Instructions to use jschwab21/Reflection-Llama-3.1-70B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jschwab21/Reflection-Llama-3.1-70B with Transformers:

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

pipe = pipeline("text-generation", model="jschwab21/Reflection-Llama-3.1-70B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jschwab21/Reflection-Llama-3.1-70B")
model = AutoModelForCausalLM.from_pretrained("jschwab21/Reflection-Llama-3.1-70B", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use jschwab21/Reflection-Llama-3.1-70B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jschwab21/Reflection-Llama-3.1-70B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jschwab21/Reflection-Llama-3.1-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jschwab21/Reflection-Llama-3.1-70B

SGLang

How to use jschwab21/Reflection-Llama-3.1-70B 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 "jschwab21/Reflection-Llama-3.1-70B" \
    --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": "jschwab21/Reflection-Llama-3.1-70B",
		"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 "jschwab21/Reflection-Llama-3.1-70B" \
        --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": "jschwab21/Reflection-Llama-3.1-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use jschwab21/Reflection-Llama-3.1-70B with Docker Model Runner:
```
docker model run hf.co/jschwab21/Reflection-Llama-3.1-70B
```

Reflection-Llama-3.1-70B / handler.py

jschwab21

Create handler.py

351aa3d verified almost 2 years ago

Raw

History Blame Contribute Delete

1.3 kB

	from typing import Dict, Any, List
	from transformers import AutoModelForCausalLM, AutoTokenizer
	import torch

	class EndpointHandler():
	def __init__(self, path=""):
	# Load the model in FP16 to reduce memory usage while retaining performance.
	self.model = AutoModelForCausalLM.from_pretrained(path, torch_dtype=torch.float16)
	self.tokenizer = AutoTokenizer.from_pretrained(path)

	def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
	"""
	data args:
	inputs (str): The text input or prompts for the model
	Return:
	A list containing the generated responses.
	"""
	# Extract the input text from the request
	inputs = data.get("inputs", "")
	if not inputs:
	return [{"error": "No input provided"}]

	# Tokenize the input and run the model to generate output
	tokens = self.tokenizer(inputs, return_tensors="pt").to(torch.float16)
	output_tokens = self.model.generate(**tokens)

	# Decode the generated tokens back to text
	output_text = self.tokenizer.decode(output_tokens[0], skip_special_tokens=True)

	# Return the generated response as a list (required format)
	return [{"generated_text": output_text}]