Instructions to use sohv/nanokimi-mini with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use sohv/nanokimi-mini with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="sohv/nanokimi-mini", trust_remote_code=True)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("sohv/nanokimi-mini", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use sohv/nanokimi-mini with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "sohv/nanokimi-mini" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "sohv/nanokimi-mini", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/sohv/nanokimi-mini
- SGLang
How to use sohv/nanokimi-mini with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "sohv/nanokimi-mini" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "sohv/nanokimi-mini", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "sohv/nanokimi-mini" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "sohv/nanokimi-mini", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use sohv/nanokimi-mini with Docker Model Runner:
docker model run hf.co/sohv/nanokimi-mini
Upload README
Browse files
README.md
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: en
|
| 3 |
+
license: mit
|
| 4 |
+
library_name: transformers
|
| 5 |
+
tags:
|
| 6 |
+
- text-generation
|
| 7 |
+
- shakespeare
|
| 8 |
+
- transformer
|
| 9 |
+
- pytorch
|
| 10 |
+
pipeline_tag: text-generation
|
| 11 |
+
model_type: kimi-k2
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# nanokimi-mini
|
| 15 |
+
|
| 16 |
+
This repository contains a nanoKimi model checkpoint trained on Shakespeare dataset.
|
| 17 |
+
|
| 18 |
+
## Model Details
|
| 19 |
+
|
| 20 |
+
- **Architecture**: 12 layers, 12 heads, 768 embedding dimension
|
| 21 |
+
- **Training Data**: Shakespeare dataset
|
| 22 |
+
- **Features**: Mixture of Experts (8 experts), Latent Attention
|
| 23 |
+
- **Model Type**: Kimi-K2 (custom transformer)
|
| 24 |
+
|
| 25 |
+
## Files
|
| 26 |
+
|
| 27 |
+
- `pytorch_model.bin` - Model weights
|
| 28 |
+
- `config.json` - Model configuration
|
| 29 |
+
- `src/` - Source code for model architecture
|
| 30 |
+
- `modeling_kimik2.py` - HuggingFace wrapper
|
| 31 |
+
|
| 32 |
+
## Usage
|
| 33 |
+
|
| 34 |
+
```python
|
| 35 |
+
import torch
|
| 36 |
+
import json
|
| 37 |
+
from huggingface_hub import hf_hub_download
|
| 38 |
+
|
| 39 |
+
# Download files
|
| 40 |
+
config_path = hf_hub_download(repo_id="sohv/nanokimi-mini", filename="config.json")
|
| 41 |
+
weights_path = hf_hub_download(repo_id="sohv/nanokimi-mini", filename="pytorch_model.bin")
|
| 42 |
+
|
| 43 |
+
# Load config and weights
|
| 44 |
+
with open(config_path) as f:
|
| 45 |
+
config = json.load(f)
|
| 46 |
+
|
| 47 |
+
weights = torch.load(weights_path, map_location="cpu")
|
| 48 |
+
print("Model downloaded successfully!")
|
| 49 |
+
```
|
| 50 |
+
|
| 51 |
+
## License
|
| 52 |
+
|
| 53 |
+
MIT License
|