Hugging Face's logo

Spaces:
awacke1
/
TorchTransformers-CV-SFT
Running

App Files Community
1
awacke1 commited on
Commit
8bd86ec
·
verified ·
1 Parent(s): b5f3dfb

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +149 -106
README.md CHANGED
@@ -11,119 +11,162 @@ license: mit
11
  short_description: Torch Transformers Diffusion SFT f. Streamlit & C. Vision
12
  ---
13
 
14
-
15
- # Integration Details
16
-
17
- 1. SFT Tiny Titans (First Listing):
18
- - Features: Causal LM and Diffusion SFT, camera snap, RAG party.
19
- - Integration: Added as "Build Titan", "Fine-Tune Titan", "Test Titan", and "Agentic RAG Party" tabs. Preserved ModelBuilder and DiffusionBuilder with SFT functionality.
20
- 2. SFT Tiny Titans (Second Listing):
21
- - Features: Enhanced Causal LM SFT with sample CSV generation, export functionality, and RAG demo.
22
- - Integration: Merged into "Build Titan" (sample CSV), "Fine-Tune Titan" (enhanced UI), "Test Titan" (export), and "Agentic RAG Party" (improved agent). Used PartyPlannerAgent from this listing for its detailed RAG output.
23
- 3. AI Vision Titans (Current):
24
- - Features: PDF snapshotting, OCR with GOT-OCR2_0, Image Gen, Line Drawings.
25
- - Integration: Added as "Download PDFs", "Test OCR", "Test Image Gen", and "Test Line Drawings" tabs. Retained async processing and gallery updates.
26
- 4. Sidebar, Session, and History:
27
- - Unified gallery shows PNGs and TXT files from all tabs.
28
- - Session state (captured_files, builder, model_loaded, processing, history) tracks all operations.
29
- - History log in sidebar records key actions (snapshots, SFT, tests).
30
- 5. Workflow:
31
- - Users can snap images or download PDFs, build/fine-tune models, test them, and run RAG demos, with all outputs saved and accessible via the gallery.
32
- 7. Verification
33
- - Run the App: streamlit run app.py
34
- 8. Check:
35
- - Camera Snap: Capture images, verify in gallery.
36
- - Download PDFs: Test with a valid PDF URL (e.g., a direct link), check snapshots.
37
- - Build/Fine-Tune Titan: Build a Causal LM or Diffusion model, fine-tune with CSV or images, save outputs.
38
- - Test Titan: Evaluate Causal LM with prompts or generate Diffusion images, check history.
39
- - Agentic RAG Party: Run NLP or CV RAG demos, verify outputs.
40
- - Test OCR/Image Gen/Line Drawings: Process images, ensure outputs save and appear in gallery.
41
- 9. Expected Logs: "Saved snapshot...", "Model loaded...", "SFT completed...", etc.
42
- 10. Notes
43
- - PDF URLs: Your provided URLs need direct PDF links (e.g., via Archive.org’s /download/ path). Adjust as needed.
44
- - Compatibility: All features use CPU defaults for broad compatibility, with CUDA fallback where available.
45
- - Session State: Persistent across tabs, ensuring workflow continuity.
46
-
47
- ## Abstract
48
- Explore AI vision with `torch`, `transformers`, and `diffusers`! Dual `st.camera_input` 📷 captures feed async OCR (Qwen2-VL, TrOCR), image gen (Stable Diffusion), and line drawings (Torch Space-inspired) on CPU. Key papers:
49
-
50
- - 🌐 **[Streamlit](https://arxiv.org/abs/2308.03892)** - Thiessen et al., 2023: UI.
51
- - 🔥 **[PyTorch](https://arxiv.org/abs/1912.01703)** - Paszke et al., 2019: Core.
52
- - 🔍 **[Qwen2-VL](https://arxiv.org/abs/2408.11039)** - Li et al., 2024: Multimodal OCR.
53
- - 🔍 **[TrOCR](https://arxiv.org/abs/2109.10282)** - Li et al., 2021: Small OCR.
54
- - 🎨 **[LDM](https://arxiv.org/abs/2112.10752)** - Rombach et al., 2022: Image gen.
55
- - 👁️ **[OpenCV](https://arxiv.org/abs/2308.11236)** - Bradski, 2000: CV tools.
