# mindnlp
**Repository Path**: mindspore-lab/mindnlp
## Basic Information
- **Project Name**: mindnlp
- **Description**: MindNLP is an open source NLP library based on MindSpore.
- **Primary Language**: Unknown
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No
## Statistics
- **Stars**: 38
- **Forks**: 21
- **Created**: 2022-11-15
- **Last Updated**: 2026-03-08
## Categories & Tags
**Categories**: nature-language
**Tags**: None
## README
MindNLP
Run HuggingFace Models on MindSpore with Zero Code Changes
The easiest way to use 200,000+ HuggingFace models on Ascend NPU, GPU, and CPU
Quick Start β’
Features β’
Installation β’
Why MindNLP β’
Documentation
---
## π― What is MindNLP?
**MindNLP** bridges the gap between HuggingFace's massive model ecosystem and MindSpore's hardware acceleration. With just `import mindnlp`, you can run any HuggingFace model on **Ascend NPU**, **NVIDIA GPU**, or **CPU** - no code changes required.
```python
import mindnlp # That's it! HuggingFace now runs on MindSpore
from transformers import pipeline
pipe = pipeline("text-generation", model="Qwen/Qwen2-0.5B")
print(pipe("Hello, I am")[0]["generated_text"])
```
## β‘ Quick Start
### Text Generation with LLMs
```python
import mindspore
import mindnlp
from transformers import pipeline
pipe = pipeline(
"text-generation",
model="Qwen/Qwen3-8B",
ms_dtype=mindspore.bfloat16,
device_map="auto"
)
messages = [{"role": "user", "content": "Write a haiku about coding"}]
print(pipe(messages, max_new_tokens=100)[0]["generated_text"][-1]["content"])
```
### Image Generation with Stable Diffusion
```python
import mindspore
import mindnlp
from diffusers import DiffusionPipeline
pipe = DiffusionPipeline.from_pretrained(
"stable-diffusion-v1-5/stable-diffusion-v1-5",
ms_dtype=mindspore.float16
)
image = pipe("A sunset over mountains, oil painting style").images[0]
image.save("sunset.png")
```
### BERT for Text Classification
```python
import mindnlp
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
model = AutoModelForSequenceClassification.from_pretrained("bert-base-uncased")
inputs = tokenizer("MindNLP is awesome!", return_tensors="pt")
outputs = model(**inputs)
```
## β¨ Features
|
### π€ Full HuggingFace Compatibility
- **200,000+ models** from HuggingFace Hub
- **Transformers** - All model architectures
- **Diffusers** - Stable Diffusion, SDXL, ControlNet
- **Zero code changes** - Just `import mindnlp`
|
### π Hardware Acceleration
- **Ascend NPU** - Full support for Huawei AI chips
- **NVIDIA GPU** - CUDA acceleration
- **CPU** - Optimized CPU execution
- **Multi-device** - Automatic device placement
|
|
### π§ Advanced Capabilities
- **Mixed precision** - FP16/BF16 training & inference
- **Quantization** - INT8/INT4 with BitsAndBytes
- **Distributed** - Multi-GPU/NPU training
- **PEFT/LoRA** - Parameter-efficient fine-tuning
|
### π¦ Easy Integration
- **PyTorch-compatible API** via mindtorch
- **Safetensors** support for fast loading
- **Model Hub mirrors** for faster downloads
- **Comprehensive documentation**
|
## π§ͺ Mindtorch NPU Debugging
Mindtorch NPU ops are async by default. Use `torch.npu.synchronize()` when you need to block on results.
For debugging, set `ACL_LAUNCH_BLOCKING=1` to force per-op synchronization.
## π¦ Installation
```bash
# From PyPI (recommended)
pip install mindnlp
# From source (latest features)
pip install git+https://github.com/mindspore-lab/mindnlp.git
```
π Version Compatibility
| MindNLP | MindSpore | Python |
|---------|-----------|--------|
| 0.6.x | β₯2.7.1 | 3.10-3.11 |
| 0.5.x | 2.5.0-2.7.0 | 3.10-3.11 |
| 0.4.x | 2.2.x-2.5.0 | 3.9-3.11 |
## π‘ Why MindNLP?
| Feature | MindNLP | PyTorch + HF | TensorFlow + HF |
|---------|---------|--------------|-----------------|
| HuggingFace Models | β
200K+ | β
200K+ | β οΈ Limited |
| Ascend NPU Support | β
Native | β | β |
| Zero Code Migration | β
| - | β |
| Unified API | β
| β
| β |
| Chinese Model Support | β
Excellent | β
Good | β οΈ Limited |
### π Key Advantages
1. **Instant Migration**: Your existing HuggingFace code works immediately
2. **Ascend Optimization**: Native support for Huawei NPU hardware
3. **Production Ready**: Battle-tested in enterprise deployments
4. **Active Community**: Regular updates and responsive support
## πΊοΈ Supported Models
MindNLP supports **all models** from HuggingFace Transformers and Diffusers. Here are some popular ones:
| Category | Models |
|----------|--------|
| **LLMs** | Qwen, Llama, ChatGLM, Mistral, Phi, Gemma, BLOOM, Falcon |
| **Vision** | ViT, CLIP, Swin, ConvNeXt, SAM, BLIP |
| **Audio** | Whisper, Wav2Vec2, HuBERT, MusicGen |
| **Diffusion** | Stable Diffusion, SDXL, ControlNet |
| **Multimodal** | LLaVA, Qwen-VL, ALIGN |
π [View all supported models](https://mindnlp.cqu.ai/supported_models)
## π Resources
- π [Documentation](https://mindnlp.cqu.ai)
- π [Quick Start Guide](https://mindnlp.cqu.ai/quick_start)
- π [Tutorials](https://mindnlp.cqu.ai/tutorials/quick_start)
- π¬ [GitHub Discussions](https://github.com/mindspore-lab/mindnlp/discussions)
- π [Issue Tracker](https://github.com/mindspore-lab/mindnlp/issues)
## π€ Contributing
We welcome contributions! See our [Contributing Guide](https://mindnlp.cqu.ai/contribute) for details.
```bash
# Clone and install for development
git clone https://github.com/mindspore-lab/mindnlp.git
cd mindnlp
pip install -e ".[dev]"
```
## π₯ Community
Join the **MindSpore NLP SIG** (Special Interest Group) for discussions, events, and collaboration:
## β Star History
**If you find MindNLP useful, please consider giving it a star β - it helps the project grow!**
## π License
MindNLP is released under the [Apache 2.0 License](LICENSE).
## π Citation
```bibtex
@misc{mindnlp2022,
title={MindNLP: Easy-to-use and High-performance NLP and LLM Framework Based on MindSpore},
author={MindNLP Contributors},
howpublished={\url{https://github.com/mindspore-lab/mindnlp}},
year={2022}
}
```
---
Made with β€οΈ by the MindSpore Lab team