Instructions to use webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf webmp3/Sakura-Ripple-65B-Preview-GGUF:Q4_K_M

Use Docker

docker model run hf.co/webmp3/Sakura-Ripple-65B-Preview-GGUF:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use webmp3/Sakura-Ripple-65B-Preview-GGUF with Ollama:
```
ollama run hf.co/webmp3/Sakura-Ripple-65B-Preview-GGUF:Q4_K_M
```
Unsloth Desktop

How to use webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use webmp3/Sakura-Ripple-65B-Preview-GGUF with Docker Model Runner:
```
docker model run hf.co/webmp3/Sakura-Ripple-65B-Preview-GGUF:Q4_K_M
```

Lemonade

How to use webmp3/Sakura-Ripple-65B-Preview-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull webmp3/Sakura-Ripple-65B-Preview-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Sakura-Ripple-65B-Preview-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use webmp3/Sakura-Ripple-65B-Preview-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-Ripple-65B-Preview-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-Ripple-65B-Preview-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-Ripple-65B-Preview-GGUF / README.md

webmp3

Card: Q5_K_M is part of this release, remove the outdated note

5584c07 verified 7 days ago

preview code

Raw History Blame Contribute Delete

9.02 kB

---
language:
- en
- zh
license: apache-2.0
base_model: "ricursiveAI/Ripple-65B-Preview"
base_model_relation: quantized
hfqa_source_revision: da94cc013a8fda04a502f0a185083c7099ea06c1
tags:
- gguf
- quantization
- sakura
---
<p align="center"><img src="assets/sakura-logo.svg" alt="Sakura" width="560" /></p>

# Sakura · GGUF quantization of ricursiveAI/Ripple-65B-Preview

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



## 1. Model Overview

Architecture: `ricursive` · 64.8 B parameters · source formats: safetensors.

Original: **129.6 GB (BF16 GGUF, the source of these quantizations)**.

| Quant | Size | Compared with the original |
|---|---:|---|
| Q2_K | 23.7 GB | 82 % smaller |
| Q3_K_M | 31.0 GB | 76 % smaller |
| Q4_K_M | 38.8 GB | 70 % smaller |
| Q5_K_M | 45.8 GB | 65 % smaller |
## 2. Original Model / Attribution
- Original model: [ricursiveAI/Ripple-65B-Preview](https://huggingface.co/ricursiveAI/Ripple-65B-Preview)
- Quantized from: `AtomicChat/Ripple-65B-Preview-GGUF` (file `Ripple-65B-Preview-BF16.gguf`).
- Quantized by `webmp3`. Upstream authors retain their license and attribution.
## 3. Files
| File | Size | Quant |
|---|---:|---|
| `Sakura-Ripple-65B-Preview-Q2_K.gguf` | 23.7 GB | Q2_K |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | 31.0 GB | Q3_K_M |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | 38.8 GB | Q4_K_M |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | 45.8 GB | Q5_K_M |
## 4. Download and Usage
```bash
llama-cli -m <file>.gguf -p "Hello" -n 32
```
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.
<details>
<summary>Technical details: exact versions, hashes, how these were made</summary>
### Exact versions
- Original model commit: `da94cc013a8fda04a502f0a185083c7099ea06c1` ([model card at that commit](https://huggingface.co/ricursiveAI/Ripple-65B-Preview/blob/da94cc013a8fda04a502f0a185083c7099ea06c1/README.md))
- Quantized from: commit `df550a7178a87bbfb6ee2bdbd37603c553d536ee`, SHA-256 `1d7a551c5da46849db5f851e372b5b6b4af1fa1253e6a1e854cd7e77c37e9df6` (provenance verified: `True`)
### File hashes
| File | Bytes | SHA-256 | Quant |
|---|---:|---|---|
| `Sakura-Ripple-65B-Preview-Q2_K.gguf` | 23747659040 | `60b79fc65989267fcde15035ed01bb64a3915864e6e30d85159c5c1db066cc20` | Q2_K |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | 31019943200 | `fdf6815428dd2ed8ebf5d542525ca337bd1057d10d94bb74561736e24831574c` | Q3_K_M |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | 38798820640 | `7f5a149ac052e8125745d4f4b369c5e82c7ed8b43a5da1cd2b2a68fd6440f197` | Q4_K_M |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | 45757433120 | `a606978bedce45a40627773c7bd442d3c9d53c673e5c268d354bbb639950ffe4` | 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-Ripple-65B-Preview-Q2_K.gguf` | F32: 561, Q2_K: 401, Q3_K: 160, Q4_K: 80, Q6_K: 1 |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | F32: 561, Q3_K: 401, Q4_K: 233, Q5_K: 7, Q6_K: 1 |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | F32: 561, Q4_K: 561, Q6_K: 81 |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | F32: 561, Q5_K: 561, Q6_K: 81 |

