Yuichiro Kikura
Yuichiro Kikura
No, it's not because of availability of WebGL2, but the feature of reading texture values as float32 value, which is required to use WebGL texture buffers as float32 matrix buffers...
Hi, Thanks for giving a good idea. I implemented IndxedDB-based cache system (#654). As default, WebDNN try to fetch only JSON file every time and check the cache is latest...
I'll investigate. What do `True` and `False` means (Where are they from)?
Thanks. I'll investigate more.
Could you share more information (model h5 file, train script, etc.) with me if possible?
Sorry but the h5 file you shared may be the other model. its output shape is `(batch_size, 20)`.
Error message said that invalid size matrix multiplication ( `[1, 2048] * [512 10574]` ) was performed. Could you give me your model structure information (ex. [`model.summary()` in keras](https://keras.io/models/about-keras-models/)) ?
I started to track these two problems in #820 and #821.
@milhidaka I re-implement your patch in e06f903, with some extra comments. Please review it.
Thank you for your interest. Currently we don't plan to support mxnet soon because we don't have enough time, sorry. If you want to use WebDNN with mxnet now, using...