lm-deluge 0.0.42__tar.gz → 0.0.44__tar.gz

{lm_deluge-0.0.42/src/lm_deluge.egg-info → lm_deluge-0.0.44}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lm_deluge
-Version: 0.0.42
+Version: 0.0.44
 Summary: Python utility for using LLM API models.
 Author-email: Benjamin Anderson <ben@trytaylor.ai>
 Requires-Python: >=3.10

{lm_deluge-0.0.42 → lm_deluge-0.0.44}/pyproject.toml

@@ -3,7 +3,7 @@ requires = ["setuptools", "wheel"]
 
 [project]
 name = "lm_deluge"
-version = "0.0.42"
+version = "0.0.44"
 authors = [{ name = "Benjamin Anderson", email = "ben@trytaylor.ai" }]
 description = "Python utility for using LLM API models."
 readme = "README.md"

{lm_deluge-0.0.42 → lm_deluge-0.0.44}/src/lm_deluge/client.py

@@ -1,5 +1,4 @@
 import asyncio
-import random
 from typing import Any, AsyncGenerator, Callable, Literal, Self, Sequence, overload
 
 import numpy as np
@@ -105,14 +104,8 @@ class _LLMClient(BaseModel):
 
     def _get_tracker(self) -> StatusTracker:
         if self._tracker is None:
-            self._tracker = StatusTracker(
-                max_requests_per_minute=self.max_requests_per_minute,
-                max_tokens_per_minute=self.max_tokens_per_minute,
-                max_concurrent_requests=self.max_concurrent_requests,
-                use_progress_bar=False,
-                progress_bar_disable=True,
-                progress_style=self.progress,
-            )
+            self.open()
+            assert self._tracker, "should have tracker now"
         return self._tracker
 
     @property
@@ -225,16 +218,16 @@ class _LLMClient(BaseModel):
     ):
         while True:
             async with self._capacity_lock:
-                tracker.update_capacity()
                 if tracker.check_capacity(num_tokens, retry=retry):
                     tracker.set_limiting_factor(None)
                     return
                 seconds_to_pause = tracker.seconds_to_pause
 
             if seconds_to_pause > 0:
+                print(f"Pausing for {seconds_to_pause} seconds to cool down.")
                 await asyncio.sleep(seconds_to_pause)
             else:
-                await asyncio.sleep(random.random())
+                await asyncio.sleep(30.0 / self.max_requests_per_minute)
 
     async def _execute_request(self, context: RequestContext) -> APIResponse:
         """Create and send a single API request using the provided context."""

{lm_deluge-0.0.42 → lm_deluge-0.0.44/src/lm_deluge.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lm_deluge
-Version: 0.0.42
+Version: 0.0.44
 Summary: Python utility for using LLM API models.
 Author-email: Benjamin Anderson <ben@trytaylor.ai>
 Requires-Python: >=3.10