-
... where do we spend most of the run-time?
-
Hi!
Thanks again for the awesome repo. I have a small question regarding the global batch size of DPO training reported in the paper vs used in the code base.
In the [paper](https://arxiv.org/p…
-
Currently, `N`-compounding a distribution is done by sampling the source distribution `N` times, adding the values, and using it as a single sample of a new distribution. The only way to construct a p…
-
| | |
| --- | --- |
| Bugzilla Link | [12402](https://llvm.org/bz12402) |
| Version | unspecified |
| OS | Linux |
| Depends On | llvm/llvm-project#12770 |
| CC | @etherzhhb |
## Extended Descript…
-
Hi I tried to run your code with the training command "python compute_stats/compute_spatiotemp_stats_clean_train_swin.py" but it does not work. below is the last part of the display before stopping.
…
-
### Feature request
Add a new API `Dataset.from_spark` to create a Dataset from Spark DataFrame.
### Motivation
Spark is a distributed computing framework that can handle large datasets. By support…
-
Packages required for one module are loaded into the global namespace and are visible to subsequent modules and to users. As is happening with #77, already loaded packages can conflict with subsequent…
-
Hi @darylnak,
I took a look over your implementation of the MMD - I think this is slow because it's doing a lot of looping in Python. In general, the more code you can push into built in NumPy/SciP…
-
The current `HashIndexBuilder` assumes that we are creating the index from scratch, and the persistent `HashIndex` is not well-optimized for inserting a large number of entries at once. To support mul…
-
*[COMPV INFO]: [UltMrzSdkEngine] Call: ultimateMrzSdk::UltMrzSdkEngine::init
*[COMPV INFO]: [UltMrzSdkEngine] jsonConfig: {"debug_level":"info","debug_write_input_image_enabled":false,"debug_internal…