kongds

Problem in SMART embedding

1

Thank you for provided code of SMART. SMART uses the following code to get the embeddings, which is then used to get noisy embeddings and feed bert as `inputs_embeds`. https://github.com/namisan/mt-dnn/blob/ca896ef1f9de561f1741221d2c98b4d989e3ed19/mt_dnn/matcher.py#L124...

90% sparsity QQP result

5

Hello， I find the results of CAP-m in 90% sparsity QQP is "91.6/87.7", while CAP-soft is "90.7/87.4"(bold). Is the result of CAP-m correct? 

Problem in reproduce multi-distillation approach

12

Hello, Thank you for providing code. I can get the right results of W1A1 with `bash scripts/run_glue.sh MNLI` (around 77 accuracy on MNLI) But when i reproduce the W1A1 with...