Daniel Kleine
|
bbb2a0c3d5
fixed num_workers (#229)
|
1 gadu atpakaļ |
rasbt
|
39c4a887eb
add allowed_special={"<|endoftext|>"}
|
1 gadu atpakaļ |
Sebastian Raschka
|
97ed38116a
Rename drop_resid to drop_shortcut (#136)
|
1 gadu atpakaļ |
Sebastian Raschka
|
dd51d4ad83
Make datesets and loaders compatible with multiprocessing (#118)
|
1 gadu atpakaļ |
James Holcombe
|
05718c6b94
Use instance tokenizer (#116)
|
1 gadu atpakaļ |
Sebastian Raschka
|
2de60d1bfb
Rename variable to context_length to make it easier on readers (#106)
|
1 gadu atpakaļ |
Sebastian Raschka
|
3829ccdb34
Remove reundant dropout in MLP module (#105)
|
1 gadu atpakaļ |
Sebastian Raschka
|
ca96abac8a
Set up basic test gh worklows (#79)
|
1 gadu atpakaļ |
Sebastian Raschka
|
9d6da22ebb
Update pep8 (#78)
|
1 gadu atpakaļ |
rasbt
|
da33ce8054
remove redundant unsqueeze in mask
|
1 gadu atpakaļ |
rasbt
|
87fcfd9245
mha variants
|
1 gadu atpakaļ |
rasbt
|
e0df4df433
add dropout for embedding layers
|
1 gadu atpakaļ |
rasbt
|
b827bf4eea
remove redundant double-unsequeeze
|
1 gadu atpakaļ |
rasbt
|
cdcd73ba7f
drop_last=True
|
1 gadu atpakaļ |
rasbt
|
4e68649f16
comment update
|
1 gadu atpakaļ |
rasbt
|
fe332006de
ch4 exercise solutions
|
1 gadu atpakaļ |
rasbt
|
5d1d8ce511
add shape information for clarity
|
1 gadu atpakaļ |
rasbt
|
ec312e581b
add chapter 4 code
|
1 gadu atpakaļ |