Comments (3)
Thanks for spotting this! It's a bug I think, though it almost never happens -> mlp_z
always turns 0 first before all the intermediate_z
become 0 during training. And you are correct that the FFN parts should be skipped when intermediate_z
are all zero during training.
I am refactoring the code and should have a clear version ready soon!
from cofipruning.
Changing this line to be if self.intermediate.dense is None or self.intermediate_z.sum().eq(0).item() or self.mlp_z.sum().eq(0).item():
should be sufficient.
from cofipruning.
I see! thanks for your reply๐
from cofipruning.
Related Issues (20)
- An issue when reproducing the efficiency evaluation HOT 1
- Discrepancy between my evaluation results and README for MNLI in evaluation.py HOT 4
- What's the model I should prepare and the training process? HOT 3
- Something wrong with run_FT.sh and data_dir
- problem of loading from_pretrained('princeton-nlp/CoFi-XXX') HOT 2
- [Typo]: argument names in run_qa_prune.py seems not updated HOT 1
- The pre-request and training process HOT 2
- Performance reproduction
- Detailed experiment results on RoBERTa? HOT 2
- Is this method applicable to LLAMA? HOT 1
- Does it support Marian machine translation models? HOT 1
- The initial of 'intermediate' loga HOT 1
- Too low accuracy result compared with the expected result HOT 6
- Bug or intent? HOT 2
- prunable_model_size updated only in initialize_structured_head HOT 2
- (expected_sparsity - target_sparsity) or (expected_sparsity - target_sparsity).abs() HOT 2
- Why prepruning distillation? HOT 1
- when the training is going to end,occurred error HOT 1
- what is the distillation model and where should i go to download it ?
- In the finetune stage, if the l0_module is None?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. ๐๐๐
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google โค๏ธ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from cofipruning.