Instructions to use purplesquirrelnetworks/purple-squirrel-r1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use purplesquirrelnetworks/purple-squirrel-r1 with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("purplesquirrelnetworks/purple-squirrel-r1")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

llama-cpp-python

How to use purplesquirrelnetworks/purple-squirrel-r1 with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="purplesquirrelnetworks/purple-squirrel-r1",
	filename="purple-squirrel-r1-f16.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use purplesquirrelnetworks/purple-squirrel-r1 with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf purplesquirrelnetworks/purple-squirrel-r1:F16
# Run inference directly in the terminal:
llama-cli -hf purplesquirrelnetworks/purple-squirrel-r1:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf purplesquirrelnetworks/purple-squirrel-r1:F16
# Run inference directly in the terminal:
llama-cli -hf purplesquirrelnetworks/purple-squirrel-r1:F16

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 purplesquirrelnetworks/purple-squirrel-r1:F16
# Run inference directly in the terminal:
./llama-cli -hf purplesquirrelnetworks/purple-squirrel-r1:F16

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 purplesquirrelnetworks/purple-squirrel-r1:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf purplesquirrelnetworks/purple-squirrel-r1:F16

Use Docker

docker model run hf.co/purplesquirrelnetworks/purple-squirrel-r1:F16

LM Studio
Jan

vLLM

How to use purplesquirrelnetworks/purple-squirrel-r1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "purplesquirrelnetworks/purple-squirrel-r1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "purplesquirrelnetworks/purple-squirrel-r1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/purplesquirrelnetworks/purple-squirrel-r1:F16

Ollama
How to use purplesquirrelnetworks/purple-squirrel-r1 with Ollama:
```
ollama run hf.co/purplesquirrelnetworks/purple-squirrel-r1:F16
```

Unsloth Studio new

How to use purplesquirrelnetworks/purple-squirrel-r1 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 purplesquirrelnetworks/purple-squirrel-r1 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 purplesquirrelnetworks/purple-squirrel-r1 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for purplesquirrelnetworks/purple-squirrel-r1 to start chatting

MLX LM

How to use purplesquirrelnetworks/purple-squirrel-r1 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "purplesquirrelnetworks/purple-squirrel-r1"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "purplesquirrelnetworks/purple-squirrel-r1"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "purplesquirrelnetworks/purple-squirrel-r1",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Docker Model Runner
How to use purplesquirrelnetworks/purple-squirrel-r1 with Docker Model Runner:
```
docker model run hf.co/purplesquirrelnetworks/purple-squirrel-r1:F16
```

Lemonade

How to use purplesquirrelnetworks/purple-squirrel-r1 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull purplesquirrelnetworks/purple-squirrel-r1:F16

Run and chat with the model

lemonade run user.purple-squirrel-r1-F16

List all available models

lemonade list

Purple Squirrel R1

Fine-tuned DeepSeek-R1-Distill-Llama-8B for Purple Squirrel AI Platform

Related Resources

Resource	Link
Research Paper	AIDP Neural Cloud: Distributed LLM Inference
Research Paper	AIDP Video Forge: GPU-Accelerated Video Processing
GGUF Version	purple-squirrel-r1-gguf
Multichain Edition	purple-squirrel-r1-multichain
Training Data	purple-squirrel-training
Multichain Training	multichain-day-training
LoRA Adapters	purple-squirrel-r1-multichain-lora
Coldstar Whitepaper	coldstar-whitepaper
Full Collection	Purple Squirrel AI

Model Details

Base Model: DeepSeek-R1-Distill-Llama-8B
Parameters: 8B
Context Length: 4096 tokens
Quantization: 4-bit NF4 (GGUF f16 available)
Specialization: Purple Squirrel AI platform operations

Research Papers

This model is deployed in the AIDP Neural Cloud distributed inference system and powers the AIDP Video Forge processing pipeline.

AIDP Neural Cloud — Distributed LLM Inference on Decentralized GPU Networks:

47% cost reduction vs OpenAI
28% faster latency (p50: 180ms vs 250ms)
50 req/s throughput with fault tolerance

AIDP Video Forge — GPU-Accelerated Video Processing:

10-20x faster encoding vs CPU
40-60% cost reduction vs centralized cloud
VMAF 95.8 quality score

Capabilities

Fine-tuned to excel at:

Video Analysis: AI-powered transcription and tagging
Blockchain Operations: Multi-chain NFT minting (Solana, Ethereum, Polygon)
Cloud Integration: OCI, AWS, IPFS storage operations
Video Editing: Professional workflow understanding
Platform Operations: Purple Squirrel feature guidance

Quick Start

Using Ollama

ollama pull purplesquirrelnetworks/purple-squirrel-r1
ollama run purplesquirrelnetworks/purple-squirrel-r1

Using Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "purplesquirrelnetworks/purple-squirrel-r1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

Via AIDP Neural Cloud API

import openai

client = openai.OpenAI(
    base_url="https://neural-cloud.aidp.store/v1",
    api_key="your-api-key"
)

response = client.chat.completions.create(
    model="purple-squirrel-r1",
    messages=[
        {"role": "user", "content": "Explain decentralized GPU compute"}
    ]
)
print(response.choices[0].message.content)

Additional Resources

Model Comparison — Side-by-side comparison of base DeepSeek-R1 vs Purple Squirrel R1 with example prompts and responses
Blog Post — Technical write-up covering training setup, data curation, results, and usage guide

Citation

If you use this model or the associated research, please cite:

@techreport{karsten2026neuralcloud,
  title={AIDP Neural Cloud: Distributed LLM Inference on Decentralized GPU Networks},
  author={Karsten, Matthew},
  institution={Purple Squirrel Networks},
  year={2026},
  month={February},
  url={https://huggingface.co/purplesquirrelnetworks/aidp-neural-cloud-paper}
}

@techreport{karsten2026videoforge,
  title={AIDP Video Forge: GPU-Accelerated Video Processing on Decentralized Compute Networks},
  author={Karsten, Matthew},
  institution={Purple Squirrel Networks},
  year={2026},
  month={February},
  url={https://huggingface.co/purplesquirrelnetworks/aidp-video-forge-paper}
}

Built by Purple Squirrel Networks

Downloads last month: 30

Safetensors

Model size

8B params

Tensor type

F32

MLX

Hardware compatibility

4-bit

Model tree for purplesquirrelnetworks/purple-squirrel-r1

Base model

deepseek-ai/DeepSeek-R1-Distill-Llama-8B

Quantized

(191)

this model

Dataset used to train purplesquirrelnetworks/purple-squirrel-r1

Space using purplesquirrelnetworks/purple-squirrel-r1 1

Collection including purplesquirrelnetworks/purple-squirrel-r1

Purple Squirrel AI — Models, Papers & Data

Collection

9 models, 3 papers, 3 datasets. Distributed AI, GPU video, multichain DeFi, Solana wallets. GGUF quants + LoRA + 1.3K training pairs. MIT. • 12 items • Updated Mar 2

Evaluation results

Cost Reduction vs OpenAI (%)
self-reported

47.000
p50 Latency (ms)
self-reported

180.000
Throughput (req/s)
self-reported

50.000
Encoding Speedup vs CPU (x)
self-reported

16.000
Cost Reduction vs Cloud (%)
self-reported

50.000
VMAF Quality Score
self-reported

95.800