Colin Raffel
Colin Raffel
As of now the mel fb is triangular, chroma fb is Gaussian. But, sometimes people use triangular fbs for computing chroma, and so on. I propose that instead of melfb...
## Description The example code at https://nlp.gluon.ai/model_zoo/bert/index.html for the DistilBERT model produces an exception at HEAD. ### Error Message ``` --------------------------------------------------------------------------- ValueError Traceback (most recent call last) in () 2...
In some prompt templates, there is logic like this: ``` {% set seq = [ ''Answer the question depending on the context.'', ''What is the answer?'', ] %} {{ seq...
If I run the TensorFlow version of this code (`tf_train.py`) with https://github.com/openai/iaf/pull/8 applied, I get a NaN within the first few iterations and training stops. If I remove that change,...
We don't use this dataset, so it is untested and we should remove it.
Currently, only the postprocessed model outputs are written out into a file suffixed with "predictions". This outputs an additional file suffixed with "outputs" that stores the raw model outputs, without...
By default, joblib.Parallel will memmap any numpy ndarray objects that are shared between tasks when they are bigger than 1e6 bytes, to avoid lots of memory allocation. In any experiment...
Because https://github.com/craffel/midi-dataset/blob/master/experiment_utils.py#L329 when `n_conv = 0`, `filters_per_layer[:n_conv - 1] = [64,]`.
Would look something like this: ``` Python def construct_best_network(model_file, trial_glob): """ Constructs the best-performing model according to the supplied hyperparameter optimization trials, and populates the parameters with the provided model...
Seems to make most sense to put all non-derived-data into /results.