-
Notifications
You must be signed in to change notification settings - Fork 2k
Pull requests: NVIDIA/Megatron-LM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[BUG] Preprocess_data.py does not finalize all keys
#852
opened Jun 2, 2024 by
zainsarwar865
Loading…
[BUG] wrong scale softmax for local transformer implement
#848
opened May 29, 2024 by
Superkeyv
Loading…
Fix the bug where the optimizer doesn't actually call multi_tensor_applier under float16.
#847
opened May 29, 2024 by
Gstdioh
Loading…
Fix Bug: Configuring Datasets with train-data-path, valid-data-path, test-data-path
#840
opened May 27, 2024 by
Eisenhower
Loading…
[Fix] Assertion to check if
num_layers
is divisible by the pipeline size
#823
opened May 13, 2024 by
kenkenpa2126
Loading…
Fix incorrect
src
argument in broadcast_params
function
#796
opened Apr 26, 2024 by
Yuxin-CV
Loading…
fix loading distributed checkpoint when enable auto-detect-ckpt-format but disable use-dist-ckpt
#794
opened Apr 24, 2024 by
imh966
Loading…
fix a mistake when check if num_layers dividable by vpp
#781
opened Apr 16, 2024 by
constroy
Loading…
[very simple change] Remove duplicated code
stale
No activity in 60 days on issue or PR
#765
opened Apr 3, 2024 by
NoelBird
Loading…
fix new bucket when param require new bucket
stale
No activity in 60 days on issue or PR
#762
opened Apr 2, 2024 by
wangxicoding
Loading…
Updated No activity in 60 days on issue or PR
fused_kernels
import path
stale
#760
opened Mar 31, 2024 by
Yazeed7
Loading…
use new methods for communication
stale
No activity in 60 days on issue or PR
#758
opened Mar 30, 2024 by
mayank31398
Loading…
drop redundant check
stale
No activity in 60 days on issue or PR
#757
opened Mar 30, 2024 by
mayank31398
Loading…
Fix typo in README.md
stale
No activity in 60 days on issue or PR
#751
opened Mar 26, 2024 by
HashiamKadhim
Loading…
Support S3 checkpointing for the torch strategy in distributed checkpointing
#748
opened Mar 22, 2024 by
jrocmar
Loading…
Previous Next
ProTip!
Filter pull requests by the default branch with base:main.