Instructions to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF: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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF: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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

Use Docker

docker model run hf.co/bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

Ollama
How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with Ollama:
```
ollama run hf.co/bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
```

Unsloth Studio

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF to start chatting

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

Run Hermes

hermes

OpenClaw new

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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 "bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-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"

Docker Model Runner
How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with Docker Model Runner:
```
docker model run hf.co/bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M
```

Lemonade

How to use bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull bartowski/Kwaipilot_KAT-Coder-V2.5-Dev-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Kwaipilot_KAT-Coder-V2.5-Dev-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

KAT-Coder-V2.5-Dev vs Qwen 3.6-35B-A3B

by Kirin1997 - opened 5 days ago

Discussion

Kirin1997

5 days ago

Kat vs stock Qwen — my mini benchmark

I ran kat against stock Qwen on a custom 6-task suite (miniSkillbench): constraint retention, dependency graphs, state machines, decoy bugfixes — it's not exactly syntax trivia, but rather reasoning depth and procedural reliability. Pi agent was used in all the tests

Additionally, the repo also includes runs on Composer 2.5, I ran it just because I felt like it)

Takeaway: on these tests, kat consistently looked genuinely stronger than stock Qwen. Recommended temperature 1.0 performed worse than base 0.6 in this benchmark. KAT's quantisation is slightly worse than stock's, but that is rather minus to the stock

Caveats (important):
I wrote the tasks and graders myself;
very few runs: 2× kat, 1× stock Qwen;
tasks target intellectual depth, no API/syntax knowledge check

The repo is open — reproduce it and push back if you disagree

Kirin1997

5 days ago

•

edited 5 days ago

What's really interesting here - KAT was able to match Composer 2.5 in the raw reasoning and details attention in this test, but it wrote obviously worse code leaving alone time it took to complete the tasks

Kirin1997 changed discussion status to closed 5 days ago

Kirin1997 changed discussion status to open 5 days ago

NeoHuggingF

4 days ago

Interesting, thanks for sharing the results. Could you please also share the quantization used (I assume you have use GGUFs, right) and the exact serving parameters (llama.cpp?). Thank you. Regards

hawken93

2 days ago

What presence-penalty do you test with? The upstream model says 1.5 but I'm sceptical

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment