Joshua Peterson
Joshua Peterson
I was able to resume just by setting load_path and running the same code to start training. Make sure to backup your current weights so you don't lose/overwrite them if...
Just the directory, not any of the filenames (e.g., './my_saved_models_path')
@stefanuddenberg has implemented this and a few other improvements. @jodeleeuw: should I ask him to submit a pull request?
I'm getting varying scores that doesn't always preserve the rank in the table. Are these supposed to be averaged to say 10 runs of training a model? They don't seem...
Any benchmarks for this? Also, how big is the file size given style/script tags are maintained? Might also be a good idea to allow inputting less than the entire set...
Not currently. If you want to try a batch and submit a pull request with the estimate in the readme, I will merge it.
Yep, different implementation.
@aakova thanks can you submit a pull request with the fix?
is this fully backwards compatible the old code / json dumping?