Kaynağa Gözat

Perf: set timeout for building chunks. (#8940)

### What problem does this PR solve?


### Type of change

- [x] Performance Improvement
tags/v0.20.0
Kevin Hu 3 ay önce
ebeveyn
işleme
c783d90ba3
No account linked to committer's email address
3 değiştirilmiş dosya ile 14 ekleme ve 10 silme
  1. 8
    6
      api/utils/api_utils.py
  2. 1
    4
      graphrag/general/index.py
  3. 5
    0
      rag/svr/task_executor.py

+ 8
- 6
api/utils/api_utils.py Dosyayı Görüntüle

@@ -676,12 +676,14 @@ async def is_strong_enough(chat_model, embedding_model):
@timeout(30, 2)
async def _is_strong_enough():
nonlocal chat_model, embedding_model
with trio.fail_after(3):
_ = await trio.to_thread.run_sync(lambda: embedding_model.encode(["Are you strong enough!?"]))
with trio.fail_after(30):
res = await trio.to_thread.run_sync(lambda: chat_model.chat("Nothing special.", [{"role":"user", "content": "Are you strong enough!?"}], {}))
if res.find("**ERROR**") >= 0:
raise Exception(res)
if embedding_model:
with trio.fail_after(3):
_ = await trio.to_thread.run_sync(lambda: embedding_model.encode(["Are you strong enough!?"]))
if chat_model:
with trio.fail_after(30):
res = await trio.to_thread.run_sync(lambda: chat_model.chat("Nothing special.", [{"role":"user", "content": "Are you strong enough!?"}], {}))
if res.find("**ERROR**") >= 0:
raise Exception(res)

# Pressure test for GraphRAG task
async with trio.open_nursery() as nursery:

+ 1
- 4
graphrag/general/index.py Dosyayı Görüntüle

@@ -20,7 +20,7 @@ import trio

from api import settings
from api.utils import get_uuid
from api.utils.api_utils import timeout, is_strong_enough
from api.utils.api_utils import timeout
from graphrag.light.graph_extractor import GraphExtractor as LightKGExt
from graphrag.general.graph_extractor import GraphExtractor as GeneralKGExt
from graphrag.general.community_reports_extractor import CommunityReportsExtractor
@@ -49,9 +49,6 @@ async def run_graphrag(
embedding_model,
callback,
):
# Pressure test for GraphRAG task
await is_strong_enough(chat_model, embedding_model)

start = trio.current_time()
tenant_id, kb_id, doc_id = row["tenant_id"], str(row["kb_id"]), row["doc_id"]
chunks = []

+ 5
- 0
rag/svr/task_executor.py Dosyayı Görüntüle

@@ -184,6 +184,7 @@ def set_progress(task_id, from_page=0, to_page=-1, prog=None, msg="Processing...
except Exception:
logging.exception(f"set_progress({task_id}), progress: {prog}, progress_msg: {msg}, got exception")


async def collect():
global CONSUMER_NAME, DONE_TASKS, FAILED_TASKS
global UNACKED_ITERATOR
@@ -229,6 +230,7 @@ async def get_storage_binary(bucket, name):
return await trio.to_thread.run_sync(lambda: STORAGE_IMPL.get(bucket, name))


@timeout(60*40, 1)
async def build_chunks(task, progress_callback):
if task["size"] > DOC_MAXIMUM_SIZE:
set_progress(task["id"], prog=-1, msg="File size exceeds( <= %dMb )" %
@@ -541,6 +543,7 @@ async def do_handle_task(task):
try:
# bind embedding model
embedding_model = LLMBundle(task_tenant_id, LLMType.EMBEDDING, llm_name=task_embedding_id, lang=task_language)
await is_strong_enough(None, embedding_model)
vts, _ = embedding_model.encode(["ok"])
vector_size = len(vts[0])
except Exception as e:
@@ -555,6 +558,7 @@ async def do_handle_task(task):
if task.get("task_type", "") == "raptor":
# bind LLM for raptor
chat_model = LLMBundle(task_tenant_id, LLMType.CHAT, llm_name=task_llm_id, lang=task_language)
await is_strong_enough(chat_model, None)
# run RAPTOR
async with kg_limiter:
chunks, token_count = await run_raptor(task, chat_model, embedding_model, vector_size, progress_callback)
@@ -566,6 +570,7 @@ async def do_handle_task(task):
graphrag_conf = task["kb_parser_config"].get("graphrag", {})
start_ts = timer()
chat_model = LLMBundle(task_tenant_id, LLMType.CHAT, llm_name=task_llm_id, lang=task_language)
await is_strong_enough(chat_model, None)
with_resolution = graphrag_conf.get("resolution", False)
with_community = graphrag_conf.get("community", False)
async with kg_limiter:

Loading…
İptal
Kaydet