Unused parameters in final_res_block in UNet
See original GitHub issueLooks like there are some unused parameters in the model. The following is an error message that I got from training it in DDP mode:
Parameters which did not receive grad for rank 14: final_res_block.time_mlp.1.bias, final_res_block.time_mlp.1.weight, norm_mid_cond.bias, norm_mid_cond.weight
On investigating, it seems like the root cause is missing c
argument to the final_res_block
here unless it is not required, in that case, the time_cond_dim
argument should have been set to None
.
Issue Analytics
- State:
- Created a year ago
- Reactions:1
- Comments:5 (3 by maintainers)
Top Results From Across the Web
CA1801: Review unused parameters (code analysis) - .NET
Unused parameters incur maintenance and performance costs. Sometimes, a violation of this rule can point to an implementation bug in the method.
Read more >Mark unused parameters in Kotlin - Stack Overflow
I was able to set @Suppress("UNUSED_PARAMETER") above the method. So all three unused arguments of four in total are covered :) – ecth....
Read more >CA1801 - Remove unused parameter code fix should not ...
For the following fragment: public struct Unit { public static bool operator ==(Unit x, Unit y) => true; public static bool operator !...
Read more >misc-unused-parameters - clang-tidy
Finds unused function parameters. Unused parameters may signify a bug in the code (e.g. when a different parameter is used instead).
Read more >Code Inspection: Unused parameter in partial method
This inspection detects unused parameters in the implementation parts of a partial method. Unused method parameters might indicate incomplete or dead code.
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
@sarvghotra ok sounds good! i’ll proceed with the accelerate or pytorch lightning integration then 😄
For now, I am just trying to use Pytorch’s DDP. If I got some good initial results, will move to DeeepSpeed.