-
Hi, I wonder did you try to use the pretrained weight on cifar rather than randomly initialized weights?
-
While working on https://github.com/huggingface/transformers/issues/8353 I discovered that `--fp16` causes a 10x+ increase in gpu memory demands.
e.g. I can run bs=12 w/o `--fp16`
```
cd exam…
-
Currently the early stopping feature requires the user to specify a validation set. Instead, the validation set should be optional. Reasonable behavior should happen if it is unspecified.
> early_s…
-
arXiv论文跟踪
-
Hi Kizmuz, First of all congratulations for this really great job!
OK, I have trained my workers, and here you are some questions:
Do you have some code to test predictions with their model?
Do…
-
Being led by @beckeroobonsai -- expand GBM by including additional predictors (e.g., weather) for the applicable periods.
-
0.841504:old
0.8428461
0.851 rm filter
0.8529791 source
0.8519965 source count
0.8630154 start yearmonth
0.8688813 start date
0.8693542 interval
0.87072 unique obj
0.8727769 browser_flag
0.8726645 obj…