remove prefillprogress event (#1550)
this should never have been a separate event, but i didnt quite communicate that well when this was merged. convert PrefillProgress to a chunk like the rest of the runner responses. tested with Llama-3.3-70B, prefill progress events still show up in the dashboard as usual
This commit is contained in:
@@ -138,7 +138,6 @@ from exo.shared.types.events import (
|
||||
Event,
|
||||
ForwarderEvent,
|
||||
IndexedEvent,
|
||||
PrefillProgress,
|
||||
TracesMerged,
|
||||
)
|
||||
from exo.shared.types.memory import Memory
|
||||
@@ -1455,22 +1454,6 @@ class API:
|
||||
await queue.send(event.chunk)
|
||||
except BrokenResourceError:
|
||||
self._text_generation_queues.pop(event.command_id, None)
|
||||
|
||||
elif isinstance(event, PrefillProgress):
|
||||
if queue := self._text_generation_queues.get(
|
||||
event.command_id, None
|
||||
):
|
||||
try:
|
||||
await queue.send(
|
||||
PrefillProgressChunk(
|
||||
model=event.model,
|
||||
processed_tokens=event.processed_tokens,
|
||||
total_tokens=event.total_tokens,
|
||||
)
|
||||
)
|
||||
except BrokenResourceError:
|
||||
self._text_generation_queues.pop(event.command_id, None)
|
||||
|
||||
if isinstance(event, TracesMerged):
|
||||
self._save_merged_trace(event)
|
||||
|
||||
|
||||
@@ -15,7 +15,6 @@ from exo.shared.types.events import (
|
||||
NodeDownloadProgress,
|
||||
NodeGatheredInfo,
|
||||
NodeTimedOut,
|
||||
PrefillProgress,
|
||||
RunnerDeleted,
|
||||
RunnerStatusUpdated,
|
||||
TaskAcknowledged,
|
||||
@@ -65,7 +64,6 @@ def event_apply(event: Event, state: State) -> State:
|
||||
| ChunkGenerated()
|
||||
| TaskAcknowledged()
|
||||
| InputChunkReceived()
|
||||
| PrefillProgress()
|
||||
| TracesCollected()
|
||||
| TracesMerged()
|
||||
): # Pass-through events that don't modify state
|
||||
|
||||
@@ -5,7 +5,7 @@ from pydantic import Field
|
||||
|
||||
from exo.shared.topology import Connection
|
||||
from exo.shared.types.chunks import GenerationChunk, InputImageChunk
|
||||
from exo.shared.types.common import CommandId, Id, ModelId, NodeId, SessionId
|
||||
from exo.shared.types.common import CommandId, Id, NodeId, SessionId
|
||||
from exo.shared.types.tasks import Task, TaskId, TaskStatus
|
||||
from exo.shared.types.worker.downloads import DownloadProgress
|
||||
from exo.shared.types.worker.instances import Instance, InstanceId
|
||||
@@ -102,13 +102,6 @@ class InputChunkReceived(BaseEvent):
|
||||
chunk: InputImageChunk
|
||||
|
||||
|
||||
class PrefillProgress(BaseEvent):
|
||||
command_id: CommandId
|
||||
model: ModelId
|
||||
processed_tokens: int
|
||||
total_tokens: int
|
||||
|
||||
|
||||
class TopologyEdgeCreated(BaseEvent):
|
||||
conn: Connection
|
||||
|
||||
@@ -155,7 +148,6 @@ Event = (
|
||||
| NodeDownloadProgress
|
||||
| ChunkGenerated
|
||||
| InputChunkReceived
|
||||
| PrefillProgress
|
||||
| TopologyEdgeCreated
|
||||
| TopologyEdgeDeleted
|
||||
| TracesCollected
|
||||
|
||||
@@ -21,12 +21,17 @@ from exo.shared.constants import EXO_MAX_CHUNK_SIZE, EXO_TRACING_ENABLED
|
||||
from exo.shared.models.model_cards import ModelId, ModelTask
|
||||
from exo.shared.tracing import clear_trace_buffer, get_trace_buffer
|
||||
from exo.shared.types.api import ImageGenerationStats
|
||||
from exo.shared.types.chunks import ErrorChunk, ImageChunk, TokenChunk, ToolCallChunk
|
||||
from exo.shared.types.chunks import (
|
||||
ErrorChunk,
|
||||
ImageChunk,
|
||||
PrefillProgressChunk,
|
||||
TokenChunk,
|
||||
ToolCallChunk,
|
||||
)
|
||||
from exo.shared.types.common import CommandId
|
||||
from exo.shared.types.events import (
|
||||
ChunkGenerated,
|
||||
Event,
|
||||
PrefillProgress,
|
||||
RunnerStatusUpdated,
|
||||
TaskAcknowledged,
|
||||
TaskStatusUpdated,
|
||||
@@ -315,11 +320,13 @@ def main(
|
||||
) -> None:
|
||||
if device_rank == 0:
|
||||
event_sender.send(
|
||||
PrefillProgress(
|
||||
ChunkGenerated(
|
||||
command_id=command_id,
|
||||
model=shard_metadata.model_card.model_id,
|
||||
processed_tokens=processed,
|
||||
total_tokens=total,
|
||||
chunk=PrefillProgressChunk(
|
||||
model=shard_metadata.model_card.model_id,
|
||||
processed_tokens=processed,
|
||||
total_tokens=total,
|
||||
),
|
||||
)
|
||||
)
|
||||
cancelled_tasks.update(cancel_receiver.collect())
|
||||
|
||||
Reference in New Issue
Block a user