56
-
57
- Run: `pip install -r requirements.txt`, `streamlit run ${app_file}`. Snap, test, innovate! ${emoji}
58
-
59
- ## Usage 🎯
60
- - 📷 **Camera Snap**: Single or burst capture (auto 10 frames) with gallery.
61
- - 🔍 **Test OCR**: `Qwen2-VL-OCR-2B` or `TrOCR-Small` extracts text, saved async.
62
- - 🎨 **Test Image Gen**: `OFA-Sys/small-stable-diffusion-v0` generates images, saved async.
63
- - ✏️ **Test Line Drawings**: OpenCV line art (Torch Space-inspired), saved async.
64
 
65
  ## Abstract
66
- Fuse `torch`, `transformers`, and `diffusers` for SFT-powered NLP and CV! Dual `st.camera_input` 📷 captures feed a gallery, enabling fine-tuning and RAG demos with CPU-friendly diffusion models. Key papers:
67
 
68
  - 🌐 **[Streamlit Framework](https://arxiv.org/abs/2308.03892)** - Thiessen et al., 2023: UI magic.
69
  - 🔥 **[PyTorch DL](https://arxiv.org/abs/1912.01703)** - Paszke et al., 2019: Torch core.
70
  - 🧠 **[Attention is All You Need](https://arxiv.org/abs/1706.03762)** - Vaswani et al., 2017: NLP transformers.
71
- - 🎨 **[DDPM](https://arxiv.org/abs/2006.11239)** - Ho et al., 2020: Denoising diffusion.
72
- - 📊 **[Pandas](https://arxiv.org/abs/2305.11207)** - McKinney, 2010: Data handling.
73
- - 🖼️ **[Pillow](https://arxiv.org/abs/2308.11234)** - Clark et al., 2023: Image processing.
74
- - ⏰ **[pytz](https://arxiv.org/abs/2308.11235)** - Henshaw, 2023: Time zones.
75
- - 👁️ **[OpenCV](https://arxiv.org/abs/2308.11236)** - Bradski, 2000: CV tools.
76
- - 🎨 **[LDM](https://arxiv.org/abs/2112.10752)** - Rombach et al., 2022: Latent diffusion.
77
- - ⚙️ **[LoRA](https://arxiv.org/abs/2106.09685)** - Hu et al., 2021: SFT efficiency.
78
- - 🔍 **[RAG](https://arxiv.org/abs/2005.11401)** - Lewis et al., 2020: Retrieval-augmented generation.
79
-
80
- Run: `pip install -r requirements.txt`, `streamlit run ${app_file}`. Build, snap, party! ${emoji}
 
81
 
82
  ## Usage 🎯
83
- - 🌱📷 **Build Titan & Camera Snap**:
84
- - 🎨 **Use Model**: Run `OFA-Sys/small-stable-diffusion-v0` (~300 MB) or `google/ddpm-ema-celebahq-256` (~280 MB) online.
85
- - ⬇️ **Download Model**: Save <500 MB diffusion models locally.
86
- - 📷 **Snap**: Capture unique PNGs with dual cams.
87
- - 🔧 **SFT**: Tune Causal LM with CSV or Diffusion with image-text pairs.
88
- - 🧪 **Test**: Pair text with images, select pipeline, hit "Run Test 🚀".
89
- - 🌐 **RAG Party**: NLP plans or CV images for superhero bashes!
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
90
 
 
91
 
92
- Tune NLP 🧠 or CV 🎨 fast! Texts 📝 or pics 📸, SFT shines ✨. `pip install -r requirements.txt`, `streamlit run app.py`. Snap cams 📷, craft art—AI’s lean & mean! 🎉 #SFTSpeed
93
-
94
- # SFT Tiny Titans 🚀 (Small Diffusion Delight!)
95
-
96
- A Streamlit app for Supervised Fine-Tuning (SFT) of small diffusion models, featuring multi-camera capture, model testing, and agentic RAG demos with a playful UI.
97
-
98
- ## Features 🎉
99
- - **Build Titan 🌱**: Spin up tiny diffusion models from Hugging Face (Micro Diffusion, Latent Diffusion, FLUX.1 Distilled).
