[QEff. Finetune]: Adding support to load checkpoints from epoch #606
Conversation
Signed-off-by: Tanisha <[email protected]>
tchawada
requested review from
ochougul,
quic-amitraj,
quic-hemagnih and
quic-rishinr
as code owners
Signed-off-by: Tanisha <[email protected]>
Signed-off-by: Tanisha <[email protected]>
| intermediate_epoch = int(train_config.from_peft_checkpoint.split("/")[-2].split("_")[-1]) - 1 | ||
| intermediate_step = int(train_config.from_peft_checkpoint.split("/")[-1].split("_")[-1]) | ||
| try: | ||
| intermediate_epoch = int(train_config.from_peft_checkpoint.split("/")[-2].split("_")[-1]) - 1 |
There was a problem hiding this comment.
Hi @tchawada , the code handles our use case. One suggestion is current exception block is broad and catches any kind of errors. Make it more specific. Please utilize Qgenie extension which helps cover all the edge cases.
There was a problem hiding this comment.
@tchawada, have you checked by passing complete_epoch_2 as --from_peft_checkpoint argument. Also, try to check the existing behavior as well.
If not please do it once.
There was a problem hiding this comment.
Yes, I tried with the path "epoch_2" and checked the overall behavior.
Signed-off-by: Tanisha <[email protected]>
quic-meetkuma
left a comment
quic-meetkuma
left a comment
There was a problem hiding this comment.
LGTM. Let me know once Ann's comment is addressed.
quic-meetkuma
left a comment
quic-meetkuma
left a comment
There was a problem hiding this comment.
LGTM. Let us merge it. Thanks Tanisha for quick fix. Please refer this PR in JIRA and close it. :)
quic-meetkuma
changed the title
Signed-off-by: Tanisha <[email protected]>
Signed-off-by: Tanisha <[email protected]> Signed-off-by: Vahid Janfaza <[email protected]>
Signed-off-by: Tanisha <[email protected]> Signed-off-by: abhishek-singh591 <[email protected]>
3 participants
No description provided.