pybeebee nielsr HF Staff commited on
Commit
ad5573b
·
verified ·
1 Parent(s): 57ea0fa

Add library_name and pipeline tag (#1)

Browse files

- Add library_name and pipeline tag (212efe5db33162d342619c2ffaf3faa082c1aaa2)


Co-authored-by: Niels Rogge <[email protected]>

Files changed (1) hide show
  1. README.md +15 -7
README.md CHANGED
@@ -1,23 +1,23 @@
1
  ---
2
- license: apache-2.0
 
3
  datasets:
4
  - yale-nlp/MDCure-72k
5
  language:
6
  - en
7
- base_model:
8
- - google/flan-t5-base
9
  tags:
10
  - multi-document
11
  - long-context
12
  - Long Context
 
 
13
  ---
14
 
15
  # MDCure-FlanT5-Large
16
 
17
-
18
  [📄 Paper](https://arxiv.org/pdf/2410.23463) | [🤗 HF Collection](https://huggingface.co/collections/yale-nlp/mdcure-6724914875e87f41e5445395) | [⚙️ GitHub Repo](https://github.com/yale-nlp/MDCure)
19
 
20
-
21
  ## Introduction
22
 
23
  **MDCure** is an effective and scalable procedure for generating high-quality multi-document (MD) instruction tuning data to improve MD capabilities of LLMs. Using MDCure, we construct a suite of MD instruction datasets complementary to collections such as [FLAN](https://github.com/google-research/FLAN) and fine-tune a variety of already instruction-tuned LLMs from the FlanT5, Qwen2, and LLAMA3.1 model families, up to 70B parameters in size. We additionally introduce **MDCureRM**, an evaluator model specifically designed for the MD setting to filter and select high-quality MD instruction data in a cost-effective, RM-as-a-judge fashion. Extensive evaluations on a wide range of MD and long-context benchmarks spanning various tasks show MDCure consistently improves performance over pre-trained baselines and over corresponding base models by up to 75.5%.
@@ -41,7 +41,9 @@ We recommend using the latest version of HF Transformers, or any `transformers>4
41
 
42
  ## Quickstart
43
 
44
- Below we provide a code snippet demonstrating how to load the tokenizer and model and generate content in response to an input context concerning multiple source documents and a related question or instruction. We strongly recommend to separate the texts and/or instruction using `\n\n` or `<doc-sep>` to maintain consistency with the format of the data used during training.
 
 
45
 
46
  ```python
47
  model = AutoModelForSeq2SeqLM.from_pretrained("yale-nlp/MDCure-FlanT5-Large", device_map='auto',torch_dtype="auto",)
@@ -50,7 +52,13 @@ tokenizer = AutoTokenizer.from_pretrained("yale-nlp/MDCure-FlanT5-Large")
50
  source_text_1 = ...
51
  source_text_2 = ...
52
  source_text_3 = ...
53
- input_text = f"{source_text_1}\n\n{source_text_2}\n\n{source_text_3}\n\nWhat happened in CHAMPAIGN regarding Lovie Smith and the 2019 defense improvements? Respond with 1-2 sentences."
 
 
 
 
 
 
54
 
55
  input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to(model.device)
56
  outputs = model.generate(input_ids)
 
1
  ---
2
+ base_model:
3
+ - google/flan-t5-base
4
  datasets:
5
  - yale-nlp/MDCure-72k
6
  language:
7
  - en
8
+ license: apache-2.0
 
9
  tags:
10
  - multi-document
11
  - long-context
12
  - Long Context
13
+ library_name: transformers
14
+ pipeline_tag: summarization
15
  ---
16
 
17
  # MDCure-FlanT5-Large
18
 
 
19
  [📄 Paper](https://arxiv.org/pdf/2410.23463) | [🤗 HF Collection](https://huggingface.co/collections/yale-nlp/mdcure-6724914875e87f41e5445395) | [⚙️ GitHub Repo](https://github.com/yale-nlp/MDCure)
20
 
 
21
  ## Introduction
22
 
23
  **MDCure** is an effective and scalable procedure for generating high-quality multi-document (MD) instruction tuning data to improve MD capabilities of LLMs. Using MDCure, we construct a suite of MD instruction datasets complementary to collections such as [FLAN](https://github.com/google-research/FLAN) and fine-tune a variety of already instruction-tuned LLMs from the FlanT5, Qwen2, and LLAMA3.1 model families, up to 70B parameters in size. We additionally introduce **MDCureRM**, an evaluator model specifically designed for the MD setting to filter and select high-quality MD instruction data in a cost-effective, RM-as-a-judge fashion. Extensive evaluations on a wide range of MD and long-context benchmarks spanning various tasks show MDCure consistently improves performance over pre-trained baselines and over corresponding base models by up to 75.5%.
 
41
 
42
  ## Quickstart
43
 
44
+ Below we provide a code snippet demonstrating how to load the tokenizer and model and generate content in response to an input context concerning multiple source documents and a related question or instruction. We strongly recommend to separate the texts and/or instruction using `
45
+
46
+ ` or `<doc-sep>` to maintain consistency with the format of the data used during training.
47
 
48
  ```python
49
  model = AutoModelForSeq2SeqLM.from_pretrained("yale-nlp/MDCure-FlanT5-Large", device_map='auto',torch_dtype="auto",)
 
52
  source_text_1 = ...
53
  source_text_2 = ...
54
  source_text_3 = ...
55
+ input_text = f"{source_text_1}
56
+
57
+ {source_text_2}
58
+
59
+ {source_text_3}
60
+
61
+ What happened in CHAMPAIGN regarding Lovie Smith and the 2019 defense improvements? Respond with 1-2 sentences."
62
 
63
  input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to(model.device)
64
  outputs = model.generate(input_ids)