{ "id": 363980, "title": "transformers4rec - Sequential and Session-Based Recommendation Systems", "url": "/competitions/otto-recommender-system/discussion/363980", "author_name": "", "post_date": "2022-11-04T00:03:01.917880500Z", "votes": 31, "comment_count": 4, "views": 0, "content": "

Hey everyone,

\n

I want to share this repository from NVIDIA that integrates with the HuggingFace transformers library and makes state-of-the-art Transformer architectures available in PyTorch for recommender systems.

\n

https://github.com/NVIDIA-Merlin/Transformers4Rec

\n

From the repository:

\n

Transformers4Rec is a flexible and efficient library for sequential and session-based recommendation and can work with PyTorch.

\n

It works as a bridge between NLP and recommender systems by integrating with one the most popular NLP frameworks HuggingFace Transformers, making state-of-the-art Transformer architectures available for RecSys researchers and industry practitioners.

\n

\"Sequential

\n
\n \n
\n

Transformers4Rec supports multiple input features and provides configurable building blocks that can be easily combined for custom architectures.

\n

You can build a fully GPU-accelerated pipeline for sequential and session-based recommendation with Transformers4Rec and its smooth integration with other components of NVIDIA Merlin: NVTabular for preprocessing and Triton Inference Server.

\n

And in their examples directory, you can find the following tutorials/examples:

\n\n
\n

I personally found it a bit hard to set up but once you have the momentum and understand the background more, it becomes easier.

\n
", "messages": [ { "id": "2016356", "postDate": "11/04/2022 00:03:01", "content": "

Hey everyone,

\n

I want to share this repository from NVIDIA that integrates with the HuggingFace transformers library and makes state-of-the-art Transformer architectures available in PyTorch for recommender systems.

\n

https://github.com/NVIDIA-Merlin/Transformers4Rec

\n

From the repository:

\n

Transformers4Rec is a flexible and efficient library for sequential and session-based recommendation and can work with PyTorch.

\n

It works as a bridge between NLP and recommender systems by integrating with one the most popular NLP frameworks HuggingFace Transformers, making state-of-the-art Transformer architectures available for RecSys researchers and industry practitioners.

\n

\"Sequential

\n
\n \n
\n

Transformers4Rec supports multiple input features and provides configurable building blocks that can be easily combined for custom architectures.

\n

You can build a fully GPU-accelerated pipeline for sequential and session-based recommendation with Transformers4Rec and its smooth integration with other components of NVIDIA Merlin: NVTabular for preprocessing and Triton Inference Server.

\n

And in their examples directory, you can find the following tutorials/examples:

\n\n
\n

I personally found it a bit hard to set up but once you have the momentum and understand the background more, it becomes easier.

\n
", "rawMarkdown": "Hey everyone,\n\nI want to share this repository from NVIDIA that integrates with the HuggingFace transformers library and makes state-of-the-art Transformer architectures available in PyTorch for recommender systems.\n\nhttps://github.com/NVIDIA-Merlin/Transformers4Rec\n\n*From the repository:*\n\nTransformers4Rec is a flexible and efficient library for sequential and session-based recommendation and can work with PyTorch.\n\nIt works as a bridge between NLP and recommender systems by integrating with one the most popular NLP frameworks [HuggingFace Transformers](https://github.com/huggingface/transformers), making state-of-the-art Transformer architectures available for RecSys researchers and industry practitioners.\n\n\"Sequential
\n
\n
Sequential and Session-based recommendation with Transformers4Rec
\n
\n\nTransformers4Rec supports multiple input features and provides configurable building blocks that can be easily combined for custom architectures.\n\nYou can build a fully GPU-accelerated pipeline for sequential and session-based recommendation with Transformers4Rec and its smooth integration with other components of [NVIDIA Merlin](https://developer.nvidia.com/nvidia-merlin): [NVTabular](https://github.com/NVIDIA-Merlin/NVTabular) for preprocessing and [Triton Inference Server](https://github.com/triton-inference-server/server).\n\nAnd in their examples directory, you can find the following tutorials/examples:\n- [Getting started - Session-based recommendation with Synthetic Data](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/getting-started-session-based)\n- [End-to-end session-based recommendation](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/end-to-end-session-based)\n- [Tutorial - End-to-End Session-Based Recommendation on GPU](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/tutorial)\n\n> I personally found it a bit hard to set up but once you have the momentum and understand the background more, it becomes easier.", "votes": null }, { "id": "2016958", "postDate": "11/04/2022 11:41:41", "content": "

Did you make it work on kaggle? If so, would you be so kind and share a short notebook :)

", "rawMarkdown": "Did you make it work on kaggle? If so, would you be so kind and share a short notebook :)", "votes": null }, { "id": "2016963", "postDate": "11/04/2022 11:56:09", "content": "

With their examples, yes I managed it to run in Kaggle, but with the competition data still in progress. If I can manage, definitely I will share.

", "rawMarkdown": "With their examples, yes I managed it to run in Kaggle, but with the competition data still in progress. If I can manage, definitely I will share.", "votes": null }, { "id": "2016976", "postDate": "11/04/2022 12:05:30", "content": "

would be nice if you could put your notebook where example works here.
\nthan i can try to make it run for comp data aswell during the weekend and share here if i get it running :)

", "rawMarkdown": "would be nice if you could put your notebook where example works here.\nthan i can try to make it run for comp data aswell during the weekend and share here if i get it running :)", "votes": null }, { "id": "2017017", "postDate": "11/04/2022 12:37:55", "content": "

There you go, made it public.
\nhttps://www.kaggle.com/code/snnclsr/transformers4rec-synthetic-data-example

", "rawMarkdown": "There you go, made it public. \nhttps://www.kaggle.com/code/snnclsr/transformers4rec-synthetic-data-example", "votes": null } ], "comments": [ { "id": 2016958, