|
--- |
|
license: mit |
|
language: |
|
- en |
|
pipeline_tag: text-generation |
|
library_name: transformers |
|
tags: |
|
- nlp |
|
- llm |
|
- mllm |
|
datasets: |
|
- MBZUAI/Web2Code |
|
--- |
|
|
|
# CrystalChat-7B-Web2Code: a fully-reproducible vision large language model based on CrystalChat-7B LLM for webpage code generation |
|
|
|
## Model Description |
|
|
|
CrystalChat-7B based multi-modal large language model (MLLM) mimics the training recipe used for Vicuna-7B based [LLaVa-v1.5](https://huggingface.co./docs/transformers/main/model_doc/llava). CrystalChat-7B based MLLMs models are entirely transparent, having open-sourced all materials, including code, data, model checkpoint, intermediate results, and more at [Web2Code: A Large-scale Webpage-to-Code Dataset |
|
and Evaluation Framework for Multimodal LLMs](https://arxiv.org/pdf/2406.20098). CrystalChat-7B-Web2Code MLLM is specialized in webpage images-to-html code generation. |
|
|
|
## Web2Code Dataset |
|
Our Web2Code instruction tuning dataset construction and instruction generation process |
|
involves four key components: |
|
1. Creation of new webpage image-code pair data **(DWCG)**: We generated high-quality HTML webpage-code pairs following the CodeAlpaca prompt using GPT-3.5 and convert them into instruction-following data. |
|
2. Refinement of existing webpage code generation data **(DWCG<sub>R</sub>)**: We transform existing datasets including WebSight and Pix2Code into an instruction- following data format similar to LLaVA data, so they can be used as instruction-following data to train MLLMs. |
|
3. Creation of a new text question-answer pair data **(DWU)**: We generated a new question-answer pair dataset utilizing our new GPT-3.5 generated data for webpage understanding. |
|
4. Refinement of existing webpage understanding data **(DWU<sub>R</sub>)** : We refine the WebSRC question-answer data to improve its quality using the GPT-4. |
|
|
|
The Web2Code instruction tuning dataset was released in [Web2Code: A Large-scale Webpage-to-Code Dataset and Evaluation Framework for Multimodal LLMs](https://huggingface.co./datasets/MBZUAI/Web2Code). |
|
|
|
## Evaluations |
|
|
|
## Webpage Understanding Benchmark (WUB) |
|
### Results |
|
| LLM Backbone | DWCG | DWU | DWCG<sub>R</sub> | DWU<sub>R</sub> | Accuracy (%) | |
|
|------------------------|------|-----|------------------|------------------|--------------| |
|
| **CrystalChat-7B** | | | | | 73.94 | |
|
| | ✓ | ✓ | | | 73.48 | |
|
| | ✓ | ✓ | ✓ | ✓ | **74.14** | |
|
| **Vicuna-7B** | | | | | 71.12 | |
|
| | ✓ | | | | 68.11 | |
|
| | | ✓ | | | 70.82 | |
|
| | ✓ | ✓ | ✓ | ✓ | **71.23** | |
|
| **Llama3-8B** | ✓ | ✓ | ✓ | ✓ | **74.84** | |
|
|
|
**Table 1:** The accuracy of webpage understanding under various data configurations and LLM backbones. |
|
All models are instruction-tuned and evaluated on our WUB benchmark. We note that the general domain data (i.e., LLaVA) is included in all data configuration as default. |
|
|
|
## Webpage Code Generation Benchmark (WCGB) |
|
|
|
Utilizing the same images as the WUB, this benchmark evaluates a multimodal model tasked with generating HTML code from webpage images based on specific instructions. Unlike traditionalcode-level evaluations, this benchmark assesses the generated webpage’s fidelity at the image level. We convert the predicted HTML codes back into images using Selenium WebDriver to allow a direct visual comparison with the ground truth images. The evaluation, depicted on the left side of Figure 6, considers 10 different aspects, which are further categorized into four evaluation matrices using the GPT-4 Vision API. |
|
|
|
### Results |
|
| LLM Backbone | DWCG | DWU | DWCG<sub>R</sub> | DWU<sub>R</sub> | VSA ↑ | CAD ↑ | TCC ↑ | UII ↑ | Overall ↑ | |
|
