|
--- |
|
title: CPU Only Zero Shot Text Classification |
|
emoji: π |
|
colorFrom: gray |
|
colorTo: purple |
|
sdk: gradio |
|
sdk_version: 5.31.0 |
|
app_file: app.py |
|
pinned: false |
|
license: apache-2.0 |
|
short_description: ' CPU-only Zero-Shot Text Classification' |
|
--- |
|
# π·οΈ Zero-Shot Text Classification |
|
[](https://huggingface.co/spaces/your-username/zero-shot-classifier) |
|
[] |
|
[](https://huggingface.co/facebook/bart-large-mnli) |
|
[](LICENSE) |
|
|
|
--- |
|
|
|
## π Overview |
|
Unlock **zero-shot classification** for any textβno fine-tuning required. |
|
Define your own label set on the fly and see how well each label matches your input, powered by **BART-MNLI** on CPU. |
|
|
|
> **AI buzzwords:** |
|
> Natural Language Inference β’ Zeroβshot Learning β’ Transformer-based NLP β’ Real-time Inference β’ Edge Deployment β’ Cloud-native Demo |
|
|
|
--- |
|
|
|
## β¨ Features |
|
|
|
| π Feature | π Description | |
|
|--------------------------------|----------------------------------------------------------------| |
|
| **π Zero-Shot** | Classify into arbitrary categories without task-specific data | |
|
| **β‘ CPU-Only Inference** | Runs on free Hugging Face Spaces (2 vCPU / 16 GB RAM) | |
|
| **ποΈ Single vs. Multi-Label** | Toggle between exclusive or overlapping labels | |
|
| **π¨ Interactive UI** | Gradio Blocks with text input, label list, mode toggle, table | |
|
| **π§ No Training Needed** | Leverages pre-trained BART-MNLI via HF Transformers | |
|
| **βοΈ Instant Deploy** | Commit three filesβSpaces auto-builds & hosts your demo | |
|
|
|
--- |
|
|
|
## ποΈ How It Works |
|
|
|
1. **User Input** β Paste any sentence or paragraph. |
|
2. **Label Definition** β Enter comma-separated candidate labels (e.g. βPositive, Negative, Questionβ). |
|
3. **Model Inference** β Pipeline computes entailment scores for each label. |
|
4. **Result Table** β Display each label with its confidence score. |
|
|
|
All done **locally on the Space**, ensuring **privacy**, **zero API cost**, and **lightning speed**. |
|
|
|
--- |
|
|
|
## π οΈ Local Development |
|
|
|
```bash |
|
git clone https://github.com/your-username/zero-shot-classifier.git |
|
cd zero-shot-classifier |
|
python3 -m venv venv && source venv/bin/activate |
|
pip install -r requirements.txt |
|
python app.py |
|
|
|
## Latest Update |
|
|
|
- Updated BART-MNLI model for better accuracy. - May 29, 2025 π |
|
- Optimized CPU inference speed. - June 12, 2025 π |
|
- Added support for custom label sets. - June 11, 2025 π |
|
- Enhanced Gradio UI responsiveness. - June 10, 2025 π |
|
- Improved multi-label classification support. π·οΈ - June 09, 2025 π |
|
- Updated BART-MNLI model for better accuracy. - June 08, 2025 π |
|
- Optimized CPU inference speed. β‘ - June 07, 2025 π |
|
- Added support for custom label sets. - June 06, 2025 π |
|
- Enhanced Gradio UI responsiveness. π - June 05, 2025 π |
|
- Improved multi-label classification support. ποΈ - June 04, 2025 π |
|
- Updated BART-MNLI model for better accuracy. ποΈ - June 03, 2025 π |
|
- Optimized CPU inference speed. β‘ - June 02, 2025 π |
|
- Added support for custom label sets. π§ - June 01, 2025 π |
|
- Enhanced Gradio UI responsiveness. π - May 31, 2025 π |
|
- Improved multi-label classification support. π·οΈ - May 30, 2025 π |
|
|
|
**Website**: https://ghostainews.com/ |
|
**Discord**: https://discord.gg/BfA23aYz |