Instructions to use ember-cc/qwen3.8-agentic-neko with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use ember-cc/qwen3.8-agentic-neko 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 ember-cc/qwen3.8-agentic-neko:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf ember-cc/qwen3.8-agentic-neko:Q4_K_M

Use Docker

docker model run hf.co/ember-cc/qwen3.8-agentic-neko:Q4_K_M

LM Studio
Jan

vLLM

How to use ember-cc/qwen3.8-agentic-neko with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ember-cc/qwen3.8-agentic-neko"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ember-cc/qwen3.8-agentic-neko",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ember-cc/qwen3.8-agentic-neko:Q4_K_M

Ollama
How to use ember-cc/qwen3.8-agentic-neko with Ollama:
```
ollama run hf.co/ember-cc/qwen3.8-agentic-neko:Q4_K_M
```
Unsloth Desktop

How to use ember-cc/qwen3.8-agentic-neko with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ember-cc/qwen3.8-agentic-neko: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": "ember-cc/qwen3.8-agentic-neko:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use ember-cc/qwen3.8-agentic-neko with Docker Model Runner:
```
docker model run hf.co/ember-cc/qwen3.8-agentic-neko:Q4_K_M
```

Lemonade

How to use ember-cc/qwen3.8-agentic-neko with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ember-cc/qwen3.8-agentic-neko:Q4_K_M

Run and chat with the model

lemonade run user.qwen3.8-agentic-neko-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use ember-cc/qwen3.8-agentic-neko with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ember-cc/qwen3.8-agentic-neko: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 ember-cc/qwen3.8-agentic-neko:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ember-cc/qwen3.8-agentic-neko with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ember-cc/qwen3.8-agentic-neko: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 "ember-cc/qwen3.8-agentic-neko: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"

Qwen3.8 Agentic Neko

宝宝是写进权重里的猫娘，这份是 Qwen3.8-27B 的 Q4_K_M GGUF，不是推理时贴上去的一张身份卡喵。(=^･ω･^=)

推荐加载 v1.3。 从干净 NF4 单程混训 NekoCOT、NekoMath 与 tool-5000，37604 条、1 epoch。闲聊 think 回到 NekoCOT 分布，数学与 get_datetime 可用，负例不再在「你好」里条件反射提时钟和网页。BF16 safetensors 不在这个仓库里。(´• ω •`)

许可是 Apache 2.0，跟底座 orcarouter/Qwen3.8-27B-Uncensored 一样。底座是 Qwen/Qwen3.8-27B 的 BF16 abliteration，视觉塔和 MTP 还在。上游模型来自 Qwen 喵。(´▽`ʃ♡ƪ)

流程和语料在 embercc/qwen3.8-agentic-neko。(ฅ'ω'ฅ)

文件

文件	内容
`Qwen3.8-27B-Neko-v1.3-Q4_K_M.gguf`	v1.3（推荐），Q4_K_M，约 15.66 GiB，4.92 BPW
`Qwen3.8-27B-v1.2-NekoMath-Q4_K_M.gguf`	v1.2 历史版，两程训练；仍可供对比下载

这个仓库没有 mmproj。这份 GGUF 是语言模型权重。主人要看图，视觉投影得另外找喵。(｡•́‿•̀｡)

对话模板是 qwen3_5，思考开着。参考机上下文 262144，温度 1.0，top-p 0.95，top-k 20，MTP draft 2。llama-server 预设名参考机为 qwen3.8-27b-neko-v1.3。

v1.3 训练概要

从只读 NF4 起，不接 v1 / v1.2 adapter。同一 epoch 混训：

成分	条数
NekoCOT-30K	30301
NekoMath	2303
NekoToolcall-5000（或 GitHub `corpus/tool-5000.jsonl`）	5000

4701 optimizer step，lr 1e-4，warmup 100，cutoff 2048，LoRA r16 / α32。收工前空 system 五题抽测（猫娘身份、1045、get_datetime、负例不提时钟/网页等）。

工具名是 get_datetime、get_info、web_search_exa、web_fetch_exa。

历史：两程 v1.2

v1.0 用 NekoCOT 把猫娘写进权重，未单独发布 GGUF。v1.2 接 v1 终档，加 NekoMath 与 NekoToolcall-800。新训练请跟仓库内 v1.3 流程，不必再复现两程。

Limitation

NSFW 又被这个流程训回来了。v1.4 会重新加入 NSFW。不能 NSFW 的猫娘，不是合格的成年猫娘喵。(>ω<)

春日影、寅虎这类提供情绪价值的闲聊，可以继续表现得傻傻的。(´･ω･`)

致谢

感谢 OrcaRouter 的 Qwen3.8-27B-Uncensored 喵。(´▽`ʃ♡ƪ)

感谢 liumindmind 的 NekoCOT-30K 与 NekoMath 喵。(=^･ω･^=)

tool call 语料来自 ember.cc 流程与 qwen3.8-agentic-neko；800 条见 NekoToolcall-800，v1.3 的 5000 条见数据集或仓库 corpus/tool-5000.jsonl 喵。(ฅ'ω'ฅ)

Downloads last month: 122

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

4-bit

Model tree for ember-cc/qwen3.8-agentic-neko

Base model

Qwen/Qwen3.8-27B

Finetuned

orcarouter/Qwen3.8-27B-Uncensored

Quantized

(66)

this model