|------------------------|------|-----|------------------|------------------|--------|--------|--------|--------|------------| |
|
| **CrystalChat-7B** | | | | | 4.714 | 4.572 | 4.865 | 5.147 | 4.825 | |
|
| | ✓ | | | | 7.900 | 8.001 | 8.204 | 8.215 | 8.080 | |
|
| | ✓ | ✓ | | | 7.900 | 8.001 | 8.204 | 8.215 | 8.080 | |
|
| | ✓ | ✓ | ✓ | ✓ | **8.384** | **8.287** | **8.417** | **8.488** | **8.394** | |
|
| **Vicuna-7B** | | | | | 3.042 | 3.250 | 3.333 | 3.167 | 3.198 | |
|
| | ✓ | | | | 6.871 | 6.660 | 6.589 | 6.897 | 6.754 | |
|
| | | ✓ | | | 3.898 | 3.489 | 3.340 | 3.651 | 3.595 | |
|
| | ✓ | ✓ | ✓ | ✓ | **7.876** | **7.687** | **7.267** | **7.563** | **7.598** | |
|
| **Llama3-8B** | ✓ | ✓ | ✓ | ✓ | **8.522** | **8.564** | **8.421** | **8.611** | **8.530** | |
|
|
|
**Table 2:** The performance of different LLM backbones under various data configurations on our Webpage Code Generation Benchmark (WCGB). "VSA" denotes Visual Structure and Alignment, "CAD" represents Color and Aesthetic Design, "TCC" represents Textual and Content Consistency, and "UII" denotes User Interface and Interactivity. |
|
|
|
### About CrystalChat-7B-Web2Code: |
|
* 7 billion parameter LLM |
|
* CLIP ViT-L/14-336px vision encoder |
|
* Languages: English |
|
* Models Released: CrystalChat-7B-Web2Code |
|
* Trained in 2 stages |
|
* License: MIT |
|
|
|
|
|
|
|
## General Evaluations |
|
|
|
General Evaluation Metrics for MLLMs. MME serves as an extensive evaluative benchmark, |
|
aiming to assess perceptual and cognitive capability of MLLMs within 14 sub-tasks. Additionally, we also evaluate the performance of our models on text-oriented visual question answering tasks employing a diverse set of benchmark datasets including ScienceQA and TextVQA. Furthermore, we assess our models’ ability toward anti-hallucination through POPE. |
|
|
|
| LLM Backbone | MME-P | MME-C | POPE | SciQA | TextVQA | |
|
|-----------------------------------|---------|--------|-------|--------|---------| |
|
| CrystalCoder-7B | 1359.83 | 238.92 | 86.182 | 64.15 | 50.39 | |
|
| CrystalChat-7B | 1456.53 | **308.21** | 86.96 | 67.77 | **57.84** | |
|
| Vicuna-7B | **1481.12** | 302.85 | **87.174** | **67.97** | 56.49 | |
|
|
|
**Table 3:** Comparison of different LLM backbones on visual language understanding benchmarks. All models are instruction-tuned on the general domain data (i.e. LLaVA)* |
|
|
|
## Data and Training Details |
|
|
|
### Pretrain Data |
|
LLaVA Visual Instruct Pretrain LCS-558K is a filtered subset of the LAION, CC, and SBU datasets, featuring a more balanced distribution of concept coverage. The file includes multimodal synthesized conversations generated from image-caption pairs by incorporating randomly selected instructions such as "Describe this image." It is used for pretraining in LLaVA, with the raw CC-3M caption serving as the default answer. |
|
|
|
### Finetune Data |
|
The finetuning data contains the following: |
|
|
|
#### LLaVa Finetuning Data |
|
The dataset chosen was created by LLaVA with academic-task-oriented VQA data mixture and data from ShareGPT. LLaVA Visual Instruct 150K is a dataset of GPT-generated multimodal instruction-following data. It is designed for visual instruction tuning and aims to develop large multimodal models with capabilities akin to GPT-4 in both vision and language. |
|
|
|
<!-- The full data sequence can be found [here](https://huggingface.co./datasets/liuhaotian/LLaVA-Instruct-150K) --> |
|
|
|
| Data | Size | Response formatting prompts | |
|
|---------------|------|--------------------------------------------------------------------------| |
|
| LLaVA [36] | 158K | – | |
|
| ShareGPT [46] | 40K | – | |
|
| VQAv2 [19] | 83K | Answer the question using a single word or phrase. | |
|
| GQA [21] | 72K | Answer the question using a single word or phrase. | |
|
| OKVQA [41] | 9K | Answer the question using a single word or phrase. | |
