server: remove warmup methods everywhere

server/reflector/processors/audio_transcript_auto.py

@@ -52,9 +52,6 @@ class AudioTranscriptAutoProcessor(AudioTranscriptProcessor):
     def off(self, callback):
         self.processor.off(callback)
 
-    async def _warmup(self):
-        return await self.processor._warmup()
-
     async def _push(self, data: AudioFile):
         return await self.processor._push(data)
 

server/reflector/processors/audio_transcript_modal.py

@@ -12,10 +12,7 @@ API will be a POST request to TRANSCRIPT_URL:
 
 """
 
-from time import monotonic
-
 import httpx
-
 from reflector.processors.audio_transcript import AudioTranscriptProcessor
 from reflector.processors.audio_transcript_auto import AudioTranscriptAutoProcessor
 from reflector.processors.types import AudioFile, Transcript, Word
@@ -27,26 +24,9 @@ class AudioTranscriptModalProcessor(AudioTranscriptProcessor):
     def __init__(self, modal_api_key: str):
         super().__init__()
         self.transcript_url = settings.TRANSCRIPT_URL + "/transcribe"
-        self.warmup_url = settings.TRANSCRIPT_URL + "/warmup"
         self.timeout = settings.TRANSCRIPT_TIMEOUT
         self.headers = {"Authorization": f"Bearer {modal_api_key}"}
 
-    async def _warmup(self):
-        try:
-            async with httpx.AsyncClient() as client:
-                start = monotonic()
-                self.logger.debug("Transcribe modal: warming up...")
-                response = await client.post(
-                    self.warmup_url,
-                    headers=self.headers,
-                    timeout=self.timeout,
-                )
-                response.raise_for_status()
-                duration = monotonic() - start
-                self.logger.debug(f"Transcribe modal: warmup took {duration:.2f}s")
-        except Exception:
-            self.logger.exception("Transcribe modal: warmup failed")
-
     async def _transcript(self, data: AudioFile):
         async with httpx.AsyncClient() as client:
             self.logger.debug(f"Try to transcribe audio {data.name}")

server/reflector/processors/base.py

@@ -5,7 +5,6 @@ from uuid import uuid4
 
 from prometheus_client import Counter, Gauge, Histogram
 from pydantic import BaseModel
-
 from reflector.logger import logger
 
 
@@ -18,7 +17,6 @@ class PipelineEvent(BaseModel):
 class Processor:
     INPUT_TYPE: type = None
     OUTPUT_TYPE: type = None
-    WARMUP_EVENT: str = "WARMUP_EVENT"
 
     m_processor = Histogram(
         "processor",
@@ -172,21 +170,12 @@ class Processor:
     def describe(self, level=0):
         logger.info("  " * level + self.__class__.__name__)
 
-    async def warmup(self):
-        """
-        Warmup the processor
-        """
-        await self._warmup()
-
     async def _push(self, data):
         raise NotImplementedError
 
     async def _flush(self):
         pass
 
-    async def _warmup(self):
-        pass
-
     @classmethod
     def as_threaded(cls, *args, **kwargs):
         """
@@ -242,12 +231,6 @@ class ThreadedProcessor(Processor):
                     if data is None:
                         await self.processor.flush()
                         break
-                    if data == self.WARMUP_EVENT:
-                        self.logger.debug(
-                            f"Warming up {self.processor.__class__.__name__}"
-                        )
-                        await self.processor.warmup()
-                        continue
                     try:
                         await self.processor.push(data)
                     except Exception:
@@ -258,9 +241,6 @@ class ThreadedProcessor(Processor):
                 finally:
                     self.queue.task_done()
 
-    async def _warmup(self):
-        await self.queue.put(self.WARMUP_EVENT)
-
     async def _push(self, data):
         await self.queue.put(data)
 
@@ -309,10 +289,6 @@ class BroadcastProcessor(Processor):
         for processor in self.processors:
             processor.set_pipeline(pipeline)
 
-    async def _warmup(self):
-        for processor in self.processors:
-            await processor.warmup()
-
     async def _push(self, data):
         for processor in self.processors:
             await processor.push(data)
@@ -352,7 +328,6 @@ class Pipeline(Processor):
     OUTPUT_TYPE = None
 
     def __init__(self, *processors: Processor):
-        self._warmed_up = False
         super().__init__()
         self.logger = logger.bind(pipeline=self.uid)
         self.logger.info("Pipeline created")
@@ -369,11 +344,6 @@ class Pipeline(Processor):
         self.INPUT_TYPE = processors[0].INPUT_TYPE
         self.OUTPUT_TYPE = processors[-1].OUTPUT_TYPE
 
-    async def _warmup(self):
-        for processor in self.processors:
-            self.logger.debug(f"Warming up {processor.__class__.__name__}")
-            await processor.warmup()
-
     async def _push(self, data):
         await self.processors[0].push(data)
 

server/reflector/processors/transcript_topic_detector.py

@@ -22,9 +22,6 @@ class TranscriptTopicDetectorProcessor(Processor):
         self.llm = LLM.get_instance()
         self.params = LLMTaskParams.get_instance(self.TASK).task_params
 
-    async def _warmup(self):
-        await self.llm.warmup(logger=self.logger)
-
     async def _push(self, data: Transcript):
         if self.transcript is None:
             self.transcript = data

server/reflector/processors/transcript_translator.py

@@ -1,5 +1,3 @@
-from time import monotonic
-
 import httpx
 from reflector.processors.base import Processor
 from reflector.processors.types import Transcript, TranslationLanguages
@@ -22,22 +20,6 @@ class TranscriptTranslatorProcessor(Processor):
         self.timeout = settings.TRANSCRIPT_TIMEOUT
         self.headers = {"Authorization": f"Bearer {settings.LLM_MODAL_API_KEY}"}
 
-    async def _warmup(self):
-        try:
-            async with httpx.AsyncClient() as client:
-                start = monotonic()
-                self.logger.debug("Translate modal: warming up...")
-                response = await client.post(
-                    settings.TRANSCRIPT_URL + "/warmup",
-                    headers=self.headers,
-                    timeout=self.timeout,
-                )
-                response.raise_for_status()
-                duration = monotonic() - start
-                self.logger.debug(f"Translate modal: warmup took {duration:.2f}s")
-        except Exception:
-            self.logger.exception("Translate modal: warmup failed")
-
     async def _push(self, data: Transcript):
         self.transcript = data
         await self.flush()