AdaptLLM commited on
Commit
d417aba
·
verified ·
1 Parent(s): 52e2781

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +112 -0
README.md ADDED
@@ -0,0 +1,112 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ configs:
3
+ - config_name: FPB
4
+ data_files:
5
+ - split: train
6
+ path: train.csv
7
+ - split: test
8
+ path: test.csv
9
+ task_categories:
10
+ - text-classification
11
+ - question-answering
12
+ - zero-shot-classification
13
+ language:
14
+ - en
15
+ tags:
16
+ - finance
17
+ ---
18
+
19
+ # Domain Adaptation of Large Language Models
20
+ This repo contains the **FPB dataset** used in our **ICLR 2024** paper [Adapting Large Language Models via Reading Comprehension](https://huggingface.co/papers/2309.09530).
21
+
22
+ We explore **continued pre-training on domain-specific corpora** for large language models. While this approach enriches LLMs with domain knowledge, it significantly hurts their prompting ability for question answering. Inspired by human learning via reading comprehension, we propose a simple method to **transform large-scale pre-training corpora into reading comprehension texts**, consistently improving prompting performance across tasks in biomedicine, finance, and law domains. **Our 7B model competes with much larger domain-specific models like BloombergGPT-50B**.
23
+
24
+ ### 🤗 We are currently working hard on developing models across different domains, scales and architectures! Please stay tuned! 🤗
25
+
26
+ **************************** **Updates** ****************************
27
+ * 2024/4/2: Released the raw data splits (train and test) of all the evaluation datasets
28
+ * 2024/1/16: 🎉 Our [research paper](https://huggingface.co/papers/2309.09530) has been accepted by ICLR 2024!!!🎉
29
+ * 2023/12/19: Released our [13B base models](https://huggingface.co/AdaptLLM/law-LLM-13B) developed from LLaMA-1-13B.
30
+ * 2023/12/8: Released our [chat models](https://huggingface.co/AdaptLLM/law-chat) developed from LLaMA-2-Chat-7B.
31
+ * 2023/9/18: Released our [paper](https://huggingface.co/papers/2309.09530), [code](https://github.com/microsoft/LMOps), [data](https://huggingface.co/datasets/AdaptLLM/law-tasks), and [base models](https://huggingface.co/AdaptLLM/law-LLM) developed from LLaMA-1-7B.
32
+
33
+
34
+ ## Domain-Specific LLaMA-1
35
+ ### LLaMA-1-7B
36
+ In our paper, we develop three domain-specific models from LLaMA-1-7B, which are also available in Huggingface: [Biomedicine-LLM](https://huggingface.co/AdaptLLM/medicine-LLM), [Finance-LLM](https://huggingface.co/AdaptLLM/finance-LLM) and [Law-LLM](https://huggingface.co/AdaptLLM/law-LLM), the performances of our AdaptLLM compared to other domain-specific LLMs are:
37
+
38
+ <p align='center'>
39
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/650801ced5578ef7e20b33d4/6efPwitFgy-pLTzvccdcP.png" width="700">
40
+ </p>
41
+
42
+ ### LLaMA-1-13B
43
+ Moreover, we scale up our base model to LLaMA-1-13B to see if **our method is similarly effective for larger-scale models**, and the results are consistently positive too: [Biomedicine-LLM-13B](https://huggingface.co/AdaptLLM/medicine-LLM-13B), [Finance-LLM-13B](https://huggingface.co/AdaptLLM/finance-LLM-13B) and [Law-LLM-13B](https://huggingface.co/AdaptLLM/law-LLM-13B).
