Instructions to use sriyuth/ari-paraphraser-1.5b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use sriyuth/ari-paraphraser-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 sriyuth/ari-paraphraser-1.5b
# Run inference directly in the terminal:
llama cli -hf sriyuth/ari-paraphraser-1.5b

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf sriyuth/ari-paraphraser-1.5b
# Run inference directly in the terminal:
llama cli -hf sriyuth/ari-paraphraser-1.5b

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 sriyuth/ari-paraphraser-1.5b
# Run inference directly in the terminal:
./llama-cli -hf sriyuth/ari-paraphraser-1.5b

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 sriyuth/ari-paraphraser-1.5b
# Run inference directly in the terminal:
./build/bin/llama-cli -hf sriyuth/ari-paraphraser-1.5b

Use Docker

docker model run hf.co/sriyuth/ari-paraphraser-1.5b

LM Studio
Jan

vLLM

How to use sriyuth/ari-paraphraser-1.5b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sriyuth/ari-paraphraser-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": "sriyuth/ari-paraphraser-1.5b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/sriyuth/ari-paraphraser-1.5b

Ollama
How to use sriyuth/ari-paraphraser-1.5b with Ollama:
```
ollama run hf.co/sriyuth/ari-paraphraser-1.5b
```
Unsloth Desktop

How to use sriyuth/ari-paraphraser-1.5b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf sriyuth/ari-paraphraser-1.5b

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": "sriyuth/ari-paraphraser-1.5b"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use sriyuth/ari-paraphraser-1.5b with Docker Model Runner:
```
docker model run hf.co/sriyuth/ari-paraphraser-1.5b
```

Lemonade

How to use sriyuth/ari-paraphraser-1.5b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull sriyuth/ari-paraphraser-1.5b

Run and chat with the model

lemonade run user.ari-paraphraser-1.5b-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use sriyuth/ari-paraphraser-1.5b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf sriyuth/ari-paraphraser-1.5b

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 sriyuth/ari-paraphraser-1.5b

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use sriyuth/ari-paraphraser-1.5b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf sriyuth/ari-paraphraser-1.5b

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 "sriyuth/ari-paraphraser-1.5b" \
  --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"

Ari Paraphraser 1.5B (RewriteBot) — v2.0 Production Release

Ari Paraphraser 1.5B is an ultra-fast, production-grade text paraphrasing and rewriting engine fine-tuned and aligned with Direct Preference Optimization (DPO). It delivers QuillBot-grade stylistic versatility, zero semantic drift, and instant sub-2-second inference on standard laptop CPUs.

🌟 What's New in This Release (Loss: 0.6136)

PAWS Integration (Google Research): Trained on Adversarial Paraphrase pairs to permanently eliminate cross-entity, subject, and gender swaps across clauses.
Strict English ASCII Hygiene: Fully eliminates foreign glyph and training artifact leakages (zero Thai suffix leakage).
1:1 Length Ratio Calibration: Short inputs stay concise without runaway hallucination, while long multi-paragraph documents (up to 1,024 tokens) generate with full 100% coverage.
5 Stylistic Modes: Standard, Fluency, Academic, Creative, and Concise.
Optimized for CPU Inference: Generates at 15–17 tokens/s on standard Intel/AMD consumer CPUs using AVX2.

📊 Benchmark Scorecard (100% Pass)

Benchmark Test Case	In $\to$ Out Words	Ratio	Entity Preservation	Status
Cancer Fertility Benchmark (143 words)	$143 \to 134$	0.94	100% (Boys $\to$ Sperm/Testicular, Women $\to$ Oocytes)	PASS
PAWS Medical Swap Inversion	$14 \to 15$	1.07	100% (Girl $\to$ Appendectomy, Boy $\to$ Physical Exam)	PASS
PAWS Route/Flight Mapping	$18 \to 18$	1.00	100% (Flight 340 late, Flight 112 on schedule)	PASS
Short Sentence 1:1 Calibration	$7 \to 8$	1.14	100% (Zero paragraph bloat)	PASS
Academic Mode Tone Shift	$17 \to 16$	0.94	100% (Formal vocabulary enhancement)	PASS
Concise / Shorten Mode	$23 \to 16$	0.70	100% (Wordy filler removed)	PASS

🎯 Supported Modes & Prompt Formats

Prefix your prompt with any of the following modes:

Mode	Prefix Example	Purpose
Standard	`[Mode: Standard] <text>`	Natural stylistic rewrite preserving exact meaning.
Fluency	`[Mode: Fluency] <text>`	Corrects irregular verbs, plurals, idioms, and sentence rhythm.
Academic	`[Mode: Academic] <text>`	Scholarly, formal, and objective research tone.
Creative	`[Mode: Creative] <text>`	Expressive vocabulary with varied clause dynamics.
Concise	`[Mode: Concise] <text>`	Punchy, condensed summary eliminating filler words.

🚀 Running with Ollama

Download ari_paraphraser_1.5b.gguf and the Modelfile into the same folder.
Build the local model:

ollama create ari-paraphraser -f Modelfile

Run directly from the command line:

ollama run ari-paraphraser "The boy underwent a routine physical examination while the girl received an emergency appendectomy."

Or run via Hugging Face:

ollama run hf.co/sriyuth/ari-paraphraser-1.5b

💻 Running via Python (llama-cpp-python)

from llama_cpp import Llama

llm = Llama(
    model_path="ari_paraphraser_1.5b.gguf",
    n_ctx=2048,
    n_threads=4
)

prompt = """<|im_start|>system
You are RewriteBot (ARI-1.5B), an expert editorial paraphrasing engine. Paraphrase the entire input text with complete coverage, authentic sentence variety, and strict factual preservation. Never swap, mix, or reassign subjects, genders, or medical procedures across clauses. Output ONLY the rewritten text.<|im_end|>
<|im_start|>user
She finished the test ahead of time.<|im_end|>
<|im_start|>assistant
"""

response = llm(prompt, max_tokens=256, temperature=0.3, stop=["<|im_end|>", "<|endoftext|>"])
print(response['choices'][0]['text'].strip())

Downloads last month: 1,820

Safetensors

Model size

2B params

Tensor type

F16

Model tree for sriyuth/ari-paraphraser-1.5b

Base model

Qwen/Qwen2.5-1.5B

Quantized

(83)

this model