-
Notifications
You must be signed in to change notification settings - Fork 2.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
PaddlePaddle Hackathon 56 提交 #1088
Conversation
merges.txt也通过百度云上传 |
@yingyibiao baidu网盘包含merges.txt,现在已经删除community/junnyu所有文件。 |
|
权重已上传至bos |
参照 #1085 的review意见修改类似问题~ |
顺便将 DialoGPT-medium,DialoGPT-large 这两个权重也导入~ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
非常不错!
看 GPTForTokenClassification
GPTForSequenceClassification
是否也能给 https://github.com/PaddlePaddle/PaddleNLP/blob/develop/examples/language_model/gpt/ 中添加一些实际使用的例子
return logits | ||
|
||
|
||
class GPTForSequenceClassification(GPTPretrainedModel): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
GPTForTokenClassification
GPTForSequenceClassification
这两个东西能否加一个例子到 https://github.com/PaddlePaddle/PaddleNLP/blob/develop/examples/language_model/gpt/ 中
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
GPTForTokenClassification
跟 BertForTokenClassification
基本一样的,只是模型不需要输入token type id
.
GPTForSequenceClassification
现已完成
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
现已添加使用GPTForTokenClassification
进行NER的例子,发现效果很差。
@yingyibiao |
已上传~ |
LGTM for community related files |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Precision | 0.484939 | | ||
Recall | 0.634716 | | ||
F1 | 0.549810 | | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这效果感觉可能有些偏低。
|
||
基于`gpt-cpm-small-cn-distill`在MSRA的NER任务上Fine-tuning后,在验证集上有如下结果: | ||
|
||
Metric | Result | |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
GPT可以解决TokenClassification问题这个事有相关paper佐证吗?@ZHUI
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Task: #1074
GPTEmbeddings
中paddle.ParamAttr
的name属性,设置了话会报错,提示说重复使用了相同的名字,单元测试无法通过,jupyter notebook中无法重复初始化。microsoft-DialoGPT-small
转换后的误差有点大,问题与之前的应该类似,其他的两个模型转换后误差正常。(使用了相同的转换代码,因此不存在转换时代码的错误。)lm_head.weight
,因为是它是与word embedding
绑定的,所以没必要转换它,如果有需要可自行修改。