4b02ff188b
* add prompt format and stopping_words for qwen mdoel * performance optimization * optimize * update * meet comments
134 lines
3.6 KiB
Python
134 lines
3.6 KiB
Python
#
|
|
# Copyright 2016 The BigDL Authors.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
# This file is adapted from
|
|
# https://github.com/oobabooga/text-generation-webui/blob/main/modules/callbacks.py
|
|
|
|
|
|
import gc
|
|
import traceback
|
|
from queue import Queue
|
|
from threading import Thread
|
|
|
|
import torch
|
|
import transformers
|
|
from transformers import is_torch_xpu_available
|
|
|
|
import modules.shared as shared
|
|
|
|
|
|
class StopNowException(Exception):
|
|
pass
|
|
|
|
|
|
class _StopEverythingStoppingCriteria(transformers.StoppingCriteria):
|
|
def __init__(self):
|
|
transformers.StoppingCriteria.__init__(self)
|
|
|
|
def __call__(self, input_ids: torch.LongTensor, _scores: torch.FloatTensor) -> bool:
|
|
return shared.stop_everything
|
|
|
|
|
|
class StopWordsCriteria(transformers.StoppingCriteria):
|
|
"""Custom `StoppingCriteria` which checks if all generated functions in the batch are completed."""
|
|
def __init__(self, stop_words, tokenizer):
|
|
self.stop_words = stop_words
|
|
self.tokenizer = tokenizer
|
|
|
|
def __call__(self, input_ids, scores, **kwargs):
|
|
"""Returns true if all generated sequences contain any of the end-of-function strings."""
|
|
text = self.tokenizer.decode(input_ids[-1][-1])
|
|
return text in self.stop_words
|
|
|
|
|
|
class Stream(transformers.StoppingCriteria):
|
|
def __init__(self, callback_func=None):
|
|
self.callback_func = callback_func
|
|
|
|
def __call__(self, input_ids, scores) -> bool:
|
|
if self.callback_func is not None:
|
|
self.callback_func(input_ids[0])
|
|
|
|
return False
|
|
|
|
|
|
class Iteratorize:
|
|
|
|
"""
|
|
Transforms a function that takes a callback
|
|
into a lazy iterator (generator).
|
|
|
|
Adapted from: https://stackoverflow.com/a/9969000
|
|
"""
|
|
|
|
def __init__(self, func, args=None, kwargs=None, callback=None):
|
|
self.mfunc = func
|
|
self.c_callback = callback
|
|
self.q = Queue()
|
|
self.sentinel = object()
|
|
self.args = args or []
|
|
self.kwargs = kwargs or {}
|
|
self.stop_now = False
|
|
|
|
def _callback(val):
|
|
if self.stop_now or shared.stop_everything:
|
|
raise StopNowException
|
|
self.q.put(val)
|
|
|
|
def gentask():
|
|
try:
|
|
ret = self.mfunc(callback=_callback, *args, **self.kwargs)
|
|
except StopNowException:
|
|
pass
|
|
except:
|
|
traceback.print_exc()
|
|
pass
|
|
|
|
clear_torch_cache()
|
|
self.q.put(self.sentinel)
|
|
if self.c_callback:
|
|
self.c_callback(ret)
|
|
|
|
self.thread = Thread(target=gentask)
|
|
self.thread.start()
|
|
|
|
def __iter__(self):
|
|
return self
|
|
|
|
def __next__(self):
|
|
obj = self.q.get(True, None)
|
|
if obj is self.sentinel:
|
|
raise StopIteration
|
|
else:
|
|
return obj
|
|
|
|
def __del__(self):
|
|
clear_torch_cache()
|
|
|
|
def __enter__(self):
|
|
return self
|
|
|
|
def __exit__(self, exc_type, exc_val, exc_tb):
|
|
self.stop_now = True
|
|
clear_torch_cache()
|
|
|
|
|
|
def clear_torch_cache():
|
|
gc.collect()
|
|
if not shared.args.cpu:
|
|
if is_torch_xpu_available():
|
|
torch.xpu.empty_cache()
|
|
else:
|
|
torch.cuda.empty_cache()
|