LLM with Streaming

Building an LLM with streaming output

View on Github

In this example, we go through a Truss that serves an LLM, and streams the output to the client.

Why Streaming?

For certain ML models, generations can take a long time. Especially with LLMs, a long output could take 10 - 20 seconds to generate. However, because LLMs generate tokens in sequence, useful output can be made available to users sooner. To support this, in Truss, we support streaming output. In this example, we build a Truss that streams the output of the Falcon-7B model.

Set up the imports and key constants

In this example, we use the HuggingFace transformers library to build a text generation model.

model/model.py

from threading import Thread
from typing import Dict

import torch
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    GenerationConfig,
    TextIteratorStreamer,
)

We use the instruct version of the Falcon-7B model, and have some defaults for inference parameters.

model/model.py

CHECKPOINT = "tiiuae/falcon-7b-instruct"
DEFAULT_MAX_NEW_TOKENS = 150
DEFAULT_TOP_P = 0.95

Define the load function

In the load function of the Truss, we implement logic involved in downloading the model and loading it into memory.

model/model.py

class Model:
    def __init__(self, **kwargs) -> None:
        self.tokenizer = None
        self.model = None

    def load(self):
        self.tokenizer = AutoTokenizer.from_pretrained(CHECKPOINT)

model/model.py

        self.tokenizer.pad_token = self.tokenizer.eos_token_id
        self.model = AutoModelForCausalLM.from_pretrained(
            CHECKPOINT,
            torch_dtype=torch.bfloat16,
            trust_remote_code=True,
            device_map="auto",
        )

Define the predict function

In the predict function of the Truss, we implement the actual inference logic. The two main steps are:

• Tokenize the input
• Call the model’s generate function, ensuring that we pass a TextIteratorStreamer. This is what gives us streaming output, and and also do this in a Thread, so that it does not block the main invocation.
• Return a generator that iterates over the TextIteratorStreamer object

model/model.py

    def predict(self, request: Dict) -> Dict:
        prompt = request.pop("prompt")
        inputs = self.tokenizer(
            prompt, return_tensors="pt", max_length=512, truncation=True, padding=True
        )
        input_ids = inputs["input_ids"].to("cuda")

Instantiate the Streamer object, which we’ll later use for returning the output to users.

model/model.py

        streamer = TextIteratorStreamer(self.tokenizer)
        generation_config = GenerationConfig(
            temperature=1,
            top_p=DEFAULT_TOP_P,
            top_k=40,
        )

When creating the generation parameters, ensure to pass the streamer object that we created previously.

model/model.py

        with torch.no_grad():
            generation_kwargs = {
                "input_ids": input_ids,
                "generation_config": generation_config,
                "return_dict_in_generate": True,
                "output_scores": True,
                "pad_token_id": self.tokenizer.eos_token_id,
                "max_new_tokens": DEFAULT_MAX_NEW_TOKENS,
                "streamer": streamer,
            }

Spawn a thread to run the generation, so that it does not block the main thread.

model/model.py

            thread = Thread(target=self.model.generate, kwargs=generation_kwargs)
            thread.start()

In Truss, the way to achieve streaming output is to return a generator that yields content. In this example, we yield the output of the streamer, which produces output and yields it until the generation is complete.

We define this inner function to create our generator.

model/model.py

            def inner():
                for text in streamer:
                    yield text
                thread.join()

            return inner()

Setting up the config.yaml

Running Falcon 7B requires torch, transformers, and a few other related libraries.

config.yaml

model_name: "LLM with Streaming"
model_metadata:
    example_model_input: {"prompt": "what is the meaning of life"}
requirements:
- torch==2.0.1
- peft==0.4.0
- scipy==1.11.1
- sentencepiece==0.1.99
- accelerate==0.21.0
- bitsandbytes==0.41.1
- einops==0.6.1
- transformers==4.31.0

Configure resources for Falcon

Note that we need an A10G to run this model.

config.yaml

resources:
  cpu: "3"
  memory: 14Gi
  use_gpu: true
  accelerator: A10G

model/model.py

from threading import Thread
from typing import Dict

import torch
from transformers import (
    AutoModelForCausalLM,
    AutoTokenizer,
    GenerationConfig,
    TextIteratorStreamer,
)

CHECKPOINT = "tiiuae/falcon-7b-instruct"
DEFAULT_MAX_NEW_TOKENS = 150
DEFAULT_TOP_P = 0.95


class Model:
    def __init__(self, **kwargs) -> None:
        self.tokenizer = None
        self.model = None

    def load(self):
        self.tokenizer = AutoTokenizer.from_pretrained(CHECKPOINT)
        self.tokenizer.pad_token = self.tokenizer.eos_token_id
        self.model = AutoModelForCausalLM.from_pretrained(
            CHECKPOINT,
            torch_dtype=torch.bfloat16,
            trust_remote_code=True,
            device_map="auto",
        )

    def predict(self, request: Dict) -> Dict:
        prompt = request.pop("prompt")
        inputs = self.tokenizer(
            prompt, return_tensors="pt", max_length=512, truncation=True, padding=True
        )
        input_ids = inputs["input_ids"].to("cuda")

        streamer = TextIteratorStreamer(self.tokenizer)
        generation_config = GenerationConfig(
            temperature=1,
            top_p=DEFAULT_TOP_P,
            top_k=40,
        )

        with torch.no_grad():
            generation_kwargs = {
                "input_ids": input_ids,
                "generation_config": generation_config,
                "return_dict_in_generate": True,
                "output_scores": True,
                "pad_token_id": self.tokenizer.eos_token_id,
                "max_new_tokens": DEFAULT_MAX_NEW_TOKENS,
                "streamer": streamer,
            }

            thread = Thread(target=self.model.generate, kwargs=generation_kwargs)
            thread.start()

            def inner():
                for text in streamer:
                    yield text
                thread.join()

            return inner()