RWKV7-G1j-2.9B-GGUF

<!-- markdownlint-disable first-line-h1 --> <!-- markdownlint-disable html --> <div align="center"> <a href="https://www.rwkv.com/"> <img src="assets/rwkv-logo.webp" width="140" alt="RWKV logo" /> </a> <h1>RWKV7-G1j-2.9B-GGUF</h1> <p><strong>RWKV-7 “Goose” for local inference with Ollama and llama.cpp</strong></p> </div> <div align="center"> <a href="https://www.rwkv.com/"><img alt="RWKV website" src="https://img.shields.io/badge/Website-RWKV-16a7c9" /></a> <a href="https://huggingface.co/BlinkDL/rwkv7-g1"><img alt="Source model" src="https://img.shields.io/badge/Source-BlinkDL%2Frwkv7--g1-ffc107" /></a> <a href="https://github.com/BlinkDL/RWKV-LM"><img alt="RWKV GitHub" src="https://img.shields.io/badge/GitHub-RWKV--LM-181717?logo=github" /></a> <a href="https://arxiv.org/abs/2503.14456v2"><img alt="RWKV-7 paper" src="https://img.shields.io/badge/Paper-arXiv%3A2503.14456-b31b1b" /></a> <a href="./LICENSE"><img alt="Apache 2.0 license" src="https://img.shields.io/badge/License-Apache--2.0-4c8bf5" /></a> </div>

---

This repository provides the RWKV-7 G1j 2.9B base model in GGUF format for local text generation. Start with Q4_K_M, the formal Ollama default for this model size, or choose another file to balance storage, runtime memory, and weight precision.

[!IMPORTANT] This is a base language model, not a safety-aligned instruction-tuned assistant. The prompt formats below provide a conversational interface, but the model may not follow instructions consistently.

About RWKV-7

RWKV-7, also called Goose, is an attention-free recurrent language model. It keeps a constant-size recurrent state instead of an attention KV cache that grows with the preceding sequence. This gives RWKV constant recurrent-state size and constant inference work per generated token, while training remains parallelizable.

The fixed-state property describes the architecture's sequence state. Total memory use still depends on the selected weights, runtime, context configuration, batch size, and hardware backend.

G1j identifies this checkpoint revision. The original checkpoint is available from BlinkDL/rwkv7-g1. That immutable source file is 5,896,273,469 bytes with SHA-256 966f3420f833532aae3fb1fd6326533b08d43d23b7b03eaa2f0694a30b64a239.

Choose a file

Smaller quantizations generally reduce storage and runtime memory at the cost of some output quality. File size is not a guarantee of total RAM or VRAM use.

File integrity

Verify a downloaded file with sha256sum <filename> on Linux or Get-FileHash <filename> -Algorithm SHA256 in PowerShell.

Run with Ollama

Install Ollama 0.33.2 or later with RWKV-7 support. The formal 2.9B setup uses the Q4_K_M file and the included Modelfile.ollama.

1. Download the model

With the Hugging Face hf CLI:

hf download shoumenchougou/RWKV7-G1j-2.9B-GGUF rwkv7-g1j-2.9b-Q4_K_M.gguf Modelfile.ollama --local-dir rwkv7-g1j-2.9b
cd rwkv7-g1j-2.9b

You can instead download both files from the repository's file browser and place them in the same directory.

2. Create and run the Ollama model

ollama create rwkv7-g1j:2.9b -f Modelfile.ollama
ollama run rwkv7-g1j:2.9b

The setup is complete when ollama create reports success and ollama run returns generated text. This release passed Q4_K_M import, embedded-template rendering in thinking and non-thinking modes, and non-empty generation with a 16,384-token context setting.

The included Modelfile selects Q4_K_M, declares the minimum Ollama version, records the license, and sets the release context and sampling defaults:

FROM ./rwkv7-g1j-2.9b-Q4_K_M.gguf
REQUIRES 0.33.2
LICENSE """
Apache License 2.0
https://www.apache.org/licenses/LICENSE-2.0
"""
PARAMETER num_ctx 16384
PARAMETER temperature 1
PARAMETER top_k 128
PARAMETER top_p 0.5
PARAMETER frequency_penalty 0.5
PARAMETER presence_penalty 0.5
PARAMETER repeat_penalty 1
PARAMETER stop """

"""

The Modelfile intentionally does not define TEMPLATE. The GGUF files must contain the validated Jinja chat template in tokenizer.chat_template; Ollama and llama.cpp can then read it directly from the selected model. Adding an external Ollama TEMPLATE would override that embedded source of truth.

Keep the stop value on three lines as shown. Ollama 0.33.2 parses the triple-quoted blank line as two newline characters; writing the escaped text \n\n instead stores literal backslashes.

See Ollama's official guides for importing a GGUF model and the complete Modelfile reference.

Run with llama.cpp

Use a recent llama.cpp build that recognizes the rwkv7 architecture. From the directory containing the GGUF file, run:

llama-cli -m rwkv7-g1j-2.9b-Q4_K_M.gguf -c 16384 -n 256 -cnv

The files were structurally validated with the locked llama.cpp GGUF tooling. Runtime validation used the included Q4_K_M file through Ollama; direct llama.cpp generation was not part of this release check.

Prompt format

Runtimes that honor the embedded chat template apply the format automatically. After conversion, template inspection must confirm the standard and thinking-mode prefixes before publication.

For raw completion interfaces, use one of these generation prefixes.

Standard response:

User: Explain why recurrent models can use constant state.

Assistant:

Thinking response:

User: Solve the problem step by step.

Assistant: <think>

If your runtime requires explicit stop conditions, use token ID 0 or the text sequence \n\nUser:.

Context length

Use 16,384 tokens as the release context setting. Some GGUF inspection tools may display 1,048,576 from the RWKV-7 architecture metadata; that value is an architectural capacity field, not a claim that this checkpoint was trained or validated at that length.

If your runtime cannot allocate the requested context, lower the context value. This reduces the amount of preceding text available to the model.

Model details

The effective head count is 40 (2,560 / 64). GGUF inspectors may also show rwkv7.attention.head_count = 0; llama.cpp reserves that field but does not use it for RWKV-7.

Compatibility and troubleshooting

Runtime support changes over time. All five files passed GGUF v3 structure, bounds, metadata, tensor-count, embedded-template, size, and SHA-256 validation. Q4_K_M additionally passed Ollama import, template rendering, and real generation.

Limitations

• Quantization can change model outputs, and the effect varies by task.
• This release does not claim systematic benchmark parity between quantizations.
• The model may produce inaccurate, biased, unsafe, or otherwise undesirable text.
• The model is not a substitute for qualified professional judgment in high-stakes settings.
• The presence of chat and thinking prompts does not make the base model reliably instruction-following or safety-aligned.

License

The model is distributed under the Apache License 2.0. Review the source repository and license terms before redistribution or deployment.

Links

• RWKV website
• Original G1j checkpoint
• RWKV-LM repository
• RWKV-7 paper
• llama.cpp
• Ollama documentation

Citation

@misc{peng2025250314456,
  title         = {RWKV-7 "Goose" with Expressive Dynamic State Evolution},
  author        = {Bo Peng and Ruichong Zhang and Daniel Goldstein and Eric Alcaide and Xingjian Du and Haowen Hou and Jiaju Lin and Jiaxing Liu and Janna Lu and William Merrill and Guangyu Song and Kaifeng Tan and Saiteja Utpala and Nathan Wilce and Johan S. Wind and Tianyi Wu and Daniel Wuttke and Christian Zhou-Zheng},
  year          = {2025},
  eprint        = {2503.14456v2},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2503.14456v2},
}