Skip to content

[Perf] Implement async token counting in proxy server #1

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[Perf] Implement async token counting in proxy server #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
27 changes: 19 additions & 8 deletions litellm/proxy/proxy_server.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5235,8 +5235,12 @@ async def moderations(
response_model=TokenCountResponse,
)
async def token_counter(request: TokenCountRequest):
""" """
"""
Count tokens for a given model and text/messages.
Uses thread pool to avoid blocking the event loop with CPU-intensive tokenization.
"""
from litellm import token_counter
from litellm.litellm_core_utils.asyncify import asyncify

global llm_router

Expand Down Expand Up @@ -5277,13 +5281,20 @@ async def token_counter(request: TokenCountRequest):
model=model_to_use, custom_tokenizer=custom_tokenizer
)

tokenizer_used = str(_tokenizer_used["type"])
total_tokens = token_counter(
model=model_to_use,
text=prompt,
messages=messages,
custom_tokenizer=_tokenizer_used, # type: ignore
)
# Define a synchronous function to do the tokenization
def _count_tokens():
tokenizer_used = str(_tokenizer_used["type"])
total_tokens = token_counter(
model=model_to_use,
text=prompt,
messages=messages,
custom_tokenizer=_tokenizer_used, # type: ignore
)
return total_tokens, tokenizer_used

# Offload the CPU-intensive tokenization to a thread pool
total_tokens, tokenizer_used = await asyncify(_count_tokens)()

return TokenCountResponse(
total_tokens=total_tokens,
request_model=request.model,
Expand Down