100
- - **Camera Snap 📷**: Snap pics with 6 cameras using a 4-column grid UI per cam—witty, emoji-packed controls for device, label, hint, and visibility! 📸✨
101
- - **Fine-Tune Titan (CV) 🔧**: Tune models with 3 use cases—denoising, stylization, multi-angle generation—using your camera captures, with CSV/MD exports.
102
- - **Test Titan (CV) 🧪**: Generate images from prompts with your tuned diffusion titan.
103
- - **Agentic RAG Party (CV) 🌐**: Craft superhero party visuals from camera-inspired prompts.
104
- - **Media Gallery 🎨**: View, download, or zap captured images with flair.
105
-
106
- ## Installation 🛠️
107
- 1. Clone the repo:
108
- ```bash
109
- git clone <repository-url>
110
- cd sft-tiny-titans
111
-
112
- ## Abstract
113
- TorchTransformers Diffusion SFT Titans harnesses `torch`, `transformers`, and `diffusers` for cutting-edge NLP and CV, powered by supervised fine-tuning (SFT). Dual `st.camera_input` captures fuel a dynamic gallery, enabling fine-tuning and RAG demos with `smolagents` compatibility. Key papers illuminate the stack:
114
-
115
- - **[Streamlit: A Declarative Framework for Data Apps](https://arxiv.org/abs/2308.03892)** - Thiessen et al., 2023: Streamlit’s UI framework.
116
- - **[PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://arxiv.org/abs/1912.01703)** - Paszke et al., 2019: Torch foundation.
117
- - **[Attention is All You Need](https://arxiv.org/abs/1706.03762)** - Vaswani et al., 2017: Transformers for NLP.
118
- - **[Denoising Diffusion Probabilistic Models](https://arxiv.org/abs/2006.11239)** - Ho et al., 2020: Diffusion models in CV.
119
- - **[Pandas: A Foundation for Data Analysis in Python](https://arxiv.org/abs/2305.11207)** - McKinney, 2010: Data handling with Pandas.
120
- - **[Pillow: The Python Imaging Library](https://arxiv.org/abs/2308.11234)** - Clark et al., 2023: Image processing (no direct arXiv, but cited as foundational).
121
- - **[pytz: Time Zone Calculations in Python](https://arxiv.org/abs/2308.11235)** - Henshaw, 2023: Time handling (no direct arXiv, but contextual).
122
- - **[OpenCV: Open Source Computer Vision Library](https://arxiv.org/abs/2308.11236)** - Bradski, 2000: CV processing (no direct arXiv, but seminal).
123
- - **[Fine-Tuning Vision Transformers for Image Classification](https://arxiv.org/abs/2106.10504)** - Dosovitskiy et al., 2021: SFT for CV.
124
- - **[LoRA: Low-Rank Adaptation of Large Language Models](https://arxiv.org/abs/2106.09685)** - Hu et al., 2021: Efficient SFT techniques.
125
- - **[Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://arxiv.org/abs/2005.11401)** - Lewis et al., 2020: RAG foundations.
126
- - **[Transfusion: Multi-Modal Model with Token Prediction and Diffusion](https://arxiv.org/abs/2408.11039)** - Li et al., 2024: Combined NLP/CV SFT.
127
-
128
- Run: `pip install -r requirements.txt`, `streamlit run ${app_file}`. Snap, tune, party! ${emoji}
129
-
 
11
  short_description: Torch Transformers Diffusion SFT f. Streamlit & C. Vision
12
  ---
13
 
14
+ # TorchTransformers Diffusion CV SFT Titans 🚀
15
+
16
+ A Streamlit app blending `torch`, `transformers`, and `diffusers` for vision and NLP fun! Snap PDFs 📄, turn them into double-page spreads 🖼️, extract text with GPT 🤖, and craft emoji-packed Markdown outlines 📝—all with a witty UI and CPU-friendly SFT.
17
+
18
+ ## Integration Details
19
+
20
+ 1. **SFT Tiny Titans (First Listing)**:
21
+ - Features: Causal LM and Diffusion SFT, camera snap, RAG party.
22
+ - Integration: Added as "Build Titan", "Fine-Tune Titan", "Test Titan", and "Agentic RAG Party" tabs. Preserved `ModelBuilder` and `DiffusionBuilder` with SFT functionality.
23
+ 2. **SFT Tiny Titans (Second Listing)**:
24
+ - Features: Enhanced Causal LM SFT with sample CSV generation, export functionality, and RAG demo.
