Hugging Face's logo

QuantFactory
/
HTML-Pruner-Llama-1B-GGUF

Transformers
GGUF
English
conversational
Model card Files
xet
 Community
munish0838 commited on
Commit
f07ec4a
·
verified ·
1 Parent(s): 323bd47

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +205 -0
README.md ADDED
@@ -0,0 +1,205 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+
4
+ language:
5
+ - en
6
+ library_name: transformers
7
+ base_model: meta-llama/Llama-3.2-1B
8
+ license: apache-2.0
9
+
10
+ ---
11
+
12
+ [![QuantFactory Banner](https://lh7-rt.googleusercontent.com/docsz/AD_4nXeiuCm7c8lEwEJuRey9kiVZsRn2W-b4pWlu3-X534V3YmVuVc2ZL-NXg2RkzSOOS2JXGHutDuyyNAUtdJI65jGTo8jT9Y99tMi4H4MqL44Uc5QKG77B0d6-JfIkZHFaUA71-RtjyYZWVIhqsNZcx8-OMaA?key=xt3VSDoCbmTY7o-cwwOFwQ)](https://hf.co/QuantFactory)
13
+
14
+
15
+ # QuantFactory/HTML-Pruner-Llama-1B-GGUF
16
+ This is quantized version of [zstanjj/HTML-Pruner-Llama-1B](https://huggingface.co/zstanjj/HTML-Pruner-Llama-1B) created using llama.cpp
17
+
18
+ # Original Model Card
19
+
20
+
21
+
22
+
23
+ ## ✨ Latest News
24
+
25
+ - [11/06/2024]: Our paper is available on arXiv. You can access it [here](https://arxiv.org/abs/2411.02959).
26
+ - [11/05/2024]: The open-source toolkit and models are released. You can apply HtmlRAG in your own RAG systems now.
27
+
28
+
29
+ ## Model Information
30
+
31
+ <p align="left">
32
+ • 📝 <a href="https://arxiv.org/abs/2411.02959" target="_blank">Paper</a> • 🤗 <a href="https://huggingface.co/zstanjj/SlimPLM-Query-Rewriting/" target="_blank">Hugging Face</a> • 🧩 <a href="https://github.com/plageon/SlimPLM" target="_blank">Github</a>
33
+ </p>
34
+
35
+ We propose HtmlRAG, which uses HTML instead of plain text as the format of external knowledge in RAG systems. To tackle the long context brought by HTML, we propose **Lossless HTML Cleaning** and **Two-Step Block-Tree-Based HTML Pruning**.
36
+
37
+ - **Lossless HTML Cleaning**: This cleaning process just removes totally irrelevant contents and compress redundant structures, retaining all semantic information in the original HTML. The compressed HTML of lossless HTML cleaning is suitable for RAG systems that have long-context LLMs and are not willing to loss any information before generation.
38
+
39
+ - **Two-Step Block-Tree-Based HTML Pruning**: The block-tree-based HTML pruning consists of two steps, both of which are conducted on the block tree structure. The first pruning step uses a embedding model to calculate scores for blocks, while the second step uses a path generative model. The first step processes the result of lossless HTML cleaning, while the second step processes the result of the first pruning step.
40
+
41
+
42
+ 🌹 If you use this model, please star our **[GitHub repository](https://github.com/plageon/HTMLRAG)** to support us. Your star means a lot!
43
+
44
+ ## 📦 Installation
45
+
46
+ Install the package using pip:
47
+ ```bash
48
+ pip install htmlrag
49
+ ```
50
+ Or install the package from source:
51
+ ```bash
52
+ pip install -e .
53
+ ```
54
+
55
+ ---
56
+
57
+ ## 📖 User Guide
58
+
59
+ ### 🧹 HTML Cleaning
60
+
61
+ ```python
62
+ from htmlrag import clean_html
63
+
64
+ question = "When was the bellagio in las vegas built?"
65
+ html = """
66
+ <html>
67
+ <head>
68
+ <title>When was the bellagio in las vegas built?</title>
69
+ </head>
70
+ <body>
71
+ <p class="class0">The Bellagio is a luxury hotel and casino located on the Las Vegas Strip in Paradise, Nevada. It was built in 1998.</p>
72
+ </body>
73
+ <div>
74
+ <div>
75
+ <p>Some other text</p>
76
+ <p>Some other text</p>
77
+ </div>
78
+ </div>
79
+ <p class="class1"></p>
80
+ <!-- Some comment -->
81
+ <script type="text/javascript">
82
+ document.write("Hello World!");
83
+ </script>
84
+ </html>
85
+ """
86
+
87
+ simplified_html = clean_html(html)
88
+ print(simplified_html)
89
+
90
+ # <html>
91
+ # <title>When was the bellagio in las vegas built?</title>
92
+ # <p>The Bellagio is a luxury hotel and casino located on the Las Vegas Strip in Paradise, Nevada. It was built in 1998.</p>
93
+ # <div>
94
+ # <p>Some other text</p>
95
