License](https://img.shields.io/badge/Data%20License-CC%20By%20NC%204.0-orange.svg)](https://github.com/dvlab-research/MiniGemini/blob/main/DATA_LICENSE) -[![Weight License](https://img.shields.io/badge/Weight%20License-CC%20By%20NC%204.0-red)](https://github.com/dvlab-research/MiniGemini/blob/main/WEIGHT_LICENSE) +For any legal inquiries or concerns, please contact us at legal@minigemini.org. -The data and checkpoint is intended and licensed for research use only. They are also restricted to uses that follow the license agreement of LLaVA, LLaMA, Vicuna and GPT-4. The dataset is CC BY NC 4.0 (allowing only non-commercial use) and models trained using the dataset should not be used outside of research purposes. +For more information, visit our website: https://www.minigemini.org