models

models copied to clipboard

[BUG] Gradients do not exist for categorical variables in the advanced example notebook

4

### Bug description When running the advanced example [notebook](https://github.com/NVIDIA-Merlin/models/blob/main/examples/06-Define-your-own-architecture-with-Merlin-Models.ipynb), the fit method is printing the following warning ```WARNING:tensorflow:Gradients do not exist for variables ['embedding_features/userId:0', 'embedding_features/movieId:0', 'embedding_features/title:0', 'embedding_features/gender:0', 'parallel_block/userId:0', 'parallel_block/movieId:0', 'parallel_block/title:0',...

sararb

[BUG] Testing in CI / local dev environments fails when all tests are run at once

3

### Bug description When we run all the unit tests in one pass, some of the tests related to sampling classes fail. These same tests pass when run individually. (We...

sararb

[FEA] Split datasets in `datasets` package chronologically

2

# 🚀 Feature request Split the datasets in `datasets` package chronologically instead of at random. ## Motivation Splitting at random is known to be a problematic splitting procedure, since it...

karlhigley

[BUG] Update MovieLens Dataset Defaults to handle memory resources

2

### Bug description Various examples in the codebase use `get_movielens` without any parameters. Using this on a GPU (e.g. RTX A3000 with 6GB memory) results in a `MemoryError`. (`std::bad_alloc: out_of_memory:...

oliverholworthy

[FEA] Create unit tests for saving and loading back the retrieval and ranking models

2

# 🚀 Feature request we are missing unit tests for saving and loading back the trained models. This is important for inference step and to run inference examples without issues....

rnyak

[TASK] Metrics that require predicted item ids (e.g. ItemCoverageAt, PopularityBiasAt, NoveltyAt) are not supported in model fit() and evaluate()

2

### Bug description Metrics that require predicted item ids (e.g. ItemCoverageAt, PopularityBiasAt, NoveltyAt) are not supported in model fit() and evaluate(). Currently, they can be used only after model.fit() is...

gabrielspmoreira

### Description Description NVIDIA won the WSDM 2021 competition. As we added new features and libraries, we want to add the solution as Merlin (Models) example: http://www.recsyschallenge.com/2022/

bschifferer

# 🚀 Feature request It'd be useful to add a MultiClassClassificationTask example with single-hot or sequential datasets. We can use RecSys'22 Dressipi challenge dataset for that example.

rnyak

# 🚀 Feature request Currently `generate_data()` does not generate a dataset based on the cardinalities in a given schema. It rather generates a dataset by drawing samples from a log-normal...

rnyak

[FEA] Add post_block layer to MultiClassClassificationTask class

1

# 🚀 Feature request Currently we get logits as outputs if we train a `MultiClassClassificationTask()` model. It'd be useful to add an optional post-block layer in the MultiClassClassificationTask class and...

rnyak