Instructions to use rwkv-rs/rwkv7-g1-st with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use rwkv-rs/rwkv7-g1-st with Transformers:

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

pipe = pipeline("text-generation", model="rwkv-rs/rwkv7-g1-st")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("rwkv-rs/rwkv7-g1-st", device_map="auto")

RWKV

How to use rwkv-rs/rwkv7-g1-st with RWKV:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use rwkv-rs/rwkv7-g1-st with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "rwkv-rs/rwkv7-g1-st"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "rwkv-rs/rwkv7-g1-st",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/rwkv-rs/rwkv7-g1-st

SGLang

How to use rwkv-rs/rwkv7-g1-st 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 "rwkv-rs/rwkv7-g1-st" \
    --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": "rwkv-rs/rwkv7-g1-st",
		"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 "rwkv-rs/rwkv7-g1-st" \
        --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": "rwkv-rs/rwkv7-g1-st",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use rwkv-rs/rwkv7-g1-st with Docker Model Runner:
```
docker model run hf.co/rwkv-rs/rwkv7-g1-st
```

RWKV-7 G1i — Native Transformers Safetensors

This repository contains the four canonical RWKV-7 G1i checkpoints converted from BlinkDL/rwkv7-g1 to native Transformers Safetensors. Each checkpoint is a self-contained subfolder with model weights, configuration, the native RWKV World tokenizer, and chat prompt template. No remote model code is required.

Checkpoints

Checkpoint	Parameters	Context	Subfolder
RWKV-7 G1i 1.5B	1,527,668,736	16,384	`rwkv7-g1i-1.5b-20260805-ctx16384`
RWKV-7 G1i 2.9B	2,948,065,280	16,384	`rwkv7-g1i-2.9b-20260805-ctx16384`
RWKV-7 G1i 7.2B	7,199,932,416	16,384	`rwkv7-g1i-7.2b-20260805-ctx16384`
RWKV-7 G1i 13.3B	13,270,298,624	16,384	`rwkv7-g1i-13.3b-20260805-ctx16384`

The repository root intentionally does not contain a default model checkpoint. Select one of the subfolders explicitly.

Loading

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer


repo_id = "rwkv-rs/rwkv7-g1-st"
subfolder = "rwkv7-g1i-7.2b-20260805-ctx16384"

tokenizer = AutoTokenizer.from_pretrained(
    repo_id,
    subfolder=subfolder,
    trust_remote_code=False,
)
model = AutoModelForCausalLM.from_pretrained(
    repo_id,
    subfolder=subfolder,
    dtype=torch.float16,
    trust_remote_code=False,
)

RWKV-7 product computation is delegated to the published FlashRWKV2 operator package. Unsupported hardware or a missing native operator fails closed; there is no repository-local PyTorch, FLA, or CUDA-kernel fallback.

Artifact contract

Public model identity: model_type="rwkv", RwkvForCausalLM.
Architecture generation: architecture_version="rwkv7".
Model vocabulary: 65,536 entries.
Tokenizer vocabulary: IDs 0 through 65,529.
BOS and EOS share token ID 0; PAD and UNK are intentionally undefined.
Weight dtype: bfloat16.
Each subfolder contains config.json, generation_config.json, Safetensors, tokenizer files, special_tokens_map.json, and chat_template.jinja.
The root rwkv_vocab_v20230424.json is retained as the immutable canonical source vocabulary asset used to build the native tokenizer artifacts.

The included chat template is a prompt interface. It does not claim that these base checkpoints are safety aligned.

License

The converted model weights are distributed under Apache-2.0. FlashRWKV2 is a separate MIT-licensed operator package.

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for rwkv-rs/rwkv7-g1-st

Base model

BlinkDL/rwkv7-g1

Finetuned

(27)

this model