File size: 8,210 Bytes
fcc571f 496eed3 fcc571f 496eed3 b2a53d0 eac36e0 455b82a b2a53d0 455b82a b2a53d0 455b82a 496eed3 455b82a 496eed3 783ac46 496eed3 783ac46 496eed3 |
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 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 |
---
license: cc-by-nc-4.0
language:
- en
---
# Jellyfish-8B
<!-- Provide a quick summary of what the model is/does. -->
<!--
<img src="https://i.imgur.com/d8Bl04i.png" alt="PicToModel" width="330"/>
-->
<img src="https://i.imgur.com/E1vqCIw.png" alt="PicToModel" width="330"/>
## Model Details
Jellyfish-8B is a large language model equipped with 8 billion parameters.
We fine-tuned the [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) model using the datasets pertinent to data preprocessing tasks.
The training data include two parts:
* Jellyfish-13B training data
* GPT4 generated reasoning data for data preprocessing tasks.
<!-- Jellyfish-7B vs GPT-3.5-turbo wining rate by GPT4 evaluation is 56.36%. -->
More details about the model can be found in the [Jellyfish paper](https://arxiv.org/abs/2312.01678).
- **Developed by:** Haochen Zhang, Yuyang Dong, Chuan Xiao, Masafumi Oyamada
- **Contact: [email protected]**
- **Funded by:** NEC Corporation, Osaka University
- **Language(s) (NLP):** English
- **License:** Non-Commercial Creative Commons license (CC BY-NC-4.0)
- **Finetuned from model:** [Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct)
## Citation
If you find our work useful, please give us credit by citing:
```
@article{zhang2023jellyfish,
title={Jellyfish: A Large Language Model for Data Preprocessing},
author={Zhang, Haochen and Dong, Yuyang and Xiao, Chuan and Oyamada, Masafumi},
journal={arXiv preprint arXiv:2312.01678},
year={2023}
}
```
## Performance on seen tasks
| Task | Type | Dataset | Non-LLM SoTA<sup>1</sup> | GPT-3.5<sup>2</sup> | GPT-4<sup>2</sup> | GPT-4o | Jellyfish-13B | Jellyfish-7B | Jellyfish-8B |
| ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
| Entity Matching | Seen | Fodors-Zagats | 100 | 100 | 100 | | 100 | 100 | 92.68 |
| Entity Matching | Seen | Beer | 94.37| 96.30 | 100 | | 96.77 | 96.55| 96.30 |
| Entity Matching | Seen | iTunes-Amazon | 97.06| 96.43 | 100 | | 98.11 | 96.30| 92.00 |
| Entity Matching | Seen | DBLP-ACM | 98.99| 96.99 | 97.44 | | 98.98 | 98.88| 98.76 |
| Entity Matching | Seen | DBLP-GoogleScholar | 95.60| 76.12 | 91.87 | | 98.51 | 95.15| 93.20 |
| Entity Matching | Seen | Amazon-Google | 75.58| 66.53 | 74.21 | 70.91 | 81.34 | 80.83 | 74.49 |
| Entity Matching | Unseen | Walmart-Amazon | 86.76| 86.17 | 90.27 | | 89.42 | 85.64 | 89.97 |
| Entity Matching | Unseen | Abt-Buy | 89.33 | -- | 92.77 | | 89.58 | 82.38 | 92.54 |
| Data Imputation | Seen | Restaurant | 77.20| 94.19 | 97.67 | | 94.19 | 88.37 | 87.21 |
| Data Imputation | Seen | Buy | 96.50| 98.46 | 100 | | 100 | 96.62 | 92.31 |
| Data Imputation | Unseen | Filpkart | 68.00 | -- | 89.94 | | 81.68 | 79.44| 90.17 |
| Data Imputation | Unseen | Phone | 86.70| -- | 90.79 | | 87.21 | 85.00| 83.92 |
| Error Detection | Seen | Hosptial | 94.40| 90.74 | 90.74 | 44.76 | 95.59 | 96.27 | 80.72|
| Error Detection | Seen | Adult | 99.10| 92.01 | 92.01 | 83.58 | 99.33 | 91.96 | 81.72|
| Error Detection | Unseen | Flights | 81.00 | -- | 83.48 | 66.01 | 82.52 | 66.92 | 75.18 |
| Error Detection | Unseen | Rayyan | 79.00| -- | 81.95 | 68.53 | 90.65 | 69.82 | 91.54 |
| Schema Matching | Seen | Sythea | 38.50| 57.14 | 66.67 | 6.56 | 36.36 | 44.44 | 27.27 |
