Instructions to use steve-cse/MelloGPT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use steve-cse/MelloGPT with Transformers:

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

pipe = pipeline("text-generation", model="steve-cse/MelloGPT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("steve-cse/MelloGPT")
model = AutoModelForCausalLM.from_pretrained("steve-cse/MelloGPT")
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

vLLM

How to use steve-cse/MelloGPT with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/steve-cse/MelloGPT

SGLang

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

Docker Model Runner
How to use steve-cse/MelloGPT with Docker Model Runner:
```
docker model run hf.co/steve-cse/MelloGPT
```

source code for fine-tuning

by radchaneepornc - opened Feb 9, 2024

Discussion

radchaneepornc

Feb 9, 2024

Hi Steve,

I hope you're doing great! I've been checking out your work and it's really interesting. I've even tried some of it myself and it worked out pretty well 👍.

Right now, I'm getting more into research in this area and I was hoping you could help me out. Would you be able to share any source code for fine-tuning or maybe some tutorials you have? It would really help me dive deeper.

Thanks a lot for considering my request. Looking forward to hearing from you!

Best,

steve-cse

Owner Feb 9, 2024

•

edited Feb 9, 2024

Hello Radchaneeporn,

I used a tool called axolotl to streamline the fine tuning process. Check it out, https://github.com/OpenAccess-AI-Collective/axolotl
The starter guide I used was this, https://mlabonne.github.io/blog/posts/A_Beginners_Guide_to_LLM_Finetuning.html.
You can find the axolotl configuration file I wrote for mello here, https://github.com/steve-cse/mello/blob/master/mello.yml.
From my personal point of view cleaning/preparing the dataset and setting the right values during fine tuning such as epochs etc.. is more intricate and takes more time.

Hope this helps ⚡

radchaneepornc

Feb 10, 2024

Thank you so much, I appreciate your response :D

steve-cse changed discussion status to closed Feb 28, 2024

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