introduction
Project abstraction is a work-in-progress neural network project using similar algorithms for conversation modelling and physics event classification, featuring curated utterance-response data and word vector features.
Note: abstraction is a project, not a finished product.
documentation | description |
---|---|
setup | setup procedures for ROOT, abstraction and reference hardware P507 |
GPU | GPU utilities and tests |
data | data preprocessing and standardization, HEP data, conversation data |
word vectors | word vectors and related language modeling |
usage | usage of abstraction and related utilities |
ttHbb | ttH(bb) analysis |
TensorFlow | TensorFlow framework |
Torch | Torch framework |
Caffe | Caffe framework |
upcoming
Under consideration is a requirement for arcodex to ensure the existence of a response to an utterance before saving to database. In progress is changing from SkFlow to TF Learn.