Comments (13)
不需要分类个数 在'bert.py'的‘MyPro() - get_labels’里,直接把'return [0, 1]'改成你的类别名列表
from bert_chinese_pytorch.
我也能遇到了类似的问题,我把return[0,1]改成我的类别列表后,报了这个错,RuntimeError: CUDA error: device-side assert triggered。去网上查了下,是说类别的数量必须在0到n_classes之间,但是n_classes(分类个数)不知道在什么地方设置。我的类别数量有4000+个,请问有遇到类似的问题吗
from bert_chinese_pytorch.
@licheng-pro
如果你数据中的label已经转成了label id:
def get_labels(self):
return [str(i) for i in range(n_classes)]
from bert_chinese_pytorch.
@ETCartman 没有转成label id,
并且当我换成cpu后,报了这个错。RuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes' failed. at /pytorch/aten/src/THNN/generic/ClassNLLCriterion.c:93
from bert_chinese_pytorch.
@licheng-pro 需要return是lable id,而不是label!
try this:
def get_labels(self, label_path):
fo = open(label_path, 'r', encoding='utf-8')
lines = fo.readlines()
label_to_id ={}
for i, line in enumerate(lines):
label_to_id[line.strip()] = i
print(label_to_id)
fo.close()
return [str(i) for i in range(len(label_to_id))]
or just:
def get_labels(self, label_path):
fo = open(label_path, 'r', encoding='utf-8')
lines = fo.readlines()
fo.close()
return [str(i) for i in range(len(lines))]
from bert_chinese_pytorch.
@ETCartman
convert_examples_to_features方法里的这个地方才是把label变成label id吧,getlabels方法返回的应该就是label吧?
from bert_chinese_pytorch.
@licheng-pro 刚去看了下代码确实是:stuck_out_tongue:,不过我做2k个类别没有任何问题。
RuntimeError: Assertion `cur_target >= 0 && cur_target < n_classes' failed. at /pytorch/aten/src/THNN/generic/ClassNLLCriterion.c:93
我个人感觉这个错误肯定是label出了问题,比如label_path
中的数据有重复, 你可以
return list(set(label_list))
试一下
from bert_chinese_pytorch.
我已经解决了,多谢了,老哥 @ETCartman
from bert_chinese_pytorch.
不过现在好像在
model = BertForSequenceClassification.from_pretrained(args.bert_model,
cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format(
args.local_rank))
这一行会报如下的错误:
File "/root/anaconda3/envs/liu37/lib/python3.7/site-packages/pytorch_pretrained_bert/modeling.py", line 581, in from_pretrained
model = cls(config, *inputs, **kwargs)
TypeError: __init__() missing 1 required positional argument: 'num_labels'
from bert_chinese_pytorch.
不过现在好像在
model = BertForSequenceClassification.from_pretrained(args.bert_model, cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format( args.local_rank))
这一行会报如下的错误:
File "/root/anaconda3/envs/liu37/lib/python3.7/site-packages/pytorch_pretrained_bert/modeling.py", line 581, in from_pretrained model = cls(config, *inputs, **kwargs) TypeError: __init__() missing 1 required positional argument: 'num_labels'
我也是这个问题,请问你解决了吗?
from bert_chinese_pytorch.
@Zhaohaoran1997 在最后加上这个参数就可以
model = BertForSequenceClassification.from_pretrained(args.bert_model,
cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format(
args.local_rank), num_labels=len(label_list))
也可以参考我fork的代码:https://github.com/liuyijiang1994/bert_senta
from bert_chinese_pytorch.
@Zhaohaoran1997 在最后加上这个参数就可以
model = BertForSequenceClassification.from_pretrained(args.bert_model, cache_dir=PYTORCH_PRETRAINED_BERT_CACHE / 'distributed_{}'.format( args.local_rank), num_labels=len(label_list))
也可以参考我fork的代码:https://github.com/liuyijiang1994/bert_senta
我改了num_labels之后程序被kill了,请问是我写得有bug还是机器性能不足?
04/22/2019 15:58:31 - INFO - pytorch_pretrained_bert.modeling - Weights of BertForSequenceClassification not initialized from pretrained model: ['classifier.bias', 'classifier.weight'] 04/22/2019 15:58:31 - INFO - pytorch_pretrained_bert.modeling - Weights from pretrained model not used in BertForSequenceClassification: ['cls.predictions.bias', 'cls.predictions.transform.dense.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.decoder.weight', 'cls.seq_relationship.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.LayerNorm.bias'] 04/22/2019 15:58:34 - INFO - __main__ - ***** Running training ***** 04/22/2019 15:58:34 - INFO - __main__ - Num examples = 43425 04/22/2019 15:58:34 - INFO - __main__ - Batch size = 128 04/22/2019 15:58:34 - INFO - __main__ - Num steps = 3392 Epoch: 0%| | 0/10 [00:00<?, ?it/s]已杀死ion: 0%| | 0/340 [00:00<?, ?it/s]
from bert_chinese_pytorch.
@Zhaohaoran1997 看上去运行的时候已经是正常的了 也许是性能的问题
from bert_chinese_pytorch.
Related Issues (9)
- 如何使用已训练好的output_dir下的model,而不是每次都do_train? HOT 3
- > 不过现在好像在
- EOFError: Compressed file ended before the end-of-stream marker was reached HOT 7
- 请教.json文件的具体张什么样啊 HOT 1
- 报错__init__() missing 1 required positional argument: 'num_labels' HOT 1
- 您好,请问‘’Weights of BertForSequenceClassification not initialized from pretrained model: ['classifier.weight', 'classifier.bias']‘’这种情况您碰到过吗,这样权重无法加载,预训练模型的意义也发挥不出来。 HOT 1
- Error:'model' not found
- example转feature
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from bert_chinese_pytorch.