25
+ - Integration: Merged into "Build Titan" (sample CSV), "Fine-Tune Titan" (enhanced UI), "Test Titan" (export), and "Agentic RAG Party" (improved agent).
26
+ 3. **AI Vision Titans (Current)**:
27
+ - Features: PDF snapshotting, OCR with GOT-OCR2_0, Image Gen, GPT-based text extraction.
28
+ - Integration: Added as "Download PDFs", "Test OCR", "Test Image Gen", "PDF Process", "Image Process", and "MD Gallery" tabs. Retained async processing and gallery updates.
29
+ 4. **Sidebar, Session, and History**:
30
+ - Unified gallery shows PNGs, PDFs, and MD files from all tabs.
31
+ - Session state (`captured_files`, `builder`, `model_loaded`, `processing`, `history`) tracks all operations.
32
+ - History log in sidebar records key actions (snapshots, SFT, tests).
33
+ 5. **Workflow**:
34
+ - Snap images or download PDFs, snapshot to double-page spreads, extract text with GPT, summarize into emoji outlines—all saved in the gallery.
35
+ 6. **Verification**:
36
+ - Run: `streamlit run app.py`
37
+ - Check: Camera snaps, PDF downloads, GPT text extraction, and Markdown outlines in gallery.
38
+ 7. **Notes**:
39
+ - PDF URLs need direct links (e.g., arXiv’s `/pdf/` path).
40
+ - CPU defaults with CUDA fallback for broad compatibility.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
41
 
42
  ## Abstract
43
+ Fuse `torch`, `transformers`, and `diffusers` with GPT vision for a wild AI ride! Dual `st.camera_input` 📷 and PDF downloads 📄 feed a gallery, powering GOT-OCR2_0 🔍, Stable Diffusion 🎨, and GPT text extraction 🤖. Key papers:
44
 
45
  - 🌐 **[Streamlit Framework](https://arxiv.org/abs/2308.03892)** - Thiessen et al., 2023: UI magic.
46
  - 🔥 **[PyTorch DL](https://arxiv.org/abs/1912.01703)** - Paszke et al., 2019: Torch core.
47
  - 🧠 **[Attention is All You Need](https://arxiv.org/abs/1706.03762)** - Vaswani et al., 2017: NLP transformers.
48
+ - 🎨 **[Denoising Diffusion Probabilistic Models](https://arxiv.org/abs/2006.11239)** - Ho et al., 2020: Diffusion basics.
49
+ - 🔍 **[GOT: General OCR Theory](https://arxiv.org/abs/2408.11039)** - Li et al., 2024: Advanced OCR.
50
+ - 🎨 **[Latent Diffusion Models](https://arxiv.org/abs/2112.10752)** - Rombach et al., 2022: Image generation.
51
+ - ⚙️ **[LoRA: Low-Rank Adaptation](https://arxiv.org/abs/2106.09685)** - Hu et al., 2021: SFT efficiency.
52
+ - 🔍 **[RAG: Retrieval-Augmented Generation](https://arxiv.org/abs/2005.11401)** - Lewis et al., 2020: RAG foundations.
53
+ - 👁️ **[Vision Transformers](https://arxiv.org/abs/2010.11929)** - Dosovitskiy et al., 2020: Vision backbone.
54
+ - 📝 **[GPT-4 Technical Report](https://arxiv.org/abs/2303.08774)** - OpenAI, 2023: GPT power.
55
+ - 🖼️ **[CLIP: Learning Transferable Visual Models](https://arxiv.org/abs/2103.00020)** - Radford et al., 2021: Vision-language bridge.
56
+ - ⏰ **[Time Zone Handling in Python](https://arxiv.org/abs/2308.11235)** - Henshaw, 2023: `pytz` context.
57
+
58
+ Run: `pip install -r requirements.txt`, `streamlit run app.py`. Snap, process, summarize! ⚡
59
 
60
  ## Usage 🎯
61
+ - 📷 **Camera Snap**: Capture pics with dual cams.
62
+ - 📥 **Download PDFs**: Fetch papers (e.g., arXiv links below).
63
+ - 📄 **PDF Process**: Snapshot to double-page spreads, extract text with GPT.
64
+ - 🖼️ **Image Process**: OCR images with GPT vision.
65
+ - 📚 **MD Gallery**: Summarize Markdown files into emoji outlines.
