-
Notifications
You must be signed in to change notification settings - Fork 2
/
eval.py
162 lines (131 loc) · 5.38 KB
/
eval.py
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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
import torch
from transformers import AutoTokenizer, AutoModel,AutoModelForCausalLM, LlamaTokenizer, LlamaForCausalLM,T5Tokenizer, T5ForConditionalGeneration
import torch
import argparse
import json
from tqdm import tqdm
import os
import copy
def get_model_tokenizer(model_path):
if 'llama' in model_path.lower():
tokenizer = LlamaTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path,torch_dtype=torch.float16, device_map="auto")
else:
tokenizer = AutoTokenizer.from_pretrained(model_path,trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_path,torch_dtype=torch.float16, device_map="auto",trust_remote_code=True)
model.eval()
return model,tokenizer
parser = argparse.ArgumentParser()
parser.add_argument('--model_path',
default="",
required=False)
parser.add_argument('--res_path',
default="",
required=False)
parser.add_argument('--rank_path',
default="",
required=False)
parser.add_argument('--data_name',
default='msmarco')
args = parser.parse_args()
model_path=args.model_path
data_name=args.data_name
IGNORE_INDEX = -100
DEFAULT_PAD_TOKEN = "[PAD]"
DEFAULT_EOS_TOKEN = "</s>"
DEFAULT_BOS_TOKEN = "</s>"
DEFAULT_UNK_TOKEN = "</s>"
bsz=8
prompt='Document: {doc} Query:'
model,tokenizer=get_model_tokenizer(model_path)
if 'qwen' in model_path.lower():
tokenizer.pad_token_id = tokenizer.eod_id
def get_num_token(text):
return len(tokenizer.encode(text))
prompt_len=get_num_token(prompt)
print(f"prompt_len: {prompt_len}")
def truncation(text,length):
text=tokenizer.decode(tokenizer.encode(text,max_length=length, add_special_tokens=False))
return text
def _tokenize_fn(strings):
"""Tokenize a list of strings."""
tokenized_list = [
tokenizer(
text,
return_tensors="pt",
padding="longest",
)['input_ids']
for text in strings
]
input_ids = labels = [tokenized[0] for tokenized in tokenized_list]
input_ids_lens = labels_lens = [
tokenized.ne(tokenizer.pad_token_id).sum().item() for tokenized in tokenized_list
]
return dict(
input_ids=input_ids,
labels=labels,
input_ids_lens=input_ids_lens,
labels_lens=labels_lens,
)
all_examples=[]
all_sources=[]
all_qpids=[]
all_queries=[]
for line in tqdm(open(args.rank_path),desc='load data'):
ex = json.loads(line)
all_qpids.append((ex['qid'],ex['pid']))
if data_name!='arguana':
query = ex["query"].replace(DEFAULT_PAD_TOKEN,'PAD')
query_len = get_num_token(query)
passage_max_len = 512-prompt_len-query_len-10
source = prompt.format(doc = truncation(ex['passage'], passage_max_len)).replace(DEFAULT_PAD_TOKEN,'PAD')
else:
source = prompt.format(doc = truncation(ex['passage'], 256)).replace(DEFAULT_PAD_TOKEN,'PAD')
query = truncation(ex['query'], 256)
all_examples.append(source+query)
all_sources.append(source)
all_queries.append(query)
with open(args.res_path,"w") as fw:
for index in tqdm(range(0,len(all_examples),bsz)):
examples=all_examples[index:index+bsz]
sources=all_sources[index:index+bsz]
qpids=all_qpids[index:index+bsz]
queries=all_queries[index:index+bsz]
qid, pid = qpids[0]
examples_tokenized, sources_tokenized = [_tokenize_fn(strings) for strings in (examples, sources)]
input_ids = examples_tokenized["input_ids"]
labels = copy.deepcopy(input_ids)
for label, source_len in zip(labels, sources_tokenized["input_ids_lens"]):
label[:source_len] = IGNORE_INDEX
for index in range(len(input_ids)):
input_ids[index]=input_ids[index][:-1]
input_ids = torch.nn.utils.rnn.pad_sequence(input_ids, batch_first=True, padding_value=tokenizer.pad_token_id).cuda()
labels = torch.nn.utils.rnn.pad_sequence(labels, batch_first=True, padding_value=IGNORE_INDEX).cuda()
labels = labels[..., 1:].contiguous() #BL
with torch.no_grad():
lm_logits = model(input_ids=input_ids,attention_mask=input_ids.ne(tokenizer.pad_token_id))[0]
preds = torch.nn.functional.log_softmax(lm_logits,dim=-1)
label_no_ingore = torch.where(labels==-100,torch.ones(labels.shape).long().cuda(),labels)
logprobs = torch.gather(preds, -1, label_no_ingore.unsqueeze(dim=-1)).squeeze(dim=-1) # B L
indexs=(labels!=-100).long()
scores=(logprobs*indexs).sum(dim=-1)/indexs.sum(dim=-1)
scores=scores.cpu().tolist()
for index,score in enumerate(scores):
qid, pid=qpids[index]
print(" ".join([qid,"Q0",pid,"-1",str(score),model_path]),file=fw)
del lm_logits
results={}
for line in open(args.res_path):
line = line.strip().split()
qid = line[0]
pid = line[2]
score = float(line[4])
if qid not in results:
results[qid] = []
results[qid].append((pid,score))
with open(args.res_path[:-4]+"_post.res","w") as fw:
for qid in results:
res = results[qid]
sorted_res = sorted(res,key = lambda x:-x[1])
for i,item in enumerate(sorted_res):
print(" ".join([qid, "Q0", item[0], str(i), str(item[1]), 'llm']),file=fw)