Instructions to use INSAIT-Institute/BgGPT-7B-Instruct-v0.2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use INSAIT-Institute/BgGPT-7B-Instruct-v0.2 with Transformers:

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

pipe = pipeline("text-generation", model="INSAIT-Institute/BgGPT-7B-Instruct-v0.2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("INSAIT-Institute/BgGPT-7B-Instruct-v0.2")
model = AutoModelForCausalLM.from_pretrained("INSAIT-Institute/BgGPT-7B-Instruct-v0.2")
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 INSAIT-Institute/BgGPT-7B-Instruct-v0.2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "INSAIT-Institute/BgGPT-7B-Instruct-v0.2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "INSAIT-Institute/BgGPT-7B-Instruct-v0.2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/INSAIT-Institute/BgGPT-7B-Instruct-v0.2

SGLang

How to use INSAIT-Institute/BgGPT-7B-Instruct-v0.2 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 "INSAIT-Institute/BgGPT-7B-Instruct-v0.2" \
    --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": "INSAIT-Institute/BgGPT-7B-Instruct-v0.2",
		"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 "INSAIT-Institute/BgGPT-7B-Instruct-v0.2" \
        --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": "INSAIT-Institute/BgGPT-7B-Instruct-v0.2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use INSAIT-Institute/BgGPT-7B-Instruct-v0.2 with Docker Model Runner:
```
docker model run hf.co/INSAIT-Institute/BgGPT-7B-Instruct-v0.2
```

missing tokenizer.model

by vesofilev - opened Mar 4, 2024

Discussion

vesofilev

Mar 4, 2024

•

edited Mar 4, 2024

I had to copy tokenizer.model from v1.0 to be able to start the model, since I was encountering this issue:

https://github.com/ThilinaRajapakse/simpletransformers/issues/873

Is that the expected approach?

casic

Mar 4, 2024

И как го подкара ?

vesofilev

Mar 5, 2024

•

edited Mar 5, 2024

Копирах tokenizer.model oт version0.1 и след това с LLaMA-Factory

casic

Mar 5, 2024

Благодаря,
ще пробвам.

casic

Mar 5, 2024

На какъв хардуер работи ?

vesofilev

Mar 5, 2024

аз ползвах Tesla V100-SXM2-32GB, но заема малко повече от половината памет

ldilov

Mar 5, 2024

Tokenizer.model не ви трябва. Относно VRAM, помислете за quantuzation.

casic

Mar 6, 2024

При мене тръгна. Обаче лъже като дърт циганин.

aalexandrov

Institute for Computer Science, Artificial intelligence and Technology org Mar 6, 2024

tokenizer.model has been added to the repo. You were correct in using the one from v0.1. You can use one of the GGUF quantized versions if it doesn't fit on your hardware, otherwise, you need more than 15GB of VRAM to run without offloading to CPU. If you have a good GPU but not enough memory, I believe llamacpp can help with doing the opposite - GPU offloading.

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