Instructions to use richardlian/iolai-2026-4b-wvote with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use richardlian/iolai-2026-4b-wvote 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 richardlian/iolai-2026-4b-wvote:Q8_0
# Run inference directly in the terminal:
llama cli -hf richardlian/iolai-2026-4b-wvote:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf richardlian/iolai-2026-4b-wvote:Q8_0
# Run inference directly in the terminal:
llama cli -hf richardlian/iolai-2026-4b-wvote:Q8_0

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 richardlian/iolai-2026-4b-wvote:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf richardlian/iolai-2026-4b-wvote:Q8_0

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 richardlian/iolai-2026-4b-wvote:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf richardlian/iolai-2026-4b-wvote:Q8_0

Use Docker

docker model run hf.co/richardlian/iolai-2026-4b-wvote:Q8_0

LM Studio
Jan
Ollama
How to use richardlian/iolai-2026-4b-wvote with Ollama:
```
ollama run hf.co/richardlian/iolai-2026-4b-wvote:Q8_0
```
Unsloth Desktop

How to use richardlian/iolai-2026-4b-wvote with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf richardlian/iolai-2026-4b-wvote:Q8_0

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": "richardlian/iolai-2026-4b-wvote:Q8_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use richardlian/iolai-2026-4b-wvote with Docker Model Runner:
```
docker model run hf.co/richardlian/iolai-2026-4b-wvote:Q8_0
```

Lemonade

How to use richardlian/iolai-2026-4b-wvote with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull richardlian/iolai-2026-4b-wvote:Q8_0

Run and chat with the model

lemonade run user.iolai-2026-4b-wvote-Q8_0

List all available models

lemonade list

Hermes Agent

How to use richardlian/iolai-2026-4b-wvote with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf richardlian/iolai-2026-4b-wvote:Q8_0

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 richardlian/iolai-2026-4b-wvote:Q8_0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use richardlian/iolai-2026-4b-wvote with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf richardlian/iolai-2026-4b-wvote:Q8_0

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 "richardlian/iolai-2026-4b-wvote:Q8_0" \
  --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"

iolai-2026-4b-wvote / script.py

richardlian

Upload folder using huggingface_hub

010384c verified about 2 months ago

Raw

History Blame Contribute Delete

1.36 kB

	"""IOL-AI 2026 submission: q4b-wvote-v1.

	E33 confidence-weighted votes: 12 samples, per-item raw-model logprob weighting (T=0.35).

	Recipe (every knob is read by v2/config.py::Config.from_env):
	BACKEND=parallel
	FEWSHOT=0
	LLAMA_CTX=16384
	MAX_NEW=700
	N_SAMPLES=12
	PROMPT_MODE=direct
	REFINE_ROUNDS=0
	SLOTS=16
	THINK=0
	TRANSCRIPT=
	VOTE_LP_T=0.35
	VOTE_TEMP=1.0
	VOTE_TOP_P=0.95
	VOTE_WEIGHT=full
	"""

	import os
	import sys

	sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
	HERE = os.path.dirname(os.path.abspath(__file__))
	os.environ.setdefault("GGUF_PATH", os.path.join(HERE, "model", "model-Q8_0.gguf"))
	os.environ.setdefault("LLAMA_PARALLEL", os.path.join(HERE, "bin", "llama-iolbatch"))
	os.environ.setdefault("BACKEND", "parallel")
	os.environ.setdefault("SLOTS", "16")
	os.environ.setdefault("LLAMA_CTX", "16384")
	os.environ.setdefault("VOTE_TEMP", "1.0")
	os.environ.setdefault("VOTE_TOP_P", "0.95")
	os.environ.setdefault("PROMPT_MODE", "direct")
	os.environ.setdefault("N_SAMPLES", "12")
	os.environ.setdefault("MAX_NEW", "700")
	os.environ.setdefault("FEWSHOT", "0")
	os.environ.setdefault("THINK", "0")
	os.environ.setdefault("REFINE_ROUNDS", "0")
	os.environ.setdefault("TRANSCRIPT", "")
	os.environ.setdefault("VOTE_WEIGHT", "full")
	os.environ.setdefault("VOTE_LP_T", "0.35")

	from v2.script import main # noqa: E402

	main()