Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Vectorize asynchronously with ThreadPoolExecutor #56

Merged
merged 2 commits into from
Apr 21, 2023
Merged

Vectorize asynchronously with ThreadPoolExecutor #56

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1c39258
vectorize asynchronously with ThreadPoolExecutor
alexcannan Apr 21, 2023
4527559
unbounded executor workers
alexcannan Apr 21, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion app.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,7 @@ def startup_event():

@app.get("/.well-known/live", response_class=Response)
@app.get("/.well-known/ready", response_class=Response)
def live_and_ready(response: Response):
async def live_and_ready(response: Response):
response.status_code = status.HTTP_204_NO_CONTENT


Expand Down
12 changes: 10 additions & 2 deletions vectorizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,5 @@
import asyncio
from concurrent.futures import ThreadPoolExecutor
import math
from typing import Optional
import torch
Expand Down Expand Up @@ -33,6 +35,7 @@ class Vectorizer:
cuda_core: str
model_type: str
direct_tokenize: bool
executor: ThreadPoolExecutor

def __init__(self, model_path: str, cuda_support: bool, cuda_core: str, cuda_per_process_memory_fraction: float, model_type: str, architecture: str, direct_tokenize: bool):
self.cuda = cuda_support
Expand All @@ -52,8 +55,10 @@ def __init__(self, model_path: str, cuda_support: bool, cuda_core: str, cuda_per

self.tokenizer = self.model_delegate.create_tokenizer(model_path)

self.executor = ThreadPoolExecutor()

def tokenize(self, text:str):
return self.tokenizer(text, padding=True, truncation=True, max_length=500,
return self.tokenizer(text, padding=True, truncation=True, max_length=500,
add_special_tokens = True, return_tensors="pt")

def get_embeddings(self, batch_results):
Expand All @@ -65,7 +70,7 @@ def get_batch_results(self, tokens, text):
def pool_embedding(self, batch_results, tokens, config):
return self.model_delegate.pool_embedding(batch_results, tokens, config)

async def vectorize(self, text: str, config: VectorInputConfig):
def _vectorize(self, text: str, config: VectorInputConfig):
with torch.no_grad():
if self.direct_tokenize:
# create embeddings without tokenizing text
Expand All @@ -92,6 +97,9 @@ async def vectorize(self, text: str, config: VectorInputConfig):
batch_sum_vectors += self.pool_embedding(batch_results, tokens, config)
return batch_sum_vectors.detach() / num_sentences

async def vectorize(self, text: str, config: VectorInputConfig):
return await asyncio.wrap_future(self.executor.submit(self._vectorize, text, config))


class HFModel:

Expand Down