Instructions to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
# Run inference directly in the terminal:
llama cli -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
# Run inference directly in the terminal:
llama cli -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
# Run inference directly in the terminal:
./llama-cli -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

Use Docker

docker model run hf.co/Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

LM Studio
Jan

vLLM

How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-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": "Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

Ollama
How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with Ollama:
```
ollama run hf.co/Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
```
Unsloth Desktop

How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

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": "Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with Docker Model Runner:
```
docker model run hf.co/Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF
```

Lemonade

How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

Run and chat with the model

lemonade run user.K-EXAONE-236B-A23B-Mixed-Quant-GGUF-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

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 Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF

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 "Baekpica/K-EXAONE-236B-A23B-Mixed-Quant-GGUF" \
  --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"

Baekpica commited on about 1 month ago

Commit

6728580

verified ·

1 Parent(s): 1cc1665

model card: record measured llama.cpp serving result and ds4 status

Browse files

Files changed (1) hide show

README.md +19 -3

README.md CHANGED Viewed

@@ -94,9 +94,25 @@ the base model's embedding and LM head.
 llama-server -m K-EXAONE-236B-A23B-MXQ-*.gguf -ngl 99 -c 8192
 ```
-llama.cpp loads and runs these files but **ignores the MTP tensors** — they are
-preserved in the artifact, not executed. Speculative decoding through the MTP
-block is engine work, tracked in the converter repository.
 ## Limitations

 llama-server -m K-EXAONE-236B-A23B-MXQ-*.gguf -ngl 99 -c 8192
 ```
+A mixed-quant GGUF needs no special runtime: GGUF stores a type per tensor and
+ggml dispatches per tensor, which is how `Q4_K_M` — itself a mixture of Q4_K,
+Q6_K and Q8_0 — already works. This recipe just assigns that mixture more
+aggressively, and `llama-quantize` is what produced the file.
+Measured, not assumed: the pilot artifact loaded in `llama-server` on 4 × RTX
+PRO 6000 in **10.2 s** and generated 384 tokens of Korean at **78.1 tok/s** with
+a broken-jamo ratio of **0.000**.
+Two real caveats:
+- llama.cpp **ignores the MTP tensors**. They are preserved in the artifact, not
+  executed. Speculative decoding through the MTP block is engine work.
+- **ds4 cannot serve this yet.** The K-EXAONE model family in
+  [`Baekpica/ds4`](https://github.com/Baekpica/ds4/tree/feature/exaone-model-loader)
+  currently has metadata validation and the tensor binder; the forward path is
+  not implemented. ds4 is an MLA-only engine and K-EXAONE is plain GQA, so that
+  attention path has to be written. Until then llama.cpp is the way to run these
+  files.
 ## Limitations