+ # <p>Some other text</p>
96
+ # </div>
97
+ # </html>
98
+ ```
99
+
100
+
101
+ ### 🌲 Build Block Tree
102
+
103
+ ```python
104
+ from htmlrag import build_block_tree
105
+
106
+ block_tree, simplified_html = build_block_tree(simplified_html, max_node_words=10)
107
+ for block in block_tree:
108
+ print("Block Content: ", block[0])
109
+ print("Block Path: ", block[1])
110
+ print("Is Leaf: ", block[2])
111
+ print("")
112
+
113
+ # Block Content: <title>When was the bellagio in las vegas built?</title>
114
+ # Block Path: ['html', 'title']
115
+ # Is Leaf: True
116
+ #
117
+ # Block Content: <div>
118
+ # <p>Some other text</p>
119
+ # <p>Some other text</p>
120
+ # </div>
121
+ # Block Path: ['html', 'div']
122
+ # Is Leaf: True
123
+ #
124
+ # Block Content: <p>The Bellagio is a luxury hotel and casino located on the Las Vegas Strip in Paradise, Nevada. It was built in 1998.</p>
125
+ # Block Path: ['html', 'p']
126
+ # Is Leaf: True
127
+ ```
128
+
129
+ ### ✂️ Prune HTML Blocks with Embedding Model
130
+
131
+ ```python
132
+ from htmlrag import EmbedHTMLPruner
133
+
134
+ embed_html_pruner = EmbedHTMLPruner(embed_model="bm25")
135
+ block_rankings = embed_html_pruner.calculate_block_rankings(question, simplified_html, block_tree)
136
+ print(block_rankings)
137
+
138
+ # [0, 2, 1]
139
+
140
+ from transformers import AutoTokenizer
141
+
142
+ chat_tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.1-70B-Instruct")
143
+
144
+ max_context_window = 60
145
+ pruned_html = embed_html_pruner.prune_HTML(simplified_html, block_tree, block_rankings, chat_tokenizer, max_context_window)
146
+ print(pruned_html)
147
+
148
+ # <html>
149
+ # <title>When was the bellagio in las vegas built?</title>
150
+ # <p>The Bellagio is a luxury hotel and casino located on the Las Vegas Strip in Paradise, Nevada. It was built in 1998.</p>
151
+ # </html>
152
+ ```
153
+
154
+
155
+ ### ✂️ Prune HTML Blocks with Generative Model
156
+
157
+ ```python
158
+ from htmlrag import GenHTMLPruner
159
+
160
+ ckpt_path = "zstanjj/HTML-Pruner-Llama-1B"
161
+ gen_embed_pruner = GenHTMLPruner(gen_model=ckpt_path, max_node_words=10)
162
+ block_rankings = gen_embed_pruner.calculate_block_rankings(question, pruned_html)
163
+ print(block_rankings)
164
+
165
+ # [1, 0]
166
+
167
+ max_context_window = 32
168
+ pruned_html = gen_embed_pruner.prune_HTML(pruned_html, block_tree, block_rankings, chat_tokenizer, max_context_window)
169
+ print(pruned_html)
170
+
171
+ # <p>The Bellagio is a luxury hotel and casino located on the Las Vegas Strip in Paradise, Nevada. It was built in 1998.</p>
172
+ ```
173
+
174
+ ## Results
175
+
176
+ - **Results for [HTML-Pruner-Phi-3.8B](https://huggingface.co/zstanjj/HTML-Pruner-Phi-3.8B) and [HTML-Pruner-Llama-1B](https://huggingface.co/zstanjj/HTML-Pruner-Llama-1B) with Llama-3.1-70B-Instruct as chat model**.
177
+
178
+ | Dataset | ASQA | HotpotQA | NQ | TriviaQA | MuSiQue | ELI5 |
179
+ |------------------|-----------|-----------|-----------|-----------|-----------|-----------|
180
+ | Metrics | EM | EM | EM | EM | EM | ROUGE-L |
181
+ | BM25 | 49.50 | 38.25 | 47.00 | 88.00 | 9.50 | 16.15 |
182
+ | BGE | 68.00 | 41.75 | 59.50 | 93.00 | 12.50 | 16.20 |
183
+ | E5-Mistral | 63.00 | 36.75 | 59.50 | 90.75 | 11.00 | 16.17 |
184
+ | LongLLMLingua | 62.50 | 45.00 | 56.75 | 92.50 | 10.25 | 15.84 |
185
+ | JinaAI Reader | 55.25 | 34.25 | 48.25 | 90.00 | 9.25 | 16.06 |
186
+ | HtmlRAG-Phi-3.8B | **68.50** | **46.25** | 60.50 | **93.50** | **13.25** | **16.33** |
187
+ | HtmlRAG-Llama-1B | 66.50 | 45.00 | **60.75** | 93.00 | 10.00 | 16.25 |
188
+
189
+ ---
190
+
191
+ ## 📜 Citation
192
+
193
+ ```bibtex
194
+ @misc{tan2024htmlraghtmlbetterplain,
195
+ title={HtmlRAG: HTML is Better Than Plain Text for Modeling Retrieved Knowledge in RAG Systems},
196
+ author={Jiejun Tan and Zhicheng Dou and Wen Wang and Mang Wang and Weipeng Chen and Ji-Rong Wen},
197
+ year={2024},
198
+ eprint={2411.02959},
199
+ archivePrefix={arXiv},
200
+ primaryClass={cs.IR},
201
+ url={https://arxiv.org/abs/2411.02959},
202
+ }
203
+ ```
204
+
205
+