Instructions to use edoggy/poc-qwen2.5-coder-7b-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use edoggy/poc-qwen2.5-coder-7b-v1 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 edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Use Docker

docker model run hf.co/edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use edoggy/poc-qwen2.5-coder-7b-v1 with Ollama:
```
ollama run hf.co/edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M
```
Unsloth Desktop

How to use edoggy/poc-qwen2.5-coder-7b-v1 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

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": "edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use edoggy/poc-qwen2.5-coder-7b-v1 with Docker Model Runner:
```
docker model run hf.co/edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M
```

Lemonade

How to use edoggy/poc-qwen2.5-coder-7b-v1 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Run and chat with the model

lemonade run user.poc-qwen2.5-coder-7b-v1-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use edoggy/poc-qwen2.5-coder-7b-v1 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf edoggy/poc-qwen2.5-coder-7b-v1: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 edoggy/poc-qwen2.5-coder-7b-v1:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use edoggy/poc-qwen2.5-coder-7b-v1 with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf edoggy/poc-qwen2.5-coder-7b-v1: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 "edoggy/poc-qwen2.5-coder-7b-v1: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"

poc-qwen2.5-coder-7b-v1

Describe a smart-contract bug, give it the project's test setup, and it writes a Foundry test that tries to prove the bug. Runs locally with Ollama, so your code stays private.

It writes drafts. Expect to finish and verify the test yourself.

Quick start

You need Ollama and Foundry.

pipx install huggingface_hub    # or: pip install -U huggingface_hub (inside a venv)
hf download edoggy/poc-qwen2.5-coder-7b-v1 --local-dir ~/poc-v1
cd ~/poc-v1
ollama create poc-7b-v1 -f Modelfile

Then, from your Foundry project root, with your prompt in prompt.txt:

ollama run --nowordwrap poc-7b-v1 "$(cat prompt.txt)" | sed '/^```/d' > test/PoC_1.t.sol
forge test --match-path test/PoC_1.t.sol -vvv

The system prompt and context size are built into the Modelfile. The prompt layout is the same as v5's; see the v5 card for the layout and how to verify a result.

Results

Tested on 37 bugs from 15 projects the model never saw during training. Each model worked alone, and every passing test was read by hand.

model	proved	progress (0-5)	compile errors
Qwen2.5-Coder-7B (base)	0	0.95	11
v1	0	0.78	18

v1 compiles less often than its base model. An earlier version of this card reported "1 of 22 proved" on an older, smaller test set; on the set above that pass did not hold up when read, so we report 0.

Limits

Foundry projects only. It proves bugs you describe; it doesn't find them.
Check your bounty platform's rules on AI-assisted work.

Trained (LoRA on Qwen2.5-Coder-7B-Instruct) on 58 PoCs from 47 projects, from public Code4rena and Sherlock reports. Credit to the wardens who found the bugs.