Instructions to use LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct

SGLang

How to use LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct 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 "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct" \
    --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": "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct",
		"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 "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct" \
        --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": "LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct with Docker Model Runner:
```
docker model run hf.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct
```

support ollama ??

#14

by skju - opened Aug 30, 2024

Discussion

skju

Aug 30, 2024

i coverted this to .gguf file or used another link .gguf file
but do not support in ollama (version 0.3.8)

ollama create : success
ollama run : failed

yireun

Sep 9, 2024

•

edited Sep 9, 2024

Here is simple guidelines for using the EXAONE model on ollama:

Download the EXAONE model from HuggingFace, and save to /path/to/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct.
Llamafy the EXAONE model by referring to the followings.
- maywell/EXAONE-3.0-7.8B-Instruct-Llamafied
- CarrotAI/EXAONE-3.0-7.8B-Instruct-Llamafied-cpu
Create the EXAONE Modelfile. See https://github.com/ollama/ollama/blob/main/docs/modelfile.md for more information. This is an example of the EXAONE Modelfile.


# Set the base model.
FROM /path/to/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct-Llamafied

# Set the parameter values according to your application.
PARAMETER stop "[|endofturn|]"
PARAMETER num_predict -2
PARAMETER top_k 1

# Set the template.
TEMPLATE """{{ if .System }}[|system|]{{ .System }}[|endofturn|]
{{ end }}{{ if .Prompt }}[|user|]{{ .Prompt }}
{{ end }}[|assistant|]{{ .Response }}[|endofturn|]
"""

# Set the system prompt.
SYSTEM """You are EXAONE model from LG AI Research, a helpful assistant."""

# Set the license.
LICENSE """EXAONE AI Model License Agreement 1.1 - NC """

Convert the EXAONE model saved as pyTorch safetensors to ollama. To quantize the EXAONE model, you can add --quantize flag. Please refer to https://github.com/ollama/ollama/blob/main/docs/import.md for the quantization flag.
$ ollama create exaone3 -f <the EXAONE Modelfile>

Good luck to you.

skju

Sep 9, 2024

•

edited Sep 9, 2024

@yireun Thank you for your reply. But when I tried
Again, the following message occurs.

ollama create : ok
ollama run : failed
"Error: llama runner process has terminated: this model is not supported by your version of Ollama. You may need to upgrade"

I think the llama.cpp library inside ollama should be updated.

yireun

Sep 9, 2024

@skju
Sorry, I missed the step of llamafying the EXAONE model.
I updated above posted guidelines. Would you try again?

skju

Sep 10, 2024

Thank you , but ....ㅜ.ㅜ
I received the "maywell/EXAONE-3.0-7.8B-Instruct-Llamafied" weights

but
I got a error message.
Error: no safetensors or torch files found

The files are already there, also Modelfile path is right!

please make a gguf file for ollama

yireun

Sep 10, 2024

Would you check your path again?
According to https://github.com/ollama/ollama/blob/4a8069f9c4c8cb761cd6c10ca5f4be6af21fa0ae/cmd/cmd.go#L222,
the error "Error: no safetensors or torch files found" occurs when ollama cannot find files "model*.safetensors".

When I used the maywell/EXAONE-3.0-7.8B-Instruct-Llamafied weights, no errors occurred.

If you use an ollama docker container, those two paths should point the paths in container.
- FROM <the EXAONE-Llamafied model path>
- $ ollama create exaone3 -f <the EXAONE Modelfile>

Good luck to you.