### Test builds

| File | llama.cpp build | Executable SHA-256 |
|---|---|---|
| `Sakura-Ripple-65B-Preview-Q2_K.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |

</details>

## 6. License Information

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

<details>
<summary>中文说明 · 樱花 Sakura（点击展开）</summary>

<!-- hfqa-zh-begin -->
## 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：`ricursive` · 64.8 B 参数 · 源格式：safetensors。

原始模型：**129.6 GB（BF16 GGUF，这些量化版本的来源）**。

| 量化 | 大小 | 相比原始模型 |
|---|---:|---|
| Q2_K | 23.7 GB | 小 82 % |
| Q3_K_M | 31.0 GB | 小 76 % |
| Q4_K_M | 38.8 GB | 小 70 % |
| Q5_K_M | 45.8 GB | 小 65 % |
## 2. 原始模型与署名
- 原始模型：[ricursiveAI/Ripple-65B-Preview](https://huggingface.co/ricursiveAI/Ripple-65B-Preview)
- 量化来源：`AtomicChat/Ripple-65B-Preview-GGUF`（文件 `Ripple-65B-Preview-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-Ripple-65B-Preview-Q2_K.gguf` | 23.7 GB | Q2_K |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | 31.0 GB | Q3_K_M |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | 38.8 GB | Q4_K_M |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | 45.8 GB | Q5_K_M |
## 4. 下载与使用
```bash
llama-cli -m <file>.gguf -p "Hello" -n 32
```
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。
<details>
<summary>技术细节：确切版本、哈希、制作方式</summary>
### 确切版本
- 原始模型提交：`da94cc013a8fda04a502f0a185083c7099ea06c1`（[该提交的模型卡](https://huggingface.co/ricursiveAI/Ripple-65B-Preview/blob/da94cc013a8fda04a502f0a185083c7099ea06c1/README.md)）
- 量化来源：提交 `df550a7178a87bbfb6ee2bdbd37603c553d536ee`，SHA-256 `1d7a551c5da46849db5f851e372b5b6b4af1fa1253e6a1e854cd7e77c37e9df6`（出处已验证：`True`）
### 文件哈希
| 文件 | 字节数 | SHA-256 | 量化 |
|---|---:|---|---|
| `Sakura-Ripple-65B-Preview-Q2_K.gguf` | 23747659040 | `60b79fc65989267fcde15035ed01bb64a3915864e6e30d85159c5c1db066cc20` | Q2_K |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | 31019943200 | `fdf6815428dd2ed8ebf5d542525ca337bd1057d10d94bb74561736e24831574c` | Q3_K_M |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | 38798820640 | `7f5a149ac052e8125745d4f4b369c5e82c7ed8b43a5da1cd2b2a68fd6440f197` | Q4_K_M |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | 45757433120 | `a606978bedce45a40627773c7bd442d3c9d53c673e5c268d354bbb639950ffe4` | 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-Ripple-65B-Preview-Q2_K.gguf` | F32: 561, Q2_K: 401, Q3_K: 160, Q4_K: 80, Q6_K: 1 |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | F32: 561, Q3_K: 401, Q4_K: 233, Q5_K: 7, Q6_K: 1 |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | F32: 561, Q4_K: 561, Q6_K: 81 |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | F32: 561, Q5_K: 561, Q6_K: 81 |

### 测试版本

| 文件 | llama.cpp 版本 | 可执行文件 SHA-256 |
|---|---|---|
| `Sakura-Ripple-65B-Preview-Q2_K.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |
| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` | `b12189-80d2c0c2c` | `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` |