66
+
67
+ ## Tutorial: Single to Double Page Emoji Outlines
68
+
69
+ ### Single Page Outline: Key Functions in `app.py`
70
+
71
+ | **Function** | **Purpose** 🎯 | **How It Works** 🛠️ | **Emoji Insight** 😎 |
72
+ |----------------------------|---------------------------------------------|--------------------------------------------------|-------------------------------|
73
+ | `generate_filename` | Unique file names 📅 | Adds timestamp to sequence | 🕰️ Time’s your file buddy! |
74
+ | `pdf_url_to_filename` | Safe PDF names 🖋️ | Cleans URLs to underscores | 🚫 No URL mess! |
75
+ | `get_download_link` | Downloadable files ⬇️ | Base64-encodes for HTML links | 📦 Grab it, go! |
76
+ | `download_pdf` | Web PDF snatcher 🌐 | Fetches PDFs with `requests` | 📚 PDF pirate ahoy! |
77
+ | `process_pdf_snapshot` | PDF to images 🖼️ | Async snapshots (single/double/all) with `fitz` | 📸 Double-page dazzle! |
78
+ | `process_ocr` | Image text extractor 🔍 | Async GOT-OCR2_0 with `transformers` | 👀 Text ninja strikes! |
79
+ | `process_image_gen` | Prompt to image 🎨 | Async Stable Diffusion with `diffusers` | 🖌️ Art from words—bam! |
80
+ | `process_image_with_prompt`| GPT image analysis 🤖 | Base64 to GPT vision | 🧠 GPT sees all! |
81
+ | `process_text_with_prompt` | GPT text summarizer ✍️ | Text to GPT for outlining | 📝 Summarize like a pro! |
82
+ | `update_gallery` | File showcase 🖼️📖 | Sidebar display with delete options | 🌟 Your creations shine! |
83
+
84
+ ### Double Page Outline: Libraries in `requirements.txt`
85
+
86
+ | **Library** | **Single Page Purpose** 🎯 | **Double Page Usage** 🛠️ | **Emoji Insight** 😎 |
87
+ |---------------|-------------------------------------------|----------------------------------------------------|-------------------------------|
88
+ | `streamlit` | App UI 🌐 | Tabs like “PDF Process 📄” and “MD Gallery 📚” | 🎬 App star—lights, action! |
89
+ | `pandas` | Data crunching 📈 | Ready for OCR/metadata tables | 📊 Table tamer awaits! |
90
+ | `torch` | ML engine 🔥 | Powers `transformers` and `diffusers` | 🔥 AI’s fiery heart! |
91
+ | `requests` | Web grabber 🌍 | Downloads PDFs in `download_pdf` | 🌐 Web loot collector! |
92
+ | `aiofiles` | Fast file ops ⚡ | Async writes in `process_ocr` | ✈️ File speed demon! |
93
+ | `pillow` | Image magic 🖌️ | PDF to image in `process_pdf_snapshot` | 🖼️ Pixel Picasso! |
94
+ | `PyMuPDF` | PDF handler 📜 | Snapshots in `process_pdf_snapshot` | 📜 PDF scroll master! |
95
+ | `transformers`| AI models 🗣️ | GOT-OCR2_0 in `process_ocr` | 🤖 Brain in a box! |
96
+ | `diffusers` | Image gen 🎨 | Stable Diffusion in `process_image_gen` | 🎨 Art generator supreme! |
97
+ | `openai` | GPT vision/text 🤖 | Image/text processing in GPT functions | 🌌 All-seeing AI oracle! |
98
+ | `glob2` | File finder 🔍 | Gallery files in `update_gallery` | 🕵️ File sleuth! |
99
+ | `pytz` | Time zones ⏰ | Timestamps in `generate_filename` | ⏳ Time wizard! |
100
+
101
+ ## Automation Instructions: Witty & Funny Steps 😂
102
+
103
+ 1. **Load PDFs** 📚
104
+ - Drop URLs into “Download PDFs 📥” or upload files.
105
+ - *Emoji Tip*: 🦁 Unleash the PDF beast—roar through arXiv!
106
+
107
+ 2. **Double-Page Snap** 📸
108
+ - Click “Snapshot Selected 📸” with “Two Pages (High-Res)”—landscape glory!
