Instructions to use ccore/Qwen3.5-4B-3D-Coder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use ccore/Qwen3.5-4B-3D-Coder 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 ccore/Qwen3.5-4B-3D-Coder:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ccore/Qwen3.5-4B-3D-Coder:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ccore/Qwen3.5-4B-3D-Coder:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ccore/Qwen3.5-4B-3D-Coder: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 ccore/Qwen3.5-4B-3D-Coder:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf ccore/Qwen3.5-4B-3D-Coder:Q4_K_M

Use Docker

docker model run hf.co/ccore/Qwen3.5-4B-3D-Coder:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use ccore/Qwen3.5-4B-3D-Coder with Ollama:
```
ollama run hf.co/ccore/Qwen3.5-4B-3D-Coder:Q4_K_M
```
Unsloth Desktop

How to use ccore/Qwen3.5-4B-3D-Coder with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ccore/Qwen3.5-4B-3D-Coder: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": "ccore/Qwen3.5-4B-3D-Coder:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use ccore/Qwen3.5-4B-3D-Coder with Docker Model Runner:
```
docker model run hf.co/ccore/Qwen3.5-4B-3D-Coder:Q4_K_M
```

Lemonade

How to use ccore/Qwen3.5-4B-3D-Coder with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ccore/Qwen3.5-4B-3D-Coder:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.5-4B-3D-Coder-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use ccore/Qwen3.5-4B-3D-Coder with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ccore/Qwen3.5-4B-3D-Coder: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 ccore/Qwen3.5-4B-3D-Coder:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ccore/Qwen3.5-4B-3D-Coder with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ccore/Qwen3.5-4B-3D-Coder: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 "ccore/Qwen3.5-4B-3D-Coder: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"

Advised harness & config?

by kaiami - opened 7 days ago

Discussion

kaiami

7 days ago

•

edited 7 days ago

Please can you advise which harness you used and which configuration of the Q8 model you used for this to work?
In Unsloth it seems unable to successfully one-shot any of your examples, it will usually fail somewhere down the chain with the edit-file tool call, go into a loop or the resulting HTML is incomplete or has errors.
I haven't been able to get anything to successfully work from this so I hope you can provide some insight on what I might be doing wrong 😊

ccore

Owner 7 days ago

hey! thanks for giving the model a try! :)

i'm using llama.cpp b11889 with its built-in webui, full gpu offload, flash attention enabled and f16 kv cache.

my sampling settings are temperature 0.6, top k 20, top p 0.95 and min p 0. batch size is 512, microbatch size is 128, with just one parallel slot.

one important thing is to set the output limit to -1 (unlimited), otherwise the generation might get cut off before the html is complete. of course, it's still limited by the available context window.

i recommend around 72k context if your hardware can handle it, since some of the training samples go up to 72k tokens. 64k should already cover most cases though.

for my tests, i simply ask the model to generate a complete standalone html file directly in chat. i haven't tested unsloth's edit-file tool workflow, so that might explain some of the issues you're seeing.

also, could you try deleting your current gguf and downloading it again from the model page? you might have downloaded one of the earlier versions, and i've already uploaded improved checkpoints.

let me know if that helps! i'd be happy to figure out what's going wrong if you're still having trouble. thanks again for testing it!

kaiami

6 days ago

Thanks for your reply, indeed the edit-file tool was the issue, having it generate the HTML inline as a code block worked instead.

kaiami changed discussion status to closed 6 days ago

ccore

Owner 6 days ago

glad to hear it worked! haha

quite a few people have been running into the same issue, so i'm happy we figured it out.

the model was trained mainly to generate complete responses, so it works best when producing a full html file from scratch. you can also ask it to rewrite the entire html with changes or improvements, and that works too.

the tool calling and edit-file workflows aren't very reliable yet, since the training was focused on complete code generation rather than those kinds of interactions.

that's definitely something i'd like to improve in future training. thanks again for testing and sharing the feedback!

TimeLordRaps

5 days ago

what thinking do you set it to for your results?

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment