Instructions to use webmp3/Sakura-tencent-WeVisCut-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use webmp3/Sakura-tencent-WeVisCut-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Use Docker

docker model run hf.co/webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use webmp3/Sakura-tencent-WeVisCut-GGUF with Ollama:
```
ollama run hf.co/webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M
```
Unsloth Desktop

How to use webmp3/Sakura-tencent-WeVisCut-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use webmp3/Sakura-tencent-WeVisCut-GGUF with Docker Model Runner:
```
docker model run hf.co/webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M
```

Lemonade

How to use webmp3/Sakura-tencent-WeVisCut-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Sakura-tencent-WeVisCut-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use webmp3/Sakura-tencent-WeVisCut-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use webmp3/Sakura-tencent-WeVisCut-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "webmp3/Sakura-tencent-WeVisCut-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Sakura · GGUF quantization of tencent/WeVisCut

Independent community quantization. Not an official release of the upstream author.

1. Model Overview

Architecture: qwen35 · source formats: gguf.

Original: 17.9 GB (BF16 GGUF, the source of these quantizations).

Quant	Size	Compared with the original
Q2_K	3.8 GB	79 % smaller
Q3_K_M	4.6 GB	74 % smaller
Q4_K_M	5.6 GB	69 % smaller
Q5_K_M	6.5 GB	64 % smaller

2. Original Model / Attribution

Original model: tencent/WeVisCut
Submitted repository: prithivMLmods/tencent-WeVisCut-GGUF
Declared base models: tencent/WeVisCut (relation: quantized)
Quantized from: prithivMLmods/tencent-WeVisCut-GGUF (file WeVisCut.BF16.gguf).
Quantized by webmp3. Upstream authors retain their license and attribution.

3. Files

File	Size	Quant
`Sakura-tencent-WeVisCut-Q2_K.gguf`	3.8 GB	Q2_K
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	4.6 GB	Q3_K_M
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	5.6 GB	Q4_K_M
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	6.5 GB	Q5_K_M
`WeVisCut.mmproj-bf16.gguf`	0.9 GB	vision projector

WeVisCut.mmproj-bf16.gguf is the vision projector of the original GGUF repository prithivMLmods/tencent-WeVisCut-GGUF, copied unchanged. Use it together with any of the files above to give the model images. Image input has not been tested for this release.

4. Download and Usage

llama-cli -m <file>.gguf -p "Hello" -n 32

# with images
llama-mtmd-cli -m <file>.gguf --mmproj WeVisCut.mmproj-bf16.gguf --image <picture>.jpg -p "Describe this image."

Needs a llama.cpp build that supports this architecture. If a quantization is split into numbered parts, keep them together and load the first one.

5. Verification

GGUF structure valid: True
Real inference test (every file loaded by llama.cpp and used to generate text once): yes
No coding benchmark, perplexity or quality score is claimed unless a measured value is listed here.

Technical details: exact versions, hashes, how these were made

Exact versions

Original model commit: cbfba25be41607a6e7e52bbd2069aee2b7d788b6 (model card at that commit)
Vision projector: commit cbfba25be41607a6e7e52bbd2069aee2b7d788b6 of prithivMLmods/tencent-WeVisCut-GGUF, SHA-256 221661c4c2ee748448110d564d78b7fb0cd3f2e32516b6f28848594a4ca4b225
Quantized from: commit cbfba25be41607a6e7e52bbd2069aee2b7d788b6, SHA-256 b24c79f365d8775548fed8fb8d2cdafc490196bab06f380ef93be5c8bda06c2c (provenance verified: True)

File hashes

File	Bytes	SHA-256	Quant
`Sakura-tencent-WeVisCut-Q2_K.gguf`	3827262528	`205e22f7eeb1ae41db8818834c42297606cbe9fb1256e1278c6d4f4d90e76f30`	Q2_K
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	4623524928	`cc692f00debbf3212855257cb0c7a5a28ec366880e92c91b2189034f03516b87`	Q3_K_M
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	5629109312	`d6770c1ca433ff8d9b6885405333f0f4fbdf265adfa2a74a39947f79d145a2c6`	Q4_K_M
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	6467970112	`eb406c44385882b1eea987d652733912227cbe871f9ca3a756a9f81ae7d11956`	Q5_K_M

How these were made

This route performs existing-GGUF requantization from the pinned BF16 source GGUF named above. This job did not convert the upstream BF16/safetensors checkpoint.

No importance-matrix calibration was run by this job.

The quant label names the llama.cpp preset. Stock llama.cpp keeps some tensors (for example normalization weights) in higher precision, so not every tensor is stored in the named type; the table shows what each file really contains.

