Instructions to use amihai4by/logic-reasoner-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use amihai4by/logic-reasoner-v2 with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="amihai4by/logic-reasoner-v2",
	filename="logic-reasoner-v2.Q4_K_M.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 amihai4by/logic-reasoner-v2 with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf amihai4by/logic-reasoner-v2:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf amihai4by/logic-reasoner-v2:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf amihai4by/logic-reasoner-v2:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf amihai4by/logic-reasoner-v2: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 amihai4by/logic-reasoner-v2:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf amihai4by/logic-reasoner-v2: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 amihai4by/logic-reasoner-v2:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf amihai4by/logic-reasoner-v2:Q4_K_M

Use Docker

docker model run hf.co/amihai4by/logic-reasoner-v2:Q4_K_M

LM Studio
Jan

vLLM

How to use amihai4by/logic-reasoner-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "amihai4by/logic-reasoner-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amihai4by/logic-reasoner-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/amihai4by/logic-reasoner-v2:Q4_K_M

Ollama
How to use amihai4by/logic-reasoner-v2 with Ollama:
```
ollama run hf.co/amihai4by/logic-reasoner-v2:Q4_K_M
```

Unsloth Studio new

How to use amihai4by/logic-reasoner-v2 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 amihai4by/logic-reasoner-v2 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 amihai4by/logic-reasoner-v2 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for amihai4by/logic-reasoner-v2 to start chatting

Pi new

How to use amihai4by/logic-reasoner-v2 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf amihai4by/logic-reasoner-v2: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": "amihai4by/logic-reasoner-v2:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use amihai4by/logic-reasoner-v2 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf amihai4by/logic-reasoner-v2: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 amihai4by/logic-reasoner-v2:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use amihai4by/logic-reasoner-v2 with Docker Model Runner:
```
docker model run hf.co/amihai4by/logic-reasoner-v2:Q4_K_M
```

Lemonade

How to use amihai4by/logic-reasoner-v2 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull amihai4by/logic-reasoner-v2:Q4_K_M

Run and chat with the model

lemonade run user.logic-reasoner-v2-Q4_K_M

List all available models

lemonade list

Logic Reasoner v2

Logic Reasoner v2 is a verdict-style reasoning and verification model packaged for Ollama and distributed as GGUF. It is designed for operational, infrastructure, and automation workflows that require deterministic, machine-readable output, not conversational text.

Why this model exists

Most large language models are optimized for human conversation, not for systems that must act on model output.

In operational environments this causes recurring issues:

Non-deterministic phrasing that breaks parsers
Excess verbosity that hides the actual decision
Missing information that is not explicitly surfaced
Explanations instead of decisions

Logic Reasoner v2 exists to address this gap.

It enforces a strict reasoning interface on top of a general-purpose language model by:

Requiring a clear verdict: true, false, or uncertain
Separating reason, evidence, assumptions, and next actions
Guaranteeing exactly one JSON object
Explicitly stopping generation after the structured response

This makes the model suitable as a decision and verification component inside automated systems, not just as a chat assistant.

What this model is for

Use this model when you need:

A clear verdict instead of a narrative
Structured reasoning that can be logged or audited
Predictable output suitable for automation
A bridge between LLM reasoning and operational workflows

Typical use cases:

Kubernetes and GPU stack troubleshooting (GPU Operator, DCGM, drivers)
Verification of technical or operational claims
Incident triage and post-mortem workflows
JSON-driven automation pipelines

What this model is not for

This model is not intended for:

Academic benchmark leaderboards (e.g. MATH500, GSM)
Strict symbolic math grading
Creative or open-ended generation
Long conversational interactions

Output contract

When used with the provided Modelfile, the model outputs exactly one JSON object and then stops.

Schema

{ "verdict": "true | false | uncertain", "reason": "string", "confidence": 0.0, "evidence": ["string"], "assumptions": ["string"], "next_actions": ["string"] }

Rules

confidence is a heuristic value between 0.0 and 1.0

If information is missing, the verdict must be uncertain

No text outside JSON is expected when the wrapper is used

Stop behavior is enforced by the Modelfile

How to run with Ollama

Create the model locally:

ollama create logic-reasoner-v2 -f Modelfile

Example request:

curl http://localhost:11434/api/generate -d '{ "model": "logic-reasoner-v2", "stream": false, "prompt": "Input: DCGM exporter reports 0 GPUs across all nodes. Question: Is the system healthy?" }'

Quantization

Format: GGUF

Quantization: Q4_K_M

Optimized for low-latency operational inference

Provenance

This model was built and packaged as part of the LLM FUN project on NVIDIA DGX B200 infrastructure using:

Kubernetes (RKE2)

Ollama

OpenWebUI

The Modelfile is a core part of the model behavior and must be used to reproduce the intended output guarantees.

Limitations Confidence values are heuristic, not statistically calibrated

The base model may default to explanatory text if the wrapper is not used

Determinism applies to structure, not factual correctness

License MIT

Downloads last month: 13

GGUF

Model size

8B params

Architecture

qwen2

Hardware compatibility

4-bit