Instructions to use recursal/QRWKV7-7B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use recursal/QRWKV7-7B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="recursal/QRWKV7-7B-Instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("recursal/QRWKV7-7B-Instruct", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use recursal/QRWKV7-7B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "recursal/QRWKV7-7B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "recursal/QRWKV7-7B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/recursal/QRWKV7-7B-Instruct

SGLang

How to use recursal/QRWKV7-7B-Instruct 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 "recursal/QRWKV7-7B-Instruct" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "recursal/QRWKV7-7B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "recursal/QRWKV7-7B-Instruct" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "recursal/QRWKV7-7B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use recursal/QRWKV7-7B-Instruct with Docker Model Runner:
```
docker model run hf.co/recursal/QRWKV7-7B-Instruct
```

Can you convert Qwen 3.5 9b?

by Regrin - opened 22 days ago

Discussion

Regrin

22 days ago

Hello!
I really like the RWKV architecture, but RWKV-7 7.2b produces very poor results.
I was interested in this converted model. But here's the problem: it's based on Qwen 2.5, which is a very outdated model.
Maybe you can convert Qwen 3.5 9b?

Although, it's better not to do that. You'd better wait until Qwen 3.6 9b comes out and convert it. It will be better that way.

I would be very grateful!
I would also like you to make the RWKV hidden state larger, if possible. RWKV-7 7.2b has about 16 megabytes, and I think that's a bit small. It doesn't hold context well...

Thanks in advance!

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment