mit-han-lab / hardware-aware-transformers

[ACL'20] HAT: Hardware-Aware Transformers for Efficient Natural Language Processing

https://hat.mit.edu

mit-han-lab/hardware-aware-transformers Issues

Question about the SubTransformers sampling process.
Updated 7 months ago
Used version of `fairseq`
Updated 2 years ago
Question about the latency on Raspberry Pi
Closed 2 years ago1
lower loss but the BLEU is 0
Updated 2 years ago
What is the method used to sample training examples for the MLP latency predictor?
Closed 3 years ago2
Does the generated latency count in the embedding lookup table and the last output layers ?
Closed 3 years ago2
Latency predictor relative error instead of absolute error
Closed 3 years ago1
Training new SuperTransformer - calculating number of SubTransformer combinations?
Closed 3 years ago2
how to use the processed data in your code?
Closed 3 years ago
About the Quantization Friendly.
Closed 3 years ago1
Quantization on HAT.
Closed 4 years ago4
Error in step 2.3 (Evolutionary search with latency constraint)
Closed 4 years ago5
Question on how to evaluate inherited SubTransformers.
Closed 4 years ago11
question about number of parameters
Closed 4 years ago2
questions about the search & training process
Closed 4 years ago1
RAM in the used Raspberry Pi
Closed 4 years ago2
One question
Closed 4 years ago2