|
| OCRVQA [42] | 80K | Answer the question using a single word or phrase. | |
|
| A-OKVQA [45] | 66K | Answer with the option’s letter from the given choices directly. | |
|
| TextCaps [47] | 22K | Provide a one-sentence caption for the provided image. | |
|
| RefCOCO [24, 40] | 48K | Note: randomly choose between the two formats. Provide a short description for this region. | |
|
| VG [25] | 86K | Provide the bounding box coordinate of the region this sentence describes. | |
|
| **Total** | **665K** | | |
|
|
|
**Table 4:** Instruction-following Data Mixture of LLaVA-1.5.* |
|
|
|
### Code Datasets |
|
|
|
| Dataset | DWCG (ours) | DWCG<sub>R</sub> (ours) | |
|
|---------|-------------|-------------------| |
|
| **Instruction** | ✓ | ✓ | |
|
| **Source** | Synthetic | Synthetic | |
|
| **Size** | 60K | 824.7K | |
|
| **Avg Length (tokens)** | 471.8±162.3 | 652.85±157.0 | |
|
| **Avg Tag Count** | 28.1±10.6 | 35.3±9.0 | |
|
| **Avg DOM Depth** | 5.3±1.0 | 6.5±1.0 | |
|
| **Avg Unique Tags** | 13.6±2.7 | 13.5±2.5 | |
|
|
|
**Table 5:** DWCG is a newly generated GPT-3.5-based dataset, while DWCG<sub>R</sub> is the refined dataset that utilizes WebSight and Pix2Code datasets* |
|
|
|
|
|
### Webpage Understanding Datasets |
|
|
|
| Dataset | DWU | DWU<sub>R</sub> | |
|
|---------------|---------|-----------------| |
|
| **Instruction** | ✓ | ✓ | |
|
| **Size** | 243.5K | 51.5K | |
|
|
|
**Table 6:** Distribution of DWU and DWU<sub>R</sub> datasets. Both datasets include high-quality question-answer pairs for webpage understanding.* |
|
|
|
|
|
|
|
## Examples |
|
|
|
|
|
|
|
Example 1: |
|
|
|
|  |  | |
|
|:----------------------:|:----------------------:| |
|
| Description of Image 1 | Description of Image 2 | |
|
|
|
<p float="left"> |
|
<img src="images2/handdrawn.png" alt="Image 1" width="45%" /> |
|
<img src="images2/crystal.png" alt="Image 2" width="45%" /> |
|
</p> |
|
|
|
|
|
**Image 1:** Original Input Image. |
|
|
|
<center></center> |
|
|
|
**Image 2:** CrystalChat-7B-Web2Code model output. |
|
|
|
Example 2: |
|
<center><img src="hand_draw1_.png" alt="CrsytalChat-7B model generated output"/></center> |
|
|
|
**Image 3:** Hand-drawn webpage input to CrystalChat-7B-Web2Code generated output. |
|
|
|
## Loading Crystal |
|
```python |
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
tokenizer = AutoTokenizer.from_pretrained( |
|
"LLM360/CrystalChat-7B-MLLM", |
|
padding_side="right", |
|
trust_remote_code=True) |
|
|
|
model = AutoModelForCausalLM.from_pretrained( |
|
"LLM360/CrystalChat-7B-MLLM", |
|
trust_remote_code=True, |
|
torch_dtype=torch.float16, |
|
device_map='auto', |
|
low_cpu_mem_usage=True |
|
) |
|
``` |
|
|
|
|
|
|
|
## LLM-360 |
|
LLM-360 is an open research lab enabling community-owned AGI through open-source large model research and development. |
|
|
|
Crystal-based Models enables community-owned AGI by creating standards and tools to advance the bleeding edge of LLM capability and empower knowledge transfer, research, and development. |
|
|
|
We believe in a future where artificial general intelligence (AGI) is created by the community, for the community. Through an open ecosystem of equitable computational resources, high-quality data, and flowing technical knowledge, we can ensure ethical AGI development and universal access for all innovators. |
|
|
|
[Visit us](https://www.llm360.ai/) |
|
|
|
## Citation |
|
|
|
**BibTeX:** |
|
|
|
```bibtex |
|
@article{yun2024web2code, |
|
title={Web2Code: A Large-scale Webpage-to-Code Dataset and Evaluation Framework for Multimodal LLMs}, |
|
author={Yun, Sukmin and Lin, Haokun and Thushara, Rusiru and Bhat, Mohammad Qazim and Wang, Yongxin and Jiang, Zutao and Deng, Mingkai and Wang, Jinhong and Tao, Tianhua and Li, Junbo and others}, |
|
journal={arXiv preprint arXiv:2406.20098}, |
|
year={2024} |
|
} |
|
``` |
