-
```
When I use interpolate-ngram to interpolate two models by CM or GLI with
perplexity optimization, I get following faults:
1st:
interpolate-ngram -lm "model1.lm, model2.lm" -smoothing ModKN -inte…
-
```
I have a background unigram model (bg.arpa), some additional training data
(train.txt) and some dev text (dev.txt). I want to create an interpolated
unigram that optimizes the perplexity of dev.tx…
-
```
I have a background unigram model (bg.arpa), some additional training data
(train.txt) and some dev text (dev.txt). I want to create an interpolated
unigram that optimizes the perplexity of dev.tx…
-
```
When I use interpolate-ngram to interpolate two models by CM or GLI with
perplexity optimization, I get following faults:
1st:
interpolate-ngram -lm "model1.lm, model2.lm" -smoothing ModKN -inte…
-
Hello!
I had finished the above commands before the last evaluation.
I have two question:
1.I don't know how to key the last command
"./eval.py --out_file "
the output file must be created fir…
-
首先感谢作者团队的贡献!作者的思路给我了很大的启发,在复现过程中发现一些问题:
1、一个带有标签的低频单词被unk遮掉后,形成的【tag,unk】会不会影响最终的整体效果;
2、单纯了训练了一下Language Model看结果我不知道是好是坏,本来是CV,忽然NLP可能不太理解。所以期待作者能帮我解答一下
。
-
****Version****
pyabsa==2.3.1
torch==1.13.0
transformers==4.29.0
****Describe the bug****
I used to load my custom state_dict from my windows system and the loading procedure worked.
However, …
-
> Our code also allows you to turn your suffix array index into *an efficient n-gram language model, which can be used to generate text* or compute the perplexity of a given text.
Looking for a sim…
-
Thanks for your great job.I have some trouble when i prepare the data.I do not know where can I download these files (matterport_category_mappings.tsv and object_norm_inv_perplexity.npy). And I am co…
-
谢谢你开发的这个小程序,解决了我的一个苦恼。
建议做如下优化:
shift+space可以输出为space,而不是禁用,这也符合一般的按键逻辑,只是空格健没有大写的空格而已。
比如我经常写这样的SQL语句:select * from
当我按住shift不松时,会先按*号,再按space,然后才会松手shift,如果禁用的话,空格就出不来。
按键精灵我测试过,可以实现(把shift+s…