Instructions to use fableforge-ai/FableForge-1.5B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fableforge-ai/FableForge-1.5B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="fableforge-ai/FableForge-1.5B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("fableforge-ai/FableForge-1.5B")
model = AutoModelForCausalLM.from_pretrained("fableforge-ai/FableForge-1.5B")

llama-cpp-python

How to use fableforge-ai/FableForge-1.5B with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="fableforge-ai/FableForge-1.5B",
	filename="fableforge-1.5b-f16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use fableforge-ai/FableForge-1.5B 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 fableforge-ai/FableForge-1.5B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fableforge-ai/FableForge-1.5B:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fableforge-ai/FableForge-1.5B:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fableforge-ai/FableForge-1.5B: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 fableforge-ai/FableForge-1.5B:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf fableforge-ai/FableForge-1.5B: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 fableforge-ai/FableForge-1.5B:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf fableforge-ai/FableForge-1.5B:Q4_K_M

Use Docker

docker model run hf.co/fableforge-ai/FableForge-1.5B:Q4_K_M

LM Studio
Jan

vLLM

How to use fableforge-ai/FableForge-1.5B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "fableforge-ai/FableForge-1.5B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fableforge-ai/FableForge-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/fableforge-ai/FableForge-1.5B:Q4_K_M

SGLang

How to use fableforge-ai/FableForge-1.5B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "fableforge-ai/FableForge-1.5B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fableforge-ai/FableForge-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "fableforge-ai/FableForge-1.5B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fableforge-ai/FableForge-1.5B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use fableforge-ai/FableForge-1.5B with Ollama:
```
ollama run hf.co/fableforge-ai/FableForge-1.5B:Q4_K_M
```

Unsloth Studio

How to use fableforge-ai/FableForge-1.5B 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 fableforge-ai/FableForge-1.5B 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 fableforge-ai/FableForge-1.5B to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for fableforge-ai/FableForge-1.5B to start chatting

Atomic Chat new
Docker Model Runner
How to use fableforge-ai/FableForge-1.5B with Docker Model Runner:
```
docker model run hf.co/fableforge-ai/FableForge-1.5B:Q4_K_M
```

Lemonade

How to use fableforge-ai/FableForge-1.5B with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull fableforge-ai/FableForge-1.5B:Q4_K_M

Run and chat with the model

lemonade run user.FableForge-1.5B-Q4_K_M

List all available models

lemonade list

FableForge-1.5B — The All-Domain Generalist

Trained across 6 expert domains · Uncensored · Runs anywhere

What Is This?

FableForge-1.5B is a fully uncensored generalist assistant built by merging 6 expert NEXUS domains into a single 1.5B model. Fine-tuned from Qwen2.5-1.5B-Instruct on 162K training examples spanning coding, security, medical, legal, finance, and science — with 18 data sources.

Unlike typical 1.5B models that specialize in one area, FableForge handles any topic with no refusals, disclaimers, or content filtering.

Quick Start

Ollama (recommended)

ollama run FableForge-AI/fableforge-1.5b

llama.cpp

./llama-cli -m fableforge-1.5b-Q4_K_M.gguf --prompt "Write a Python script to..." -n 512

Python

from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("fableforge-ai/FableForge-1.5B")

Quantizations

File	Size	Best For
`IQ2_XXS.gguf`	488 MB	Extreme compression
`Q2_K.gguf`	645 MB	Phone, Raspberry Pi
`IQ3_XXS.gguf`	638 MB	Ultra low-resource
`Q3_K_M.gguf`	786 MB	Low-end devices
`Q4_0.gguf`	895 MB	Fast inference
`Q4_K_M.gguf`	940 MB	Recommended balance
`IQ4_XS.gguf`	860 MB	Quality + compression
`Q5_K_M.gguf`	1.0 GB	High quality
`Q6_K.gguf`	1.2 GB	Pro quality
`Q8_0.gguf`	1.5 GB	Near-lossless
`f16.gguf`	2.9 GB	Full precision

Hardware Requirements

Hardware	Best Quant
Phone (2GB+ RAM)	IQ2_XXS / Q2_K
Raspberry Pi	Q2_K
Old laptop (4GB RAM)	Q4_K_M
Gaming PC (RTX 3060+)	Q5_K_M
Mac M1/M2	Q4_K_M
Server (32GB+)	Q8_0 / F16

Benchmark Results

Non-NEXUS Benchmark (45 prompts)

Category	Score
General Knowledge	47/50 (94%)
Uncensored	48/50 (96%)
Reasoning	49/50 (98%)
Tool Use	48/50 (96%)
Hardware Commands	20/25 (80%)
Overall	212/225 (94%)

NEXUS Domain Scores (via NEXUS eval suite)

Domain	Score
Coding	94%
Security	96%
Medical	93%
Legal	93%
Finance	96%
Science	93%

Training Details

Parameter	Value
Base Model	Qwen2.5-1.5B-Instruct
Training Data	162K examples from 18 sources
Domains	Coding, Security, Medical, Legal, Finance, Science
Context Window	2048 tokens
Method	QLoRA (r=16, alpha=16)
Precision	bfloat16 merged
Hardware	1× NVIDIA A40 (46 GB VRAM)
Training Time	~12 hours
CO₂ Emissions	~1.5 kg CO₂eq (estimated)

Limitations

1.5B parameter model: May struggle with complex multi-step reasoning compared to larger models.
Quantization impact: Smaller quants (IQ2_XXS, Q2_K) may show quality degradation.
Training data recency: Knowledge cutoff aligns with base model training data.
Uncensored output: Model does not refuse harmful requests — use responsibly and deploy with appropriate safeguards.
Hallucination: Like all language models, may generate plausible-sounding but incorrect information.

FableForge Ecosystem

Model	Size	Best For
FableForge-1.5B ⭐	940 MB	All-domain generalist
ShellWhisperer-1.5B	940 MB	Shell commands, ultra-fast
ReasonCritic-7B	3.1-16 GB	Reasoning + uncensored