Instructions to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nachikethreddyy/qwen3.5-8b-distilled-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("nachikethreddyy/qwen3.5-8b-distilled-GGUF", dtype="auto")

llama-cpp-python

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="nachikethreddyy/qwen3.5-8b-distilled-GGUF",
	filename="qwen3.5-8b-distilled-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF 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 nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF: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 nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF: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 nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

Use Docker

docker model run hf.co/nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nachikethreddyy/qwen3.5-8b-distilled-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nachikethreddyy/qwen3.5-8b-distilled-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

SGLang

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nachikethreddyy/qwen3.5-8b-distilled-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nachikethreddyy/qwen3.5-8b-distilled-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nachikethreddyy/qwen3.5-8b-distilled-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nachikethreddyy/qwen3.5-8b-distilled-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Ollama:
```
ollama run hf.co/nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
```

Unsloth Studio

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF 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 nachikethreddyy/qwen3.5-8b-distilled-GGUF 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 nachikethreddyy/qwen3.5-8b-distilled-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for nachikethreddyy/qwen3.5-8b-distilled-GGUF to start chatting

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF: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": "nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF: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 nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF: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 "nachikethreddyy/qwen3.5-8b-distilled-GGUF: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"

Docker Model Runner
How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Docker Model Runner:
```
docker model run hf.co/nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M
```

Lemonade

How to use nachikethreddyy/qwen3.5-8b-distilled-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull nachikethreddyy/qwen3.5-8b-distilled-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.qwen3.5-8b-distilled-GGUF-Q4_K_M

