Instructions to use marcusmi4n/abeja-qwen2.5-7b-japanese-qnn with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use marcusmi4n/abeja-qwen2.5-7b-japanese-qnn with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="marcusmi4n/abeja-qwen2.5-7b-japanese-qnn", device_map="auto")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("marcusmi4n/abeja-qwen2.5-7b-japanese-qnn", dtype="auto", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use marcusmi4n/abeja-qwen2.5-7b-japanese-qnn with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/marcusmi4n/abeja-qwen2.5-7b-japanese-qnn

SGLang

How to use marcusmi4n/abeja-qwen2.5-7b-japanese-qnn 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 "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn" \
    --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": "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn",
		"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 "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn" \
        --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": "marcusmi4n/abeja-qwen2.5-7b-japanese-qnn",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use marcusmi4n/abeja-qwen2.5-7b-japanese-qnn with Docker Model Runner:
```
docker model run hf.co/marcusmi4n/abeja-qwen2.5-7b-japanese-qnn
```

abeja-qwen2.5-7b-japanese-qnn

File size: 910 Bytes

dd03ad6

{
  "model_name": "ABEJA-Qwen2.5-7b-Japanese-QNN",
  "version": "1.0.0",
  "description": "ABEJA Qwen 2.5 7B Japanese model compiled for Qualcomm QNN (Mock)",
  "language": "japanese",
  "hardware_target": "Qualcomm NPU (Snapdragon 8cx Gen 2+)",
  "quantization": "INT8",
  "models": {
    "prefill": {
      "path": "prefill/",
      "description": "Prefill model for initial token generation"
    },
    "token_generation": {
      "path": "token_gen/",
      "description": "Token generation model for subsequent tokens"
    }
  },
  "requirements": {
    "qnn_sdk_version": "2.18.0.240127",
    "onnxruntime_qnn": "latest",
    "python_version": ">=3.8"
  },
  "usage": {
    "inference": "Use onnxruntime-qnn with QNNExecutionProvider",
    "deployment": "Deploy to Qualcomm NPU hardware"
  },
  "notes": "This is a mock QNN compilation for demonstration. Real deployment requires QNN SDK installation."
}