Instructions to use DuckyBlender/bawialniagpt with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DuckyBlender/bawialniagpt with Transformers:

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

pipe = pipeline("text-generation", model="DuckyBlender/bawialniagpt", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DuckyBlender/bawialniagpt", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("DuckyBlender/bawialniagpt", trust_remote_code=True, device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use DuckyBlender/bawialniagpt with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/DuckyBlender/bawialniagpt

SGLang

How to use DuckyBlender/bawialniagpt 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 "DuckyBlender/bawialniagpt" \
    --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": "DuckyBlender/bawialniagpt",
		"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 "DuckyBlender/bawialniagpt" \
        --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": "DuckyBlender/bawialniagpt",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DuckyBlender/bawialniagpt with Docker Model Runner:
```
docker model run hf.co/DuckyBlender/bawialniagpt
```

BawialniaGPT: A Fine-Tuned Phi-3 Model for... Well, Something

Warning: This model is a joke and should not be used for any practical purposes. It was trained on a low-quality dataset and is only intended for entertainment purposes. It's basically an over-engineered markov algorithm.

Model Description

BawialniaGPT is a fine-tuned QLora model based on the Phi-3 model. It was trained on the Bawialnia Telegram Group Dataset, which is low quality and lacks context. Despite the challenges, the model was trained for approximately 8 hours on an RTX 4060 GPU, because why not? And yes, it is horribly overtrained.

Model Performance

The model's performance is... questionable. It's not entirely clear what the model is good at, but it's definitely not good at generating coherent or meaningful text. In fact, it just generates random polish-ish garbage. This is not a bug, it's a feature.

Model Statistics

Training time: ~8 hours (3 epochs)
GPU: RTX 4060
Dataset: Bawialnia Telegram Group Dataset (because why not?)
Architecture: PHI-3
Fine-tuning: QLora

Model Limitations

The model is not suitable for any practical applications.
The model may generate nonsensical or offensive responses.
The model may not respond at all, or respond with complete gibberish.

Usage

If you're feeling adventurous, you can use the model to generate text. Just don't say I didn't warn you.

Disclaimer

The creators of this model are not responsible for any damage, confusion, or frustration caused by using this model. You have been warned.

Downloads last month: 5

Safetensors

Model size

4B params

Tensor type

BF16