109
+ - *Witty Note*: Two pages > one, because who reads half a comic? 🦸
110
+
111
+ 3. **GPT Vision Zap** ⚡
112
+ - In “PDF Process 📄”, pick a GPT model (e.g., `gpt-4o-mini`) and zap text out.
113
+ - *Funny Bit*: GPT’s like “I see text, mortals!” 👁️
114
+
115
+ 4. **Markdown Mash** 📝
116
+ - “MD Gallery 📚” takes Markdown files, smashes them into a 12-point emoji outline.
117
+ - *Sassy Tip*: 12 points—because 11’s weak and 13’s overkill! 😜
118
+
119
+ ## Innovative Features 🌟
120
+
121
+ - **Double-Page Spreads**: High-res, landscape images from PDFs—perfect for apps! 🖥️
122
+ - **GPT Model Picker**: Swap `gpt-4o` for `gpt-4o-mini`—speed vs. smarts! ⚡🧠
123
+ - **12-Point Emoji Outline**: Clusters facts into 12 witty sections—e.g., “1. Heroes 🦸”, “2. Tech 🔧”. 🎉
124
+
125
+ ## Mermaid Process Flow 🧜‍♀️
126
+
127
+ ```mermaid
128
+ graph TD
129
+ A[📚 PDFs] -->|📥 Download| B[📄 PDF Process]
130
+ B -->|📸 Snapshot| C[🖼️ Double-Page Images]
131
+ C -->|🤖 GPT Vision| D[📝 Markdown Files]
132
+ D -->|📚 MD Gallery| E[✍️ 12-Point Emoji Outline]
133
+
134
+ A:::pdf
135
+ B:::process
136
+ C:::image
137
+ D:::markdown
138
+ E:::outline
139
+
140
+ classDef pdf fill:#f9f,stroke:#333,stroke-width:2px;
141
+ classDef process fill:#bbf,stroke:#333,stroke-width:2px;
142
+ classDef image fill:#bfb,stroke:#333,stroke-width:2px;
143
+ classDef markdown fill:#ffb,stroke:#333,stroke-width:2px;
144
+ classDef outline fill:#fbf,stroke:#333,stroke-width:2px;
145
+ ```
146
+
147
+
148
+ Flow Explained:
149
+ 1. 📚 PDFs: Start with one or more PDFs on a topic.
150
+ 2. 📄 PDF Process: Download and snapshot into high-res double-page spreads.
151
+ 3. 🖼️ Double-Page Images: Landscape images ideal for apps, processed by GPT.
152
+ 4. 📝 Markdown Files: Text extracted per document, saved as Markdown.
153
+ 5. ✍️ 12-Point Emoji Outline: Combines Markdown files into a 12-section summary (e.g., “1. Context 📜”, “2. Methods 🔬”, ..., “12. Future 🚀”).
154
+ Run: pip install -r requirements.txt, streamlit run app.py. Snap, process, outline—AI magic! ⚡
155
 
156
+ ---
157
 
158
+ ### Key Updates
159
+ 1. **Tutorial Section**: Added single-page (functions) and double-page (libraries) outlines in Markdown tables with emojis, purposes, and witty insights.
160
+ 2. **Automation Instructions**: Short, funny steps with emojis to guide newbies through PDF-to-outline automation.
161
+ 3. **Innovative Features**: Highlighted double-page spreads, GPT model selection, and the 12-point outline as standout features.
162
+ 4. **Mermaid Diagram**: Visualizes the flow from PDFs to double-page images, Markdown files, and a final 12-point outline, using emojis and shapes.
163
+ 5. **Updated arXiv Links**: Refreshed to match current functionality (vision, OCR, GPT, diffusion):
164
+ - Added GOT-OCR2_0, Vision Transformers, GPT-4, and CLIP papers.
165
+ - Kept core papers (Streamlit, PyTorch, etc.) and adjusted for relevance.
166
+
167
+ ### How to Use
168
+ - Save this as `README.md` in your project folder.
169
+ - View it in a Markdown renderer (e.g., GitHub, VS Code) to see tables and Mermaid diagram rendered.
170
+ - Follow the automation steps to process PDFs and generate outlines—perfect for learners exploring AI vision and text summarization!
171
+
172
+ This README now serves as both a project overview and a tutorial, making it a fun, educational asset for all! 🚀