</details>

## 6. 许可证信息

源许可证：`apache-2.0`。apache-2.0 allows redistribution with attribution
<!-- hfqa-zh-end -->

完整独立的中文版：[README_zh.md](README_zh.md)

</details>

	---
	language:
	- en
	- zh
	license: apache-2.0
	base_model: "ricursiveAI/Ripple-65B-Preview"
	base_model_relation: quantized
	hfqa_source_revision: da94cc013a8fda04a502f0a185083c7099ea06c1
	tags:
	- gguf
	- quantization
	- sakura
	---

	<p align="center"><img src="assets/sakura-logo.svg" alt="Sakura" width="560" /></p>

	# Sakura · GGUF quantization of ricursiveAI/Ripple-65B-Preview

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



	## 1. Model Overview

	Architecture: `ricursive` · 64.8 B parameters · source formats: safetensors.

	Original: 129.6 GB (BF16 GGUF, the source of these quantizations).

	\| Quant \| Size \| Compared with the original \|
	\|---\|---:\|---\|
	\| Q2_K \| 23.7 GB \| 82 % smaller \|
	\| Q3_K_M \| 31.0 GB \| 76 % smaller \|
	\| Q4_K_M \| 38.8 GB \| 70 % smaller \|
	\| Q5_K_M \| 45.8 GB \| 65 % smaller \|

	## 2. Original Model / Attribution

	- Original model: [ricursiveAI/Ripple-65B-Preview](https://huggingface.co/ricursiveAI/Ripple-65B-Preview)
	- Quantized from: `AtomicChat/Ripple-65B-Preview-GGUF` (file `Ripple-65B-Preview-BF16.gguf`).
	- Quantized by `webmp3`. Upstream authors retain their license and attribution.

	## 3. Files

	\| File \| Size \| Quant \|
	\|---\|---:\|---\|
	\| `Sakura-Ripple-65B-Preview-Q2_K.gguf` \| 23.7 GB \| Q2_K \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| 31.0 GB \| Q3_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| 38.8 GB \| Q4_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| 45.8 GB \| Q5_K_M \|



	## 4. Download and Usage

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

	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.



	<details>
	<summary>Technical details: exact versions, hashes, how these were made</summary>

	### Exact versions

	- Original model commit: `da94cc013a8fda04a502f0a185083c7099ea06c1` ([model card at that commit](https://huggingface.co/ricursiveAI/Ripple-65B-Preview/blob/da94cc013a8fda04a502f0a185083c7099ea06c1/README.md))
	- Quantized from: commit `df550a7178a87bbfb6ee2bdbd37603c553d536ee`, SHA-256 `1d7a551c5da46849db5f851e372b5b6b4af1fa1253e6a1e854cd7e77c37e9df6` (provenance verified: `True`)

	### File hashes

	\| File \| Bytes \| SHA-256 \| Quant \|
	\|---\|---:\|---\|---\|
	\| `Sakura-Ripple-65B-Preview-Q2_K.gguf` \| 23747659040 \| `60b79fc65989267fcde15035ed01bb64a3915864e6e30d85159c5c1db066cc20` \| Q2_K \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| 31019943200 \| `fdf6815428dd2ed8ebf5d542525ca337bd1057d10d94bb74561736e24831574c` \| Q3_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| 38798820640 \| `7f5a149ac052e8125745d4f4b369c5e82c7ed8b43a5da1cd2b2a68fd6440f197` \| Q4_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| 45757433120 \| `a606978bedce45a40627773c7bd442d3c9d53c673e5c268d354bbb639950ffe4` \| 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-Ripple-65B-Preview-Q2_K.gguf` \| F32: 561, Q2_K: 401, Q3_K: 160, Q4_K: 80, Q6_K: 1 \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| F32: 561, Q3_K: 401, Q4_K: 233, Q5_K: 7, Q6_K: 1 \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| F32: 561, Q4_K: 561, Q6_K: 81 \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| F32: 561, Q5_K: 561, Q6_K: 81 \|

