Instructions to use YTan2000/Qwen3.6-27B-TQ3_4S with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use YTan2000/Qwen3.6-27B-TQ3_4S with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="YTan2000/Qwen3.6-27B-TQ3_4S",
	filename="Qwen3.6-27B-TQ3_4S.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": [
				{
					"type": "text",
					"text": "Describe this image in one sentence."
				},
				{
					"type": "image_url",
					"image_url": {
						"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
					}
				}
			]
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use YTan2000/Qwen3.6-27B-TQ3_4S with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf YTan2000/Qwen3.6-27B-TQ3_4S
# Run inference directly in the terminal:
llama-cli -hf YTan2000/Qwen3.6-27B-TQ3_4S

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf YTan2000/Qwen3.6-27B-TQ3_4S
# Run inference directly in the terminal:
llama-cli -hf YTan2000/Qwen3.6-27B-TQ3_4S

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 YTan2000/Qwen3.6-27B-TQ3_4S
# Run inference directly in the terminal:
./llama-cli -hf YTan2000/Qwen3.6-27B-TQ3_4S

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 YTan2000/Qwen3.6-27B-TQ3_4S
# Run inference directly in the terminal:
./build/bin/llama-cli -hf YTan2000/Qwen3.6-27B-TQ3_4S

Use Docker

docker model run hf.co/YTan2000/Qwen3.6-27B-TQ3_4S

LM Studio
Jan

vLLM

How to use YTan2000/Qwen3.6-27B-TQ3_4S with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "YTan2000/Qwen3.6-27B-TQ3_4S"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YTan2000/Qwen3.6-27B-TQ3_4S",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/YTan2000/Qwen3.6-27B-TQ3_4S

Ollama
How to use YTan2000/Qwen3.6-27B-TQ3_4S with Ollama:
```
ollama run hf.co/YTan2000/Qwen3.6-27B-TQ3_4S
```

Unsloth Studio new

How to use YTan2000/Qwen3.6-27B-TQ3_4S 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 YTan2000/Qwen3.6-27B-TQ3_4S 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 YTan2000/Qwen3.6-27B-TQ3_4S to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for YTan2000/Qwen3.6-27B-TQ3_4S to start chatting

Pi new

How to use YTan2000/Qwen3.6-27B-TQ3_4S with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf YTan2000/Qwen3.6-27B-TQ3_4S

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": "Qwen3.6-27B-TQ3_4S"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use YTan2000/Qwen3.6-27B-TQ3_4S with Docker Model Runner:
```
docker model run hf.co/YTan2000/Qwen3.6-27B-TQ3_4S
```

Lemonade

How to use YTan2000/Qwen3.6-27B-TQ3_4S with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull YTan2000/Qwen3.6-27B-TQ3_4S

Run and chat with the model

lemonade run user.Qwen3.6-27B-TQ3_4S-{{QUANT_TAG}}

List all available models

lemonade list

YTan2000 commited on 16 days ago

Commit

5d0ccf6

verified ·

1 Parent(s): 4b4605d

Update model card for text-only and image-to-text usage

Browse files

Files changed (1) hide show

README.md +53 -4

README.md CHANGED Viewed

@@ -24,7 +24,10 @@ language:
 ## TQ3_4S Release
-This repository packages the model as a TurboQuant `TQ3_4S` GGUF for local deployment.
 ## Runtime Compatibility
@@ -36,7 +39,8 @@ This quant requires a TurboQuant-capable runtime. For llama.cpp, use the `turbo-
 | File | Quant | Size |
 | --- | --- | ---: |
-| `Qwen3.6-27B-TQ3_4S.gguf` | TQ3_4S | ~13.0 GB |
 | `chat_template.jinja` | chat template | text |
 | `thumbnail.png` | model card image | png |
@@ -62,11 +66,44 @@ Prompt processing:
 - Use a TurboQuant-capable llama.cpp build for best performance.
 - For llama.cpp, the intended runtime is the `turbo-tan/llama.cpp-tq3` fork.
-- The upstream family is multimodal-capable, but the public 27B repos used here do not currently expose a separate GGUF `mmproj` artifact.
 - For llama.cpp chat usage, keep `--jinja` enabled so the bundled chat template is honored.
 - Upstream guidance recommends keeping at least `128K` context when possible for reasoning-heavy workloads. On smaller local GPUs, reduce context as needed to fit memory.
 - Upstream default sampling guidance differs between thinking and non-thinking mode; follow the official Qwen card if you are trying to reproduce base-model behavior.
 ## Recommended llama.cpp Settings
 Default prompt-processing settings on 16 GB:
@@ -81,11 +118,23 @@ llama-bench \
   -p 2048 -n 0 -r 3
 ```
-Default chat/server settings:
 ```bash
 llama-server \
   -m Qwen3.6-27B-TQ3_4S.gguf \
   --host 127.0.0.1 --port 8080 \
   -ngl 99 -c 4096 -np 1 \
   -ctk q4_0 -ctv tq3_0 -fa on \

