daniel-gallo naraim issues

daniel-gallo / naraim

3 stars 0 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Make AutoAugment work as non-final transformtion

#24 daniel-gallo opened 4 months ago
0
Bump tornado from 6.4 to 6.4.1

#23 dependabot[bot] opened 4 months ago
0
Ablate the learning rate for attentive probe

#22 robertdvdk opened 5 months ago
0
Check probe batch size 512 vs 1024

#21 robertdvdk opened 5 months ago
0
Do attentive probing after 50k steps of pretraining, 100k, 150k, ..., 500k

#20 robertdvdk opened 5 months ago
0
Implement gradient accumulation

#19 robertdvdk opened 5 months ago
0
Check whether performance of the probe keeps improving even if the pre-training is overfitting

#18 robertdvdk opened 5 months ago
0
Implement per-channel normalisation

#17 robertdvdk opened 5 months ago
0
Implement visualization of error maps in the tensorboard logger

#16 robertdvdk opened 5 months ago
0
Make training completely reproducible (seed dataloader, and when resuming from a checkpoint, also load the dataloader state)

#15 robertdvdk closed 5 months ago
0
Implement autoregressive generation of images

#14 robertdvdk opened 5 months ago
0
Find out why we need to prepend 2 black patches in the visualisation

#13 robertdvdk opened 5 months ago
0
Disable prefix sampling for evaluation

#12 daniel-gallo opened 5 months ago
0
Native resolution training

#11 robertdvdk opened 5 months ago
0
Data parallelism

#10 robertdvdk opened 5 months ago
0
Check classification head

#9 robertdvdk opened 5 months ago
0
Fix tests for ImageNet on CI/CD

#8 daniel-gallo opened 6 months ago
0
Training Metric Logging

#7 phlippe closed 6 months ago
0
Metrics - Device Get

#6 phlippe closed 6 months ago
1
Eval RNG

#5 phlippe closed 6 months ago
0
Logging HParams

#4 phlippe closed 6 months ago
0
Eval Metrics

#3 phlippe closed 6 months ago
0
Output Head

#2 phlippe closed 6 months ago
0
Use bfloat16

#1 daniel-gallo closed 6 months ago
0