Name | Modified | Size | Downloads / Week |
---|---|---|---|
Parent folder | |||
mbapy-0.1.0.tar.gz | 2023-05-22 | 42.2 kB | |
README.md | 2023-05-22 | 2.4 kB | |
v0.1.0 source code.tar.gz | 2023-05-22 | 176.9 kB | |
v0.1.0 source code.zip | 2023-05-22 | 194.1 kB | |
Totals: 4 Items | 415.5 kB | 0 |
v0.1.0
ADD
stats
test
- mannwhitneyu: func to do mannwhitneyu
FIX
stats
test
change test funcs to recive param in a uniform style as called mbapy-style
dl_torch
fix dl_torch sub-module code
bb
- CnnCfg: a config class for simple cnn block
- reshape: a nn.Module class for doing reshape in nn.Sequential
- permute: a nn.Module class for doing permute in nn.Sequential
- ScannCore: block for SCANN
- SCANN: doing MHSA with cnn style in a 4-D tensor like a typical image input
- PositionalEncoding: PositionalEncoding
- PositionwiseFeedforwardLayer: PositionwiseFeedforwardLayer
- MultiHeadAttentionLayer: MultiHeadAttentionLayer
- FastMultiHeadAttentionLayer: warpper for FlashMHA in MultiHeadAttentionLayer style
- OutMultiHeadAttentionLayer: MHSA for reduce dims, [b, l, c] -> [b, D]
- EncoderLayer: transformer layer
- OutEncoderLayer: EncoderLayer for OutMultiHeadAttentionLayer
- Trans: Transformer
- TransPE: transformer with positional encoding
- OutEncoderLayerAvg: OutEncoderLayer like, but using avgpool
- TransAvg: transformer for OutEncoderLayerAvg
- SeparableConv2d: SeparableConv2d
- ResBlock: ResBlock
- ResBlockR: ResBlock like, Identity Mappings in Deep Residual Networks : exclusive gating
- SABlock: Parallel convolution kernel with residual structure
- SABlockR: SABlock like, exclusive gating
- ScannBlock1d: edited from NonLocalBlock and scann_core
- GenCnn1d: tool func, generate cnn block
- SABlock1D: SABlock like, 3-D tensor
- SABlock1DR: SABlock like, 3-D tensor, exclusive gating
m
- TransCfg: config for a simple transformer
- LayerCfg: config for a simple layer which contains cnn and transformer
- calcu_q_len: tool func for caulcu aq_len for Out-style transformer
- COneDLayer: 3-D tensor
- MAlayer: 4-D tensor
- MAvlayer: MAlayer like, with avgpool down-sampling
- SCANlayer: MAlayer like, with SCANN
- str2net: dict for get blocks and layers easily
- MATTPBase: model using MA-style layer
- COneD: COneDLayer
- MATTP: MAlayer
- MAvTTP: MAvlayer
- MATTPE: MAlayer with pos_embedding
- SCANNTTP: SCANlayer
- MATTP_ViT: MAlayer with ViT performance
data
- denarmalize_img: tool func for denarmalizing a image
- SubDataSet: a high-level Dataset Loader