princeton-nlp CoFiPruning issues

princeton-nlp / CoFiPruning

[ACL 2022] Structured Pruning Learns Compact and Accurate Models https://arxiv.org/abs/2204.00408

MIT License

192 stars 31 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

In the finetune stage, if the l0_module is None?

#59 SHUSHENGQIGUI opened 5 months ago
0
what is the distillation model and where should i go to download it ?

#58 SHUSHENGQIGUI opened 5 months ago
0
when the training is going to end,occurred error

#57 zll0000 opened 10 months ago
1
Why prepruning distillation?

#56 mpiorczynski opened 1 year ago
1
(expected_sparsity - target_sparsity) or (expected_sparsity - target_sparsity).abs()

#55 hyx1999 closed 1 year ago
2
prunable_model_size updated only in initialize_structured_head

#54 tatami-galaxy closed 1 year ago
2
Bug or intent?

#53 mpiorczynski closed 1 year ago
2
Too low accuracy result compared with the expected result

#52 xtchon opened 1 year ago
6
The initial of 'intermediate' loga

#51 secretu opened 1 year ago
1
Does it support Marian machine translation models?

#50 Khaled-Elsaka opened 1 year ago
1
Is this method applicable to LLAMA?

#49 Daisy5296 opened 1 year ago
1
Detailed experiment results on RoBERTa?

#48 ROIM1998 closed 1 year ago
2
Performance reproduction

#47 oijt894 closed 1 year ago
0
The pre-request and training process

#46 oijt894 opened 1 year ago
2
[Typo]: argument names in run_qa_prune.py seems not updated

#45 sbwww opened 1 year ago
1
problem of loading from_pretrained('princeton-nlp/CoFi-XXX')

#44 twwwwx opened 1 year ago
2
Something wrong with run_FT.sh and data_dir

#43 gaishun opened 1 year ago
0
What's the model I should prepare and the training process?

#42 gaishun closed 1 year ago
3
Typo fix

#41 eldarkurtic closed 1 year ago
0
Discrepancy between my evaluation results and README for MNLI in evaluation.py

#40 TinaChen95 opened 1 year ago
4
An issue when reproducing the efficiency evaluation

#39 ROIM1998 closed 1 year ago
1
Add matplotlib to requirements

#38 eldarkurtic closed 1 year ago
0
Generating predictions with CoFi models

#37 eldarkurtic closed 1 year ago
2
Removing the already-pruned parts in the model may cause some changes in the outputs

#36 backspacetg closed 1 year ago
3
Introducee random teacher layer sets

#35 zhangzhenyu13 closed 2 years ago
0
layer-distillation: teacher layer sets selection?

#34 zhangzhenyu13 closed 2 years ago
4
Error in finetune with pruned model--AttributeError: 'NoneType' object has no attribute 'forward'`

#33 zoetu closed 1 year ago
3
Where can i see the detail of distillation?

#32 zoetu closed 2 years ago
0
How to get the loss of `lagrangian_regularization`

#31 CaffreyR closed 1 year ago
5
Update README.md

#30 CaffreyR closed 2 years ago
2
Experimental results

#29 iMountTai closed 1 year ago
2
How to prune the model from the very begigning?

#28 CaffreyR closed 1 year ago
4
training error about qnli

#27 iMountTai closed 1 year ago
4
Maybe confusing description of the distillation constraint

#26 sbwww closed 2 years ago
1
About the diag() and distillation in your paper

#25 CaffreyR closed 2 years ago
6
About the upper layer in your paper

#24 CaffreyR closed 2 years ago
2
More numbers on other sparsities

#23 GeneZC closed 2 years ago
6
potential bug loading a pruned model with no masks

#22 ctsan closed 1 year ago
2
Student model initialization

#21 slawek-ib closed 2 years ago
2
added black and wandb to requirements.txt and removed hardcoded path to glue.py in run_glue_prune.py

#20 ketulib closed 2 years ago
0
Pruning for Encoder-Decoder Architecture?

#19 Luckick closed 2 years ago
4
The usage of L_c

#18 Ther-nullptr closed 2 years ago
6
Why use 3 optimizers during training?

#17 Ther-nullptr closed 2 years ago
2
It is possible to run pruning on multiple GPUs？

#16 TonyNemo closed 2 years ago
8
Should I get a finetuned teacher model by setting `pruning_type=None`

#15 horizon86 closed 2 years ago
2
fix: fix a small typo which may cause assert error in l0module.

#14 Ther-nullptr closed 2 years ago
1
Loading a finetuned model starts from scratch?

#13 ctsan closed 2 years ago
2
Device incompatibility?

#12 ctsan closed 2 years ago
4
fix an logic bugs of none grad due to the l0_module has not yet forwarded

#11 zhangzhenyu13 closed 2 years ago
1
Fatal Logic Error found in trainer.py

#10 zhangzhenyu13 closed 2 years ago
3