44
+
45
+ ## Domain-Specific LLaMA-2-Chat
46
+ Our method is also effective for aligned models! LLaMA-2-Chat requires a [specific data format](https://huggingface.co/blog/llama2#how-to-prompt-llama-2), and our **reading comprehension can perfectly fit the data format** by transforming the reading comprehension into a multi-turn conversation. We have also open-sourced chat models in different domains: [Biomedicine-Chat](https://huggingface.co/AdaptLLM/medicine-chat), [Finance-Chat](https://huggingface.co/AdaptLLM/finance-chat) and [Law-Chat](https://huggingface.co/AdaptLLM/law-chat)
47
+
48
+ ## Domain-Specific Tasks
49
+
50
+ ### Pre-templatized/Formatted Testing Splits
51
+ To easily reproduce our prompting results, we have uploaded the filled-in zero/few-shot input instructions and output completions of the test each domain-specific task: [biomedicine-tasks](https://huggingface.co/datasets/AdaptLLM/medicine-tasks), [finance-tasks](https://huggingface.co/datasets/AdaptLLM/finance-tasks), and [law-tasks](https://huggingface.co/datasets/AdaptLLM/law-tasks).
52
+
53
+ **Note:** those filled-in instructions are specifically tailored for models before alignment and do NOT fit for the specific data format required for chat models.
54
+
55
+ ### Raw Datasets
56
+ We have also uploaded the raw training and testing splits, for facilitating fine-tuning or other usages:
57
+ - [ChemProt](https://huggingface.co/datasets/AdaptLLM/ChemProt)
58
+ - [RCT](https://huggingface.co/datasets/AdaptLLM/RCT)
59
+ - [ConvFinQA](https://huggingface.co/datasets/AdaptLLM/ConvFinQA)
60
+ - [FiQA_SA](https://huggingface.co/datasets/AdaptLLM/FiQA_SA)
61
+ - [Headline](https://huggingface.co/datasets/AdaptLLM/Headline)
62
+ - [NER](https://huggingface.co/datasets/AdaptLLM/NER)
63
+ - [FPB](https://huggingface.co/datasets/AdaptLLM/FPB)
64
+
65
+ The other datasets used in our paper have already been available in huggingface, so you can directly load them with the following code:
66
+ ```python
67
+ from datasets import load_dataset
68
+
69
+ # MQP:
70
+ dataset = load_dataset('medical_questions_pairs')
71
+ # PubmedQA:
72
+ dataset = load_dataset('bigbio/pubmed_qa')
73
+ # USMLE:
74
+ dataset=load_dataset('GBaker/MedQA-USMLE-4-options', cache_dir=cache_dir)
75
+ # SCOTUS
76
+ dataset = load_dataset("lex_glue", 'scotus')
77
+ # CaseHOLD
78
+ dataset = load_dataset("lex_glue", 'case_hold')
79
+ # UNFAIR-ToS
80
+ dataset = load_dataset("lex_glue", 'unfair_tos')
81
+ ```
82
+
83
+ ## Citation
84
+ If you find our work helpful, please cite us:
85
+ ```bibtex
86
+ @inproceedings{
87
+ cheng2024adapting,
88
+ title={Adapting Large Language Models via Reading Comprehension},
89
+ author={Daixuan Cheng and Shaohan Huang and Furu Wei},
90
+ booktitle={The Twelfth International Conference on Learning Representations},
91
+ year={2024},
92
+ url={https://openreview.net/forum?id=y886UXPEZ0}
93
+ }
94
+ ```
95
+
96
+ and the original dataset:
97
+ ```bibtex
98
+ @inproceedings{FiQA-SA,
99
+ author = {Macedo Maia and
100
+ Siegfried Handschuh and
101
+ Andr{\'{e}} Freitas and
102
+ Brian Davis and
103
+ Ross McDermott and
104
+ Manel Zarrouk and
105
+ Alexandra Balahur},
106
+ title = {WWW'18 Open Challenge: Financial Opinion Mining and Question Answering},
107
+ booktitle = {{WWW} (Companion Volume)},
108
+ pages = {1941--1942},
109
+ publisher = {{ACM}},
110
+ year = {2018}
111
+ }
112
+ ```