DeepSharp.Import.Keras
Reads a model Keras 3 saved into a DeepSharp network: the .keras archive it saves a model to, or the HDF5 file it saved one to before. The layers its description names are built as DeepSharp's words in Keras's meaning build them — a batch normalisation keeping Keras's epsilon and the complement of its momentum, a window padded as TensorFlow's 'same', or 'causal' along a series, walking a series, an image or a volume in its convolutions and its poolings, global poolings and dropouts of whole channels, its group and root-mean-square normalisations, its leaky, parametric and exponential rectifiers, softmax, a first embedding, LSTM and GRU layers alone or in a Bidirectional one, and MultiHeadAttention over one series — each activation a layer of its own and the last one the loss applies itself lifted into that loss; a Functional model is read as steps, each layer taking the layers it names — an Add, a Concatenate along the last axis, several inputs handed as the columns of one row; its numbers go into the network's slots as Keras lays them out, strictly, every fault named at once where the file holds it. What no network here is built of — a stride of its own for each axis, a dilated window, channels in groups, a layer of any other kind — is refused at its layer.
Activity
- Latest release
- 3d ago
- Total releases
- 14
- Cadence
- ~2 days
- Last 12 months
- 14
Details
- License
- MIT
- First release
- Oct 01, 2026
| Version | Released | |
|---|---|---|
0.17.0
minor
| ||
0.16.0
minor
| ||
0.15.0
minor
| ||
0.14.0
minor
| ||
0.13.0
minor
| ||
0.12.0
minor
| ||
0.11.0
minor
| ||
0.10.0
minor
| ||
0.9.0
minor
| ||
0.8.0
minor
| ||
0.7.1
patch
| ||
0.7.0
minor
| ||
0.6.1
minor
| ||
0.5.0
initial
|