issues
search
CERC-AAI
/
multimodal
An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
Apache License 2.0
8
stars
2
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
add mask for pad token
#14
floatingbigcat
closed
1 year ago
0
Fix maskpad
#13
floatingbigcat
closed
1 year ago
0
Fix masking of loss
#12
kshitijkg
closed
1 year ago
1
Question about gpt-neox parallel parameter partition
#11
floatingbigcat
closed
1 year ago
1
Script to convert GPT Neox Checkpoint for magma
#10
kshitijkg
closed
1 year ago
0
Add support to HFTokenzer
#9
floatingbigcat
closed
1 year ago
0
DataLoader worker is killed by signal: Segmentation fault.
#8
floatingbigcat
closed
1 year ago
1
Webdataset support for NeoX 20B tokenizer
#7
kshitijkg
closed
1 year ago
2
Script can't find local copy of CLIP
#6
Rabbidon
closed
1 year ago
2
Training without Pipeline Parallelism
#5
kshitijkg
closed
1 year ago
3
Pythia Checkpoint Loading
#4
kshitijkg
closed
1 year ago
1
Pythia checkpoint loading
#3
kshitijkg
opened
1 year ago
0
Added option to load both CLIP and LLM
#2
kshitijkg
closed
1 year ago
0
apply is_pipe_parallel fix to summit branch
#1
floatingbigcat
closed
1 year ago
0
Previous