Instructions to use AesSedai/Kimi-K2.7-Code-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AesSedai/Kimi-K2.7-Code-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="AesSedai/Kimi-K2.7-Code-GGUF",
	filename="IQ2_S/Kimi-K2.7-Code-IQ2_S-00001-of-00008.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use AesSedai/Kimi-K2.7-Code-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
# Run inference directly in the terminal:
llama-cli -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
# Run inference directly in the terminal:
llama-cli -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

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 AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
# Run inference directly in the terminal:
./llama-cli -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

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 AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
# Run inference directly in the terminal:
./build/bin/llama-cli -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

Use Docker

docker model run hf.co/AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

LM Studio
Jan
Ollama
How to use AesSedai/Kimi-K2.7-Code-GGUF with Ollama:
```
ollama run hf.co/AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
```

Unsloth Studio

How to use AesSedai/Kimi-K2.7-Code-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 AesSedai/Kimi-K2.7-Code-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 AesSedai/Kimi-K2.7-Code-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for AesSedai/Kimi-K2.7-Code-GGUF to start chatting

How to use AesSedai/Kimi-K2.7-Code-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

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": "AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use AesSedai/Kimi-K2.7-Code-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

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 AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use AesSedai/Kimi-K2.7-Code-GGUF with Docker Model Runner:
```
docker model run hf.co/AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S
```

Lemonade

How to use AesSedai/Kimi-K2.7-Code-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull AesSedai/Kimi-K2.7-Code-GGUF:IQ2_S

Run and chat with the model

lemonade run user.Kimi-K2.7-Code-GGUF-IQ2_S

List all available models

lemonade list

Thanks for the quick quants! Reccomended mmproj?

by phakio - opened 4 days ago

Discussion

phakio

4 days ago

Time to begin another round of testing! Thanks for the quick quants!

Curious if anyone has data on the various mmproj quants and whether running q8 results in real world noticeable decline in accuracy? I mean I know that there is some loss when going from F32 to F16 to Q8, but I never really see data that supports how noticeable the change in bits is.

I'm going to go with Q3_K_L and Q8 mmproj for now, but just placing this here for anyone else to post data they may find! Once I get this model dialed in I may set up some tests to compare the various mmproj files ~

AesSedai

Owner 3 days ago

•

edited 3 days ago

Hi @phakio

I realized that the Q3_K_L wasn't imatrixed, I've redone it and updated the quant. New vs old:

Version	Quant	Size	Mixture	PPL	1-(Mean PPL(Q)/PPL(base))	KLD
OLD	Q3_K_L	459.94 GiB (3.85 BPW)	Q8_0 / Q3_K / Q3_K / Q4_0	2.118293 ± 0.009633	+5.5655%	0.072403 ± 0.000616
NEW	Q3_K_L	459.94 GiB (3.85 BPW)	Q8_0 / Q3_K / Q3_K / Q4_0	2.085207 ± 0.009398	+3.9167%	0.063007 ± 0.000544

For the mmproj quants, I haven't done an eval to compare them before. I usually stick to the F16 or BF16 though, I know that image encoder quantization is pretty drastically sharp when it does start to decline. I'd be interested in any results you have to share.

ChuckMcSneed

3 days ago

I found that when going from F32 to any quant mmproj gets badly damaged. I would say use f32 if you can, otherwise use the biggest one available.

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