 ## TQ3_4S Release
+This repository packages the model as a TurboQuant `TQ3_4S` GGUF for local deployment. It can be used in two modes:
+- **Text-only chat / coding:** use `Qwen3.6-27B-TQ3_4S.gguf` only.
+- **Image-to-text / multimodal:** use `Qwen3.6-27B-TQ3_4S.gguf` together with `mmproj.gguf`.
 ## Runtime Compatibility
 | File | Quant | Size |
 | --- | --- | ---: |
+| `Qwen3.6-27B-TQ3_4S.gguf` | TQ3_4S text model | ~13.0 GB |
+| `mmproj.gguf` | Qwen3.6-27B vision projector | ~889 MB |
 | `chat_template.jinja` | chat template | text |
 | `thumbnail.png` | model card image | png |
 - Use a TurboQuant-capable llama.cpp build for best performance.
 - For llama.cpp, the intended runtime is the `turbo-tan/llama.cpp-tq3` fork.
+- Text-only usage does not need `mmproj.gguf`.
+- Image-to-text usage requires `mmproj.gguf`; pass it with `--mmproj mmproj.gguf` when using `llama-server` or other compatible llama.cpp tools.
 - For llama.cpp chat usage, keep `--jinja` enabled so the bundled chat template is honored.
 - Upstream guidance recommends keeping at least `128K` context when possible for reasoning-heavy workloads. On smaller local GPUs, reduce context as needed to fit memory.
 - Upstream default sampling guidance differs between thinking and non-thinking mode; follow the official Qwen card if you are trying to reproduce base-model behavior.
+## Text-Only vs Image-To-Text
+### Text-only
+For normal chat, coding, and text generation, load only the main model:
+```bash
+llama-server \
+  -m Qwen3.6-27B-TQ3_4S.gguf \
+  -ngl 99 -c 4096 -np 1 \
+  -ctk q4_0 -ctv tq3_0 -fa on \
+  --jinja --reasoning off --reasoning-budget 0
+```
+### Image-to-text
+For vision/image prompts, also load the projector:
+`mmproj.gguf` was smoke-tested with the `turbo-tan/llama.cpp-tq3` `llama-server` runtime on RTX 5060 Ti. The server loaded the projector as a Qwen-VL multimodal model and `/health` returned `ok`.
+Validated smoke-test settings:
+```bash
+llama-server \
+  -m Qwen3.6-27B-TQ3_4S.gguf \
+  --mmproj mmproj.gguf \
+  -ngl 99 -c 2048 -np 1 \
+  -ctk q4_0 -ctv tq3_0 -fa on \
+  --jinja --reasoning off --reasoning-budget 0
+```
 ## Recommended llama.cpp Settings
 Default prompt-processing settings on 16 GB:
   -p 2048 -n 0 -r 3
 ```
+Default text-only chat/server settings:
+```bash
+llama-server \
+  -m Qwen3.6-27B-TQ3_4S.gguf \
+  --host 127.0.0.1 --port 8080 \
+  -ngl 99 -c 4096 -np 1 \
+  -ctk q4_0 -ctv tq3_0 -fa on \
+  --jinja
+```
+Image-to-text server settings:
 ```bash
 llama-server \
   -m Qwen3.6-27B-TQ3_4S.gguf \
+  --mmproj mmproj.gguf \
   --host 127.0.0.1 --port 8080 \
   -ngl 99 -c 4096 -np 1 \
   -ctk q4_0 -ctv tq3_0 -fa on \