ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Enable fastchat benchmark latency (#11017)

Browse source 
* enable fastchat benchmark

* add readme

* update readme

* update
This commit is contained in:
Wang, Jian4 2024-05-15 14:52:09 +08:00 committed by GitHub
parent 93d40ab127
commit 2084ebe4ee
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 9 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
python/llm/src/ipex_llm/serving/fastchat/ipex_llm_worker.py
View file

@ -71,6 +71,7 @@ class BigDLLLMWorker(BaseModelWorker):
speculative: bool = False, speculative: bool = False,
load_low_bit_model: bool = False, load_low_bit_model: bool = False,
stream_interval: int = 4, stream_interval: int = 4,
benchmark: str = "true",
): ):
super().__init__( super().__init__(
controller_addr, controller_addr,
@ -103,6 +104,10 @@ class BigDLLLMWorker(BaseModelWorker):
speculative, speculative,
load_low_bit_model, load_low_bit_model,
) )
if benchmark.lower() == "true":
from ipex_llm.utils.benchmark_util import BenchmarkWrapper
self.model = BenchmarkWrapper(self.model, do_print=True)
logger.info(f"enable benchmark successfully")
self.stream_interval = stream_interval self.stream_interval = stream_interval
self.context_len = get_context_length(self.model.config) self.context_len = get_context_length(self.model.config)
self.embed_in_truncate = embed_in_truncate self.embed_in_truncate = embed_in_truncate
@ -495,6 +500,9 @@ if __name__ == "__main__":
default=False, default=False,
help="To use self-speculative or not", help="To use self-speculative or not",
) )
parser.add_argument(
"--benchmark", type=str, default="true", help="To print model generation latency or not"
)
parser.add_argument( parser.add_argument(
"--trust-remote-code", "--trust-remote-code",
action="store_true", action="store_true",
@ -527,5 +535,6 @@ if __name__ == "__main__":
args.speculative, args.speculative,
args.load_low_bit_model, args.load_low_bit_model,
args.stream_interval, args.stream_interval,
args.benchmark,
) )
uvicorn.run(app, host=args.host, port=args.port, log_level="info") uvicorn.run(app, host=args.host, port=args.port, log_level="info")