File	Tensors by type
`Sakura-tencent-WeVisCut-Q2_K.gguf`	F32: 177, Q2_K: 177, Q3_K: 40, Q4_K: 32, Q6_K: 1
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	F32: 177, Q3_K: 177, Q4_K: 68, Q5_K: 4, Q6_K: 1
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	F32: 177, Q4_K: 217, Q6_K: 33
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	F32: 177, Q5_K: 217, Q6_K: 33

Test builds

File	llama.cpp build	Executable SHA-256
`Sakura-tencent-WeVisCut-Q2_K.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`

6. License Information

Source license: apache-2.0. apache-2.0 allows redistribution with attribution

中文说明 · 樱花 Sakura（点击展开）

1. 概述

独立的社区量化版本，并非上游作者的官方发布。

架构：qwen35 · 源格式：gguf。

原始模型：17.9 GB（BF16 GGUF，这些量化版本的来源）。

量化	大小	相比原始模型
Q2_K	3.8 GB	小 79 %
Q3_K_M	4.6 GB	小 74 %
Q4_K_M	5.6 GB	小 69 %
Q5_K_M	6.5 GB	小 64 %

2. 原始模型与署名

原始模型：tencent/WeVisCut
提交的仓库：prithivMLmods/tencent-WeVisCut-GGUF
声明的基础模型：tencent/WeVisCut（关系：quantized）
量化来源：prithivMLmods/tencent-WeVisCut-GGUF（文件 WeVisCut.BF16.gguf）。
量化者：webmp3。上游作者保留其许可证与署名。

3. 文件

文件	大小	量化
`Sakura-tencent-WeVisCut-Q2_K.gguf`	3.8 GB	Q2_K
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	4.6 GB	Q3_K_M
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	5.6 GB	Q4_K_M
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	6.5 GB	Q5_K_M
`WeVisCut.mmproj-bf16.gguf`	0.9 GB	vision projector

WeVisCut.mmproj-bf16.gguf 是原始 GGUF 仓库 prithivMLmods/tencent-WeVisCut-GGUF 的视觉投影文件，原样复制。与上方任一文件配合使用即可输入图像。本版本未测试图像输入。

4. 下载与使用

llama-cli -m <file>.gguf -p "Hello" -n 32

# with images
llama-mtmd-cli -m <file>.gguf --mmproj WeVisCut.mmproj-bf16.gguf --image <picture>.jpg -p "Describe this image."

需要支持该架构的 llama.cpp 版本。如果某个量化被拆分为带编号的多个部分，请将它们放在一起并加载第一个。

5. 验证

GGUF 结构有效：True
真实推理测试（每个文件都由 llama.cpp 加载并生成一次文本）：yes
除非此处列出了实测数值，否则不声称任何编程基准、困惑度或质量分数。

技术细节：确切版本、哈希、制作方式

确切版本

原始模型提交：cbfba25be41607a6e7e52bbd2069aee2b7d788b6（该提交的模型卡）
视觉投影文件：prithivMLmods/tencent-WeVisCut-GGUF 的提交 cbfba25be41607a6e7e52bbd2069aee2b7d788b6，SHA-256 221661c4c2ee748448110d564d78b7fb0cd3f2e32516b6f28848594a4ca4b225
量化来源：提交 cbfba25be41607a6e7e52bbd2069aee2b7d788b6，SHA-256 b24c79f365d8775548fed8fb8d2cdafc490196bab06f380ef93be5c8bda06c2c（出处已验证：True）

文件哈希

文件	字节数	SHA-256	量化
`Sakura-tencent-WeVisCut-Q2_K.gguf`	3827262528	`205e22f7eeb1ae41db8818834c42297606cbe9fb1256e1278c6d4f4d90e76f30`	Q2_K
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	4623524928	`cc692f00debbf3212855257cb0c7a5a28ec366880e92c91b2189034f03516b87`	Q3_K_M
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	5629109312	`d6770c1ca433ff8d9b6885405333f0f4fbdf265adfa2a74a39947f79d145a2c6`	Q4_K_M
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	6467970112	`eb406c44385882b1eea987d652733912227cbe871f9ca3a756a9f81ae7d11956`	Q5_K_M

制作方式

该路线从上方标明的固定版本 BF16 源 GGUF 进行现有 GGUF 的再量化。本任务没有转换上游的 BF16/safetensors 检查点。

本任务没有运行重要性矩阵（imatrix）校准。

量化标签指 llama.cpp 的预设名称。标准 llama.cpp 会把部分张量（例如归一化权重）保留为更高精度，因此并非每个张量都以所标注的类型存储；下表列出每个文件实际包含的内容。

文件	按类型的张量数
`Sakura-tencent-WeVisCut-Q2_K.gguf`	F32: 177, Q2_K: 177, Q3_K: 40, Q4_K: 32, Q6_K: 1
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	F32: 177, Q3_K: 177, Q4_K: 68, Q5_K: 4, Q6_K: 1
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	F32: 177, Q4_K: 217, Q6_K: 33
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	F32: 177, Q5_K: 217, Q6_K: 33

测试版本

文件	llama.cpp 版本	可执行文件 SHA-256
`Sakura-tencent-WeVisCut-Q2_K.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q3_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q4_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`
`Sakura-tencent-WeVisCut-Q5_K_M.gguf`	`b11669-620d2225c`	`3b7167843ffb0f3677be964dc80cfc59a0167c82e11eff62f6334f66e831f209`

6. 许可证信息

源许可证：apache-2.0。apache-2.0 allows redistribution with attribution

完整独立的中文版：README_zh.md

Downloads last month: 260

GGUF

Model size

9B params

Architecture

qwen35

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for webmp3/Sakura-tencent-WeVisCut-GGUF

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Finetuned

tencent/WeVisCut

Quantized

(2)

this model