| Schema Matching | Seen | MIMIC | 20.00| -- | 40.00 | 29.41 | 40.00 | 40.00 | 34.04|
| Schema Matching | Unseen | CMS | 50.00| -- | 19.35 | 22.22 | 59.29 | 13.79 | 56.72|
_For GPT-3.5 and GPT-4, we used the few-shot approach on all datasets. However, for Jellyfish-13B and Jellyfish-Interpreter, the few-shot approach is disabled on seen datasets and enabled on unseen datasets._
_Accuracy as the metric for data imputation and the F1 score for other tasks._
| Task | Type | Dataset | Best of non-LLM | GPT-3 | GPT-3.5 | GPT-4 | GPT-4o | Table-GPT | Jellyfish-7B | Jellyfish-8B | Jellyfish-13B |
|------|--------|----------------------|-----------------|-------|---------|-------|--------|-----------|--------------|--------------|---------------|
| Error Detection | Seen | Adult | *99.10 | 99.10 | 92.01 | 92.01 | 83.58 | -- | 77.40 | 73.74 | **99.33 |
| | | Hospital | 94.40 | **97.80 | 90.74 | 90.74 | 44.76 | -- | 94.51 | 93.40 | *95.59 |
| | Unseen | Flights | 81.00 | -- | -- | **83.48 | 66.01 | -- | 69.15 | 66.21 | *82.52 |
| | | Rayyan | 79.00 | -- | -- | *81.95 | 68.53 | -- | 75.07 | 81.06 | **90.65 |
| Data Imputation | Seen | Buy | 96.50 | 98.50 | 98.46 | **100 | **100 | -- | 98.46 | 98.46 | **100 |
| | | Restaurant | 77.20 | 88.40 | *94.19 | **97.67 | 90.70 | -- | 89.53 | 87.21 | 89.53 |
| | Unseen | Flipkart | 68.00 | -- | -- | **89.94 | 83.20 | -- | 87.14 | *87.48 | 81.68 |
| | | Phone | 86.70 | -- | -- | **90.79 | 86.78 | -- | 86.52 | 85.68 | *87.21 |
| Schema Matching | Seen | MIMIC-III | 20.00 | -- | -- | 40.00 | 29.41 | -- | **53.33 | *45.45 | 40.00 |
| | | Synthea | 38.50 | 45.20 | *57.14 | **66.67 | 6.56 | -- | 55.56 | 47.06 | 56.00 |
| | Unseen | CMS | *50.00 | -- | -- | 19.35 | 22.22 | -- | 42.86 | 38.10 | **59.29 |
| Entity Matching | Seen | Amazon-Google | 75.58 | 63.50 | 66.50 | 74.21 | 70.91 | 70.10 | **81.69 | *81.42 | 81.34 |
| | | Beer | 94.37 | **100 | 96.30 | **100 | 90.32 | 96.30 | **100.00 | **100.00 | 96.77 |
| | | DBLP-ACM | **98.99 | 96.60 | 96.99 | 97.44 | 95.87 | 93.80 | 98.65 | 98.77 | *98.98 |
| | | DBLP-GoogleScholar | *95.70 | 83.80 | 76.12 | 91.87 | 90.45 | 92.40 | 94.88 | 95.03 | **98.51 |
| | | Fodors-Zagats | **100 | **100 | **100 | **100 | 93.62 | **100 | **100 | **100 | **100 |
| | | iTunes-Amazon | 97.06 | *98.20 | 96.40 | **100 | 98.18 | 94.30 | 96.30 | 96.30 | 98.11 |
| | Unseen | Abt-Buy | 89.33 | -- | -- | **92.77 | 78.73 | -- | 86.06 | 88.84 | *89.58 |
| | | Walmart-Amazon | 86.89 | 87.00 | 86.17 | **90.27 | 79.19 | 82.40 | 84.91 | 85.24 | *89.42 |
| Avg | | | 80.44 | - | - | *84.17 | 72.58 | - | 82.74 | 81.55 | **86.02 |
## Performance on unseen tasks
### Column Type Annotation
| Dataset | RoBERTa (159 shots)<sup>1</sup> | GPT-3.5<sup>1</sup> | GPT-4 | Jellfish-13B| Jellyfish-7B | Jellyfish-8B |
| ---- | ---- | ---- | ---- | ---- | ----|----|
| SOTAB | 79.20 | 89.47 | 91.55 | 82.00 | 80.89 | 67.21|
_Few-shot is disabled for Jellyfish-13B._
1. Results from [Column Type Annotation using ChatGPT](https://arxiv.org/abs/2306.00745)
### Attribute Value Extraction
| Dataset |Stable Beluga 2 70B<sup>1</sup> | SOLAR 70B<sup>1</sup> | GPT-3.5<sup>1</sup> | GPT-4 <sup>1</sup>| Jellfish-13B | Jellyfish-7B| Jellyfish-8B |
| ---- | ---- | ---- | ---- | ---- | ---- | ----| ----|
| AE-110k | 52.10 | 49.20 | 61.30 | 55.50 | 58.12 | 76.85| 69.78|
| OA-Mine | 50.80 | 55.20 | 62.70 | 68.90 | 55.96 | 76.04| 78.83|
## Prompt Template
```
[INST]:
<prompt> (without the <>)
[\INST]]
```
|