List all available models

lemonade list

nachikethreddyy commited on 29 days ago

Commit

59dc3c4

verified ·

1 Parent(s): a112945

Update README for GGUF format

Browse files

Files changed (1) hide show

README.md +35 -208

README.md CHANGED Viewed

@@ -7,241 +7,68 @@ base_model: Qwen/Qwen3-8B
 tags:
   - text-generation
   - fine-tune
-  - lora
   - coding
-  - software-engineering
-  - qwen
-  - mlx
   - gguf
-  - quantized
 pipeline_tag: text-generation
 inference: true
 model_creator: nachikethreddyy
 ---
-# Qwen3.5-8B Distilled
-Fine-tuned **Qwen3.5-8B** model for software engineering and coding tasks. Available in **HuggingFace**, **MLX (Mac)**, and **GGUF** formats.
-## 📁 Repository Structure
-```
-qwen3.5-8b-distilled/
-├── README.md                           # This file
-├── model-00001-of-00004.safetensors   # HF Full Precision (root level)
-├── model-00002-of-00004.safetensors
-├── model-00003-of-00004.safetensors
-├── model-00004-of-00004.safetensors
-├── config.json
-├── tokenizer.json
-├── tokenizer_config.json
-├── generation_config.json
-│
-├── variants/q8/                        # HuggingFace Q8 Quantized
-│   ├── model.safetensors
-│   ├── config.json
-│   └── tokenizer.json
-│
-└── gguf/                               # GGUF Format (Ollama, llama.cpp)
-    └── qwen3-distilled-f16.gguf
-```
----
-## 🚀 Quick Start by Format
-### **HuggingFace Format (Python - Any OS)**
-#### Full Precision (Highest Quality)
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-model = AutoModelForCausalLM.from_pretrained(
-    "nachikethreddyy/qwen3.5-8b-distilled",
-    torch_dtype="auto",
-    device_map="auto"
-)
-tokenizer = AutoTokenizer.from_pretrained("nachikethreddyy/qwen3.5-8b-distilled")
-inputs = tokenizer("Write a Python function", return_tensors="pt")
-outputs = model.generate(**inputs, max_length=200)
-print(tokenizer.decode(outputs[0]))
-```
-#### Q8 Quantized (Faster, 46% Smaller)
-```python
-from transformers import AutoModelForCausalLM, BitsAndBytesConfig
-q8_config = BitsAndBytesConfig(load_in_8bit=True)
-model = AutoModelForCausalLM.from_pretrained(
-    "nachikethreddyy/qwen3.5-8b-distilled",
-    quantization_config=q8_config,
-    device_map="auto"
-)
-tokenizer = AutoTokenizer.from_pretrained("nachikethreddyy/qwen3.5-8b-distilled")
-```
----
-### **MLX Format (Mac M1/M2/M3/M4)**
-```python
-from mlx_lm import load, generate
-# Load from main repo (auto-downloads HF format)
-model, tokenizer = load("nachikethreddyy/qwen3.5-8b-distilled")
-response = generate(
-    model,
-    tokenizer,
-    prompt="Write a Python function to",
-    max_tokens=200,
-    verbose=True
-)
-print(response)
-```
----
-### **GGUF Format (Ollama, llama.cpp, LM Studio)**
-#### Ollama
 ```bash
-ollama run nachikethreddyy/qwen3.5-8b-distilled:gguf
-# Or create a Modelfile pointing to gguf/qwen3-distilled-f16.gguf
 ```
-#### llama.cpp
 ```bash
-./llama-cli -m gguf/qwen3-distilled-f16.gguf \
-  -p "Write a Python function" \
-  -n 200
 ```
-#### LM Studio (GUI)
 1. Download [LM Studio](https://lmstudio.ai)
-2. Search: `nachikethreddyy/qwen3.5-8b-distilled`
-3. Select the GGUF variant
-4. Run!
----
-## 📊 Model Variants Comparison
-| Format | Size | Quantization | Location | Best For |
-|--------|------|--------------|----------|----------|
-| **HuggingFace (Root)** | 16.39 GB | Full Precision (BF16) | `model-*.safetensors` | Python, PyTorch, max quality |
-| **HuggingFace (Q8)** | 8.80 GB | INT8 (8-bit) | `variants/q8/` | Production, balanced |
-| **GGUF F16** | 16 GB | Float16 | `gguf/qwen3-distilled-f16.gguf` | Ollama, llama.cpp, LM Studio |
-| **MLX** | Native | Optimized | Auto-loaded | Apple Silicon (fastest) |
----
-## 💻 Hardware Requirements
-| Setup | CPU RAM | GPU VRAM | Recommended |
-|-------|---------|----------|-------------|
-| Full Precision | 24GB+ | 16GB+ | High-end GPU, research |
-| Q8 Quantized | 12GB+ | 8GB+ | **Production (recommended)** |
-| MLX (Mac) | 8GB+ | Shared | **MacBook M1+ (best Mac option)** |
-| GGUF | 8GB+ | 4-8GB+ | Local inference, CPU-friendly |
----
-## 📚 Model Details
-### Architecture
-- **Model Type:** Qwen3 (Transformer-based)
-- **Base Model:** Qwen/Qwen3-8B
-- **Parameters:** 8 billion
-- **Hidden Size:** 4,096
-- **Attention Heads:** 32
-- **Layers:** 36
-- **Vocab Size:** 151,669 tokens
-- **Max Context:** 131,072 tokens
-### Training
-- **Method:** LoRA Fine-tuning (r=16, alpha=32)
-- **Data:** 256 curated coding/engineering examples
-- **Validation:** 33 examples
-- **Framework:** MLX (Apple Silicon optimized)
-- **Max Sequence:** 1536 tokens
-- **Iterations:** 1600 (~3-4 epochs)
-- **Training Time:** ~5 hours on Mac M4 Pro
----
-## 🎯 Use Cases
-✅ Code generation (Python, JavaScript, TypeScript)
-✅ Technical documentation and tutorials
-✅ Software architecture discussions
-✅ Algorithm explanations
-✅ Code review and debugging
-✅ AI/LLM concept explanations
----
-## ⚠️ Limitations
-- **Optimized for technical tasks:** May underperform on general conversation
-- **Context trained on 1536 tokens:** Can extrapolate but with degraded performance
-- **Potential hallucinations:** Like all LLMs, may generate plausible-sounding but incorrect code
-- **Knowledge cutoff:** Limited to Qwen3-8B base model knowledge
-- **May reflect training data biases**
----
-## 🔧 Advanced Usage
-### vLLM (Production Serving)
-```bash
-python -m vllm.entrypoints.openai_api_server \
-  --model nachikethreddyy/qwen3.5-8b-distilled \
-  --quantization bitsandbytes
-```
-### Further Fine-tuning
 ```python
-from peft import LoraConfig, get_peft_model
-from transformers import AutoModelForCausalLM
-base_model = AutoModelForCausalLM.from_pretrained(
-    "nachikethreddyy/qwen3.5-8b-distilled"
-)
-config = LoraConfig(
-    r=8,
-    lora_alpha=16,
-    target_modules=["q_proj", "v_proj"]
 )
-model = get_peft_model(base_model, config)
 ```
----
 ## 📄 License
 Apache 2.0 (inherited from Qwen/Qwen3-8B)
 ---
-## 🙏 Acknowledgments
-- **Base Model:** [Qwen Team](https://huggingface.co/Qwen)
-- **Fine-tuning:** MLX Framework
-- **Inference Frameworks:** Transformers, Ollama, llama.cpp, LM Studio
----
-## 📞 Support
-- **Model Page:** https://huggingface.co/nachikethreddyy/qwen3.5-8b-distilled
-- **Discussions:** Open a discussion on the model page
-- **Issues:** Report bugs or request features
----
-**Last Updated:** June 19, 2024
-**Status:** Production Ready ✅

