File size: 6,209 Bytes
7886788 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 |
#! /bin/sh
S=Xunzi-Qwen1.5-4B
U=UD_Classical_Chinese-Kyoto
test -d $U || git clone --depth=1 https://github.com/UniversalDependencies/$U
for F in train dev test
do cp $U/*-$F.conllu $F.conllu
done
test -d $S || git clone --depth=1 https://www.modelscope.cn/Xunzillm4cc/$S.git
TMP=./maker$$.py
( echo '#! /usr/bin/env deepspeed'
echo 'src="'$S'"'
echo 'tgt="KoichiYasuoka/'$S'-upos"'
) > $TMP
cat << 'EOF' >> $TMP
from transformers import AutoTokenizer,Qwen2Model,Qwen2PreTrainedModel,AutoConfig,DataCollatorForTokenClassification,TrainingArguments,Trainer
from transformers.modeling_outputs import TokenClassifierOutput
class Qwen2ForTokenClassification(Qwen2PreTrainedModel):
def __init__(self,config):
from torch import nn
super().__init__(config)
self.num_labels=config.num_labels
self.model=Qwen2Model(config)
if getattr(config,"classifier_dropout",None) is not None:
classifier_dropout=config.classifier_dropout
elif getattr(config,"hidden_dropout",None) is not None:
classifier_dropout=config.hidden_dropout
else:
classifier_dropout=0.1
self.dropout=nn.Dropout(classifier_dropout)
self.score=nn.Linear(config.hidden_size,config.num_labels)
self.post_init()
def get_input_embeddings(self):
return self.model.embed_tokens
def set_input_embeddings(self,value):
self.model.embed_tokens=value
def forward(self,input_ids=None,past_key_values=None,attention_mask=None,position_ids=None,inputs_embeds=None,labels=None,use_cache=None,output_attentions=None,output_hidden_states=None,return_dict=None):
return_dict=return_dict if return_dict is not None else self.config.use_return_dict
outputs=self.model(input_ids,past_key_values=past_key_values,attention_mask=attention_mask,position_ids=position_ids,inputs_embeds=inputs_embeds,use_cache=use_cache,output_attentions=output_attentions,output_hidden_states=output_hidden_states,return_dict=return_dict)
sequence_output=outputs[0]
sequence_output=self.dropout(sequence_output)
logits=self.score(sequence_output)
loss=None
if labels is not None:
from torch import nn
loss_fct=nn.CrossEntropyLoss()
loss=loss_fct(logits.view(-1,self.num_labels),labels.view(-1))
if not return_dict:
output=(logits,)+outputs[2:]
return ((loss,)+output) if loss is not None else output
return TokenClassifierOutput(loss=loss,logits=logits,hidden_states=outputs.hidden_states,attentions=outputs.attentions)
class UPOSFileDataset(object):
def __init__(self,conllu,tokenizer):
self.conllu=open(conllu,"r",encoding="utf-8")
self.tokenizer=tokenizer
self.seeks=[0]
self.multiword={}
label=set(["SYM"])
s=self.conllu.readline()
while s!="":
if s=="\n":
self.seeks.append(self.conllu.tell())
else:
w=s.split("\t")
if len(w)==10:
if w[0].isdecimal():
label.add(w[3] if w[5]=="_" else w[3]+"|"+w[5])
elif w[0].find("-")>0:
t=w[0].split("-")
f,j,k=w[1],[],[]
for i in range(int(t[0]),int(t[1])+1):
w=self.conllu.readline().split("\t")
j.append(w[3] if w[5]=="_" else w[3]+"|"+w[5])
k.append(w[1])
p="+".join(j)
label.add(p)
if p in self.multiword:
self.multiword[p][f]=list(k)
else:
self.multiword[p]={f:list(k)}
s=self.conllu.readline()
lid={}
for i,l in enumerate(sorted(label)):
lid[l],lid["B-"+l],lid["I-"+l]=i*3,i*3+1,i*3+2
self.label2id=lid
def __call__(*args):
lid={l:i for i,l in enumerate(sorted(set(sum([list(t.label2id) for t in args],[]))))}
for t in args:
t.label2id=lid
return lid
def __del__(self):
self.conllu.close()
__len__=lambda self:len(self.seeks)-1
def __getitem__(self,i):
self.conllu.seek(self.seeks[i])
form,upos=[],[]
while self.conllu.tell()<self.seeks[i+1]:
w=self.conllu.readline().split("\t")
if len(w)==10:
form.append(w[1])
if w[0].isdecimal():
upos.append(w[3] if w[5]=="_" else w[3]+"|"+w[5])
elif w[0].find("-")>0:
t=w[0].split("-")
u=[]
for j in range(int(t[0]),int(t[1])+1):
k=self.conllu.readline().split("\t")
u.append(k[3] if k[5]=="_" else k[3]+"|"+k[5])
upos.append("+".join(u))
v=self.tokenizer(form,add_special_tokens=False)
i,u=[],[]
for j,(x,y) in enumerate(zip(v["input_ids"],upos)):
if x!=[]:
i+=x
u+=[y] if len(x)==1 else ["B-"+y]+["I-"+y]*(len(x)-1)
if len(i)<self.tokenizer.model_max_length-3:
ids=i
upos=u
else:
ids=i[0:self.tokenizer.model_max_length-2]
upos=u[0:self.tokenizer.model_max_length-2]
return {"input_ids":ids,"labels":[self.label2id[t] for t in upos]}
tkz=AutoTokenizer.from_pretrained(src)
trainDS=UPOSFileDataset("train.conllu",tkz)
devDS=UPOSFileDataset("dev.conllu",tkz)
testDS=UPOSFileDataset("test.conllu",tkz)
lid=trainDS(devDS,testDS)
cfg=AutoConfig.from_pretrained(src,num_labels=len(lid),label2id=lid,id2label={i:l for l,i in lid.items()},ignore_mismatched_sizes=True)
dsp={"fp16":{"enabled":"auto"},"optimizer":{"type":"AdamW"},"scheduler":{"type":"WarmupLR","params":{}},"train_batch_size":"auto","train_micro_batch_size_per_gpu":"auto","zero_optimization":{"stage":3,"offload_optimizer":{"device":"cpu","pin_memory":True},"offload_param":{"device":"cpu","pin_memory":True},"overlap_comm":True,"contiguous_gradients":True,"reduce_bucket_size":"auto","stage3_prefetch_bucket_size":"auto","stage3_param_persistence_threshold":"auto","stage3_gather_16bit_weights_on_model_save":True}}
arg=TrainingArguments(num_train_epochs=3,per_device_train_batch_size=16,deepspeed=dsp,output_dir=tgt,overwrite_output_dir=True,save_total_limit=2,learning_rate=5e-05,warmup_ratio=0.1,save_safetensors=False)
trn=Trainer(args=arg,data_collator=DataCollatorForTokenClassification(tkz),model=Qwen2ForTokenClassification.from_pretrained(src,config=cfg,ignore_mismatched_sizes=True),train_dataset=trainDS)
trn.train()
trn.save_model(tgt)
tkz.save_pretrained(tgt)
EOF
chmod 755 $TMP
$TMP
exit
|