	### Test builds

	\| File \| llama.cpp build \| Executable SHA-256 \|
	\|---\|---\|---\|
	\| `Sakura-Ripple-65B-Preview-Q2_K.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|

	</details>

	## 6. License Information

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

	<details>
	<summary>中文说明 · 樱花 Sakura（点击展开）</summary>

	<!-- hfqa-zh-begin -->
	## 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：`ricursive` · 64.8 B 参数 · 源格式：safetensors。

	原始模型：129.6 GB（BF16 GGUF，这些量化版本的来源）。

	\| 量化 \| 大小 \| 相比原始模型 \|
	\|---\|---:\|---\|
	\| Q2_K \| 23.7 GB \| 小 82 % \|
	\| Q3_K_M \| 31.0 GB \| 小 76 % \|
	\| Q4_K_M \| 38.8 GB \| 小 70 % \|
	\| Q5_K_M \| 45.8 GB \| 小 65 % \|

	## 2. 原始模型与署名

	- 原始模型：[ricursiveAI/Ripple-65B-Preview](https://huggingface.co/ricursiveAI/Ripple-65B-Preview)
	- 量化来源：`AtomicChat/Ripple-65B-Preview-GGUF`（文件 `Ripple-65B-Preview-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-Ripple-65B-Preview-Q2_K.gguf` \| 23.7 GB \| Q2_K \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| 31.0 GB \| Q3_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| 38.8 GB \| Q4_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| 45.8 GB \| Q5_K_M \|

	## 4. 下载与使用

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

	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。



	<details>
	<summary>技术细节：确切版本、哈希、制作方式</summary>

	### 确切版本

	- 原始模型提交：`da94cc013a8fda04a502f0a185083c7099ea06c1`（[该提交的模型卡](https://huggingface.co/ricursiveAI/Ripple-65B-Preview/blob/da94cc013a8fda04a502f0a185083c7099ea06c1/README.md)）
	- 量化来源：提交 `df550a7178a87bbfb6ee2bdbd37603c553d536ee`，SHA-256 `1d7a551c5da46849db5f851e372b5b6b4af1fa1253e6a1e854cd7e77c37e9df6`（出处已验证：`True`）

	### 文件哈希

	\| 文件 \| 字节数 \| SHA-256 \| 量化 \|
	\|---\|---:\|---\|---\|
	\| `Sakura-Ripple-65B-Preview-Q2_K.gguf` \| 23747659040 \| `60b79fc65989267fcde15035ed01bb64a3915864e6e30d85159c5c1db066cc20` \| Q2_K \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| 31019943200 \| `fdf6815428dd2ed8ebf5d542525ca337bd1057d10d94bb74561736e24831574c` \| Q3_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| 38798820640 \| `7f5a149ac052e8125745d4f4b369c5e82c7ed8b43a5da1cd2b2a68fd6440f197` \| Q4_K_M \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| 45757433120 \| `a606978bedce45a40627773c7bd442d3c9d53c673e5c268d354bbb639950ffe4` \| 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-Ripple-65B-Preview-Q2_K.gguf` \| F32: 561, Q2_K: 401, Q3_K: 160, Q4_K: 80, Q6_K: 1 \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| F32: 561, Q3_K: 401, Q4_K: 233, Q5_K: 7, Q6_K: 1 \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| F32: 561, Q4_K: 561, Q6_K: 81 \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| F32: 561, Q5_K: 561, Q6_K: 81 \|

	### 测试版本

	\| 文件 \| llama.cpp 版本 \| 可执行文件 SHA-256 \|
	\|---\|---\|---\|
	\| `Sakura-Ripple-65B-Preview-Q2_K.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q3_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q4_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|
	\| `Sakura-Ripple-65B-Preview-Q5_K_M.gguf` \| `b12189-80d2c0c2c` \| `d6cc36e5fdc2eedd6e052dc7eb1e29d21065331e37e31f57ab2ca4720635dba2` \|

	</details>

	## 6. 许可证信息

	源许可证：`apache-2.0`。apache-2.0 allows redistribution with attribution
	<!-- hfqa-zh-end -->

	完整独立的中文版：[README_zh.md](README_zh.md)

	</details>