Instructions to use TurkuNLP/gpt3-finnish-13B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TurkuNLP/gpt3-finnish-13B with Transformers:

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

pipe = pipeline("text-generation", model="TurkuNLP/gpt3-finnish-13B")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained("TurkuNLP/gpt3-finnish-13B")
model = AutoModel.from_pretrained("TurkuNLP/gpt3-finnish-13B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TurkuNLP/gpt3-finnish-13B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TurkuNLP/gpt3-finnish-13B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TurkuNLP/gpt3-finnish-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TurkuNLP/gpt3-finnish-13B

SGLang

How to use TurkuNLP/gpt3-finnish-13B 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 "TurkuNLP/gpt3-finnish-13B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TurkuNLP/gpt3-finnish-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "TurkuNLP/gpt3-finnish-13B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TurkuNLP/gpt3-finnish-13B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TurkuNLP/gpt3-finnish-13B with Docker Model Runner:
```
docker model run hf.co/TurkuNLP/gpt3-finnish-13B
```

KeyError bloom? PredictionException 'bloom'?

by juusohugs - opened Mar 14, 2023

Discussion

juusohugs

Mar 14, 2023

Tried to deploy this model to amazon aws sagemaker. I got endpoint done but when doing inference getting KeyError 'bloom'.

Log showed one line:
mms.service.PredictionException: 'bloom'

I wonder if this is my fault or some sort of actual ModelError?

spyysalo

TurkuNLP Research Group org Mar 16, 2023

I'm afraid we haven't tried this on sagemaker. The model is internally a BloomModel, so perhaps this might related to BLOOM support? Can you run the original BLOOM models?

juusohugs

Mar 16, 2023

I'll give it a go. This might be related to config file missing piece of information. Or it might be huggingface version issue (sagemaker uses transformers_version="4.17")

juusohugs

Mar 20, 2023

No luck with BloomModels. I'm fairly certain the problem relates to transformers version.

Here's a good & detailed article on potential solution (have not tried yet) in case somebody else has the same issue:

https://towardsdatascience.com/unlock-the-latest-transformer-models-with-amazon-sagemaker-7fe65130d993

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