Instructions to use DW-ReCo/spot_llama3_4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DW-ReCo/spot_llama3_4bit with Transformers:

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

pipe = pipeline("text-generation", model="DW-ReCo/spot_llama3_4bit")

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

tokenizer = AutoTokenizer.from_pretrained("DW-ReCo/spot_llama3_4bit")
model = AutoModelForCausalLM.from_pretrained("DW-ReCo/spot_llama3_4bit", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use DW-ReCo/spot_llama3_4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DW-ReCo/spot_llama3_4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DW-ReCo/spot_llama3_4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/DW-ReCo/spot_llama3_4bit

SGLang

How to use DW-ReCo/spot_llama3_4bit 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 "DW-ReCo/spot_llama3_4bit" \
    --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": "DW-ReCo/spot_llama3_4bit",
		"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 "DW-ReCo/spot_llama3_4bit" \
        --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": "DW-ReCo/spot_llama3_4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Unsloth Desktop
Docker Model Runner
How to use DW-ReCo/spot_llama3_4bit with Docker Model Runner:
```
docker model run hf.co/DW-ReCo/spot_llama3_4bit
```

spot_llama3_4bit / handler.py

DW-ReCo

Create handler.py

ab09317 verified about 2 years ago

Raw History Blame Contribute Delete

1.52 kB

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


alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
### Instruction:
{}
### Input:
{}
### Response:
{}"""

 
class EndpointHandler:
    def __init__(self, path=""):
        # load model and processor from path
        self.model = AutoModelForCausalLM.from_pretrained(path, device_map="auto")
        self.tokenizer = AutoTokenizer.from_pretrained(path)

        with open(f"{path}/zero_shot_cot_prompt.txt", 'r') as file:
            self.instruction_prompt = file.read()

 
    def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
        sentence = data.pop("inputs",data)

        inputs = self.tokenizer(
            [
                alpaca_prompt.format(
                    self.instruction_prompt,  # instruction
                    sentence,  # input
                    "",  # output - leave this blank for generation!
                )
            ], return_tensors="pt")
        
        outputs = self.model.generate(**inputs,
                    max_new_tokens=1048,
                    use_cache=True,
                    top_p=0.1,
                    temperature=0.001)
        
        outputs = self.tokenizer.batch_decode(outputs)[0]
        response = outputs.split("### Response:")[1].split("<|end_of_text|>")[0]
 
        return [{"generated_text": response}]

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


	alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.

	### Instruction:
	{}

	### Input:
	{}

	### Response:
	{}"""


	class EndpointHandler:
	def __init__(self, path=""):
	# load model and processor from path
	self.model = AutoModelForCausalLM.from_pretrained(path, device_map="auto")
	self.tokenizer = AutoTokenizer.from_pretrained(path)

	with open(f"{path}/zero_shot_cot_prompt.txt", 'r') as file:
	self.instruction_prompt = file.read()


	def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
	sentence = data.pop("inputs",data)

	inputs = self.tokenizer(
	[
	alpaca_prompt.format(
	self.instruction_prompt, # instruction
	sentence, # input
	"", # output - leave this blank for generation!
	)
	], return_tensors="pt")

	outputs = self.model.generate(**inputs,
	max_new_tokens=1048,
	use_cache=True,
	top_p=0.1,
	temperature=0.001)

	outputs = self.tokenizer.batch_decode(outputs)[0]
	response = outputs.split("### Response:")[1].split("<\|end_of_text\|>")[0]

	return [{"generated_text": response}]