<html><head><style type="text/css"><!-- DIV {margin:0px;} --></style></head><body><div style="font-family:times new roman,new york,times,serif;font-size:12pt">Dear Srilm Users,<br><br>I have a question for perplexity computation..<br><br>I create an interpolated language model which gives lower perplexity than general baseline language model. I used <span style="font-weight: bold;">witten-bell</span> discounting for both.<br><br>But when i used <span style="font-weight: bold;">kn</span> discounting the baseline language models shows greater perplexity than the interpolated one.<br><br>Can anybody tell me, why?<br><br>Why not kn discounting also give lower perplexity...?<br><br>Thanks & Regards<br><br>Akmal<br><br><br><div><br></div></div><br>
</body></html>