Instructions to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated",
	filename="Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated.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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated 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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M

Use Docker

docker model run hf.co/Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M

LM Studio
Jan
Ollama
How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with Ollama:
```
ollama run hf.co/Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
```

Unsloth Studio

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated 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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated 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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated to start chatting

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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": "Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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 "Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated: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 Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with Docker Model Runner:
```
docker model run hf.co/Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M
```

Lemonade

How to use Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Chungulus/Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated:Q4_K_M

Run and chat with the model

lemonade run user.Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated-Q4_K_M

List all available models

lemonade list

Agents-A1-Q4_K_M-imatrix-gguf-fable5-calibrated

File size: 1,925 Bytes

93cce49

# Agents-A1 imatrix quality evaluation

This is a local sanity benchmark, not a substitute for MMLU, HellaSwag, or human review.
No files were uploaded.

## Setup

- Accuracy benchmark: `eval/quality_multiple_choice.bin`
- KL holdout text: `eval/kl_holdout.txt`
- KL context/chunks: 128 ctx, 2 chunks
- Backend: native llama.cpp, CPU-only (`-dev none -ngl 0`)

## Accuracy Retention

F16 baseline accuracy: 89.5833%

| Variant | Accuracy | Retention vs F16 | Random chance | Status |
|---|---:|---:|---:|---|
| A1-Q2_K-imatrix | 87.5000% | 97.6745% | 25.0000% | passed |
| A1-IQ3_M-imatrix | 89.5833% | 100.0000% | 25.0000% | passed |
| A1-Q3_K_M-imatrix | 89.5833% | 100.0000% | 25.0000% | passed |
| A1-Q4_K_M-imatrix | 87.5000% | 97.6745% | 25.0000% | passed |
| A1-IQ4_XS-imatrix | 87.5000% | 97.6745% | 25.0000% | passed |

## KL Divergence

Baseline logits file: `/Users/oz/Documents/AgentsA1 Quantization/a1-local-quant-pipeline/benchmark_results/quality/baseline_ctx128_chunks2.kld`
Baseline PPL on KL holdout: 13.0194

| Variant | Mean KLD | PPL ratio | PPL delta | Same top p | RMS delta p | Status |
|---|---:|---:|---:|---:|---:|---|
| A1-Q2_K-imatrix | 0.128242 | 1.158739 | 2.0667 | 81.7460% | 8.8070% | passed |
| A1-IQ3_M-imatrix | 0.045312 | 1.001578 | 0.0205 | 92.0630% | 5.0240% | passed |
| A1-Q3_K_M-imatrix | 0.046980 | 0.984125 | -0.2067 | 88.0950% | 5.2290% | passed |
| A1-Q4_K_M-imatrix | 0.015182 | 0.986453 | -0.1764 | 93.6510% | 2.7470% | passed |
| A1-IQ4_XS-imatrix | 0.020185 | 0.998199 | -0.0234 | 92.0630% | 3.1110% | passed |

## Notes

- Accuracy is a local multiple-choice benchmark generated in `eval/quality_multiple_choice.json`.
- Retention is quant accuracy divided by F16 baseline accuracy on the same local benchmark.
- KL divergence compares quant logit distributions to the F16 no-MTP GGUF baseline on `eval/kl_holdout.txt`.
- Smaller KLD is better; higher same-top-p is better.