Instructions to use fsaudm/Meta-Llama-3.1-70B-Instruct-INT8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fsaudm/Meta-Llama-3.1-70B-Instruct-INT8 with Transformers:

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

pipe = pipeline("text-generation", model="fsaudm/Meta-Llama-3.1-70B-Instruct-INT8")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

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

tokenizer = AutoTokenizer.from_pretrained("fsaudm/Meta-Llama-3.1-70B-Instruct-INT8")
model = AutoModelForCausalLM.from_pretrained("fsaudm/Meta-Llama-3.1-70B-Instruct-INT8", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use fsaudm/Meta-Llama-3.1-70B-Instruct-INT8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/fsaudm/Meta-Llama-3.1-70B-Instruct-INT8

SGLang

How to use fsaudm/Meta-Llama-3.1-70B-Instruct-INT8 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 "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8" \
    --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": "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8",
		"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 "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8" \
        --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": "fsaudm/Meta-Llama-3.1-70B-Instruct-INT8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use fsaudm/Meta-Llama-3.1-70B-Instruct-INT8 with Docker Model Runner:
```
docker model run hf.co/fsaudm/Meta-Llama-3.1-70B-Instruct-INT8
```

Unable to run with vLLM

by yaronr - opened Oct 30, 2024

Discussion

yaronr

Oct 30, 2024

Hi
I am getting the following error when running with latest vllm, docker.
Here are my runtime params:

    "command": "--port=8000 
                                --model=fsaudm/Meta-Llama-3.1-70B-Instruct-INT8
                                --tensor-parallel-size=4
                                --pipeline-parallel-size=1 
                                --disable-log-requests
                                --enable-chunked-prefill
                                --num-scheduler-steps=10
                                --enable-prefix-caching
                                --max-num-batched-tokens=16192
                                --max-model-len=16192
                --max-seq-len-to-capture=16192
                                --gpu-memory-utilization=0.95"

Here's the error:

(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229] Exception in worker VllmWorkerProcess while processing method load_model.
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229] Traceback (most recent call last):
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/executor/multiproc_worker_utils.py", line 223, in _run_worker_process
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     output = executor(*args, **kwargs)
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]              ^^^^^^^^^^^^^^^^^^^^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/worker/worker.py", line 183, in load_model
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     self.model_runner.load_model()
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/worker/multi_step_model_runner.py", line 645, in load_model
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     return self._base_model_runner.load_model()
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/worker/model_runner.py", line 1058, in load_model
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     self.model = get_model(model_config=self.model_config,
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/model_loader/__init__.py", line 19, in get_model
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     return loader.load_model(model_config=model_config,
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/model_loader/loader.py", line 402, in load_model
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     model.load_weights(self._get_all_weights(model_config, model))
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/models/llama.py", line 582, in load_weights
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     loader.load_weights(
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/models/utils.py", line 203, in load_weights
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     autoloaded_weights = list(self._load_module("", self.module, weights))
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/models/utils.py", line 182, in _load_module
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     yield from self._load_module(prefix,
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/models/utils.py", line 169, in _load_module
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     module_load_weights(weights)
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]   File "/usr/local/lib/python3.12/dist-packages/vllm/model_executor/models/llama.py", line 414, in load_weights
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]     param = params_dict[name]
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229]             ~~~~~~~~~~~^^^^^^
(VllmWorkerProcess pid=148) ERROR 10-30 01:05:12 multiproc_worker_utils.py:229] KeyError: 'layers.0.mlp.down_proj.SCB'

vijaydeshpande

Feb 6, 2025

I am also getting the same error

yaronr

Feb 6, 2025

@vijaydeshpande I started using Huggingface TGI (ghcr.io/huggingface/text-generation-inference:latest). Maybe a little bit less performant on some edge use cases, but it just works, with very little parameter tuning. It just figures it out and runs..

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