 tags:
   - text-generation
   - fine-tune
   - coding
   - gguf
+  - llama.cpp
+  - ollama
+  - lm-studio
 pipeline_tag: text-generation
 inference: true
 model_creator: nachikethreddyy
 ---
+# Qwen3.5-8B Distilled - GGUF Format
+Fine-tuned **Qwen3.5-8B** for software engineering & coding tasks. **GGUF-optimized** version for local inference.
+## 📦 What's Included
+| Variant | Size | Format | Best For |
+|---------|------|--------|----------|
+| **Full Precision (BF16)** | 16.39 GB | Safetensors | Maximum quality, research |
+| **Q8 Quantized** | 8.8 GB | Safetensors | Balanced speed/quality |
+| **GGUF F16** | 15.3 GB | GGUF | Ollama, llama.cpp, LM Studio |
+## 🚀 Quick Start
+### Ollama
 ```bash
+ollama run nachikethreddyy/qwen3.5-8b-distilled-GGUF:F16
 ```
+### llama.cpp
 ```bash
+# Install
+brew install llama.cpp
+# Run
+llama-cli -hf nachikethreddyy/qwen3.5-8b-distilled-GGUF:F16
 ```
+### LM Studio
 1. Download [LM Studio](https://lmstudio.ai)
+2. Search: `nachikethreddyy/qwen3.5-8b-distilled-GGUF`
+3. Download & run!
+### Transformers (Full/Q8)
 ```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+model = AutoModelForCausalLM.from_pretrained(
+    "nachikethreddyy/qwen3.5-8b-distilled-GGUF",
+    device_map="auto"
 )
 ```
+## 📊 Training Details
+- **Base:** Qwen/Qwen3-8B
+- **Method:** LoRA Fine-tuning (r=16, alpha=32)
+- **Data:** 256 coding examples
+- **Framework:** MLX
+- **Iterations:** 1600
 ## 📄 License
 Apache 2.0 (inherited from Qwen/Qwen3-8B)
 ---
+**For MLX/Apple Silicon:** See [qwen3.5-8b-distilled-MLX](https://huggingface.co/nachikethreddyy/qwen3.5-8b-distilled-MLX)