Instructions to use TheFloat16/Llama3-70b-Instruct-TRTLLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TheFloat16/Llama3-70b-Instruct-TRTLLM with Transformers:

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

pipe = pipeline("text-generation", model="TheFloat16/Llama3-70b-Instruct-TRTLLM")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("TheFloat16/Llama3-70b-Instruct-TRTLLM", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TheFloat16/Llama3-70b-Instruct-TRTLLM with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TheFloat16/Llama3-70b-Instruct-TRTLLM"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TheFloat16/Llama3-70b-Instruct-TRTLLM",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TheFloat16/Llama3-70b-Instruct-TRTLLM

SGLang

How to use TheFloat16/Llama3-70b-Instruct-TRTLLM 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 "TheFloat16/Llama3-70b-Instruct-TRTLLM" \
    --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": "TheFloat16/Llama3-70b-Instruct-TRTLLM",
		"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 "TheFloat16/Llama3-70b-Instruct-TRTLLM" \
        --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": "TheFloat16/Llama3-70b-Instruct-TRTLLM",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TheFloat16/Llama3-70b-Instruct-TRTLLM with Docker Model Runner:
```
docker model run hf.co/TheFloat16/Llama3-70b-Instruct-TRTLLM
```

Incompatible with tensorrt_llm==0.10.0

by oobabooga - opened Jun 27, 2024

Discussion

oobabooga

Jun 27, 2024

When I try to load one of these models, I get:

│ /home/user/miniconda3/envs/tensorrt/lib/python3.10/site-packages/tensorrt_llm/runtime/model_runner │
│ .py:85 in _builder_to_model_config                                                                    │
│                                                                                                       │
│    84 def _builder_to_model_config(config: dict) -> Tuple[ModelConfig, dict]:                         │
│ ❱  85     builder_config = config['builder_config']                                                   │
│    86     model_name = builder_config['name']                                                         │
╰───────────────────────────────────────────────────────────────────────────────────────────────────────╯
KeyError: 'builder_config'

This happens because the structure of config.json has changed from 0.9 to 0.10. It would be nice if you could update the config.json files in this repository.

matichon

TheFloat16 org Jun 27, 2024

@oobabooga
Hi bro, Thanks for your notice. I'm going to solve this issue this weekend.
Mati

shacotustra

Jun 30, 2024

If it will compatible with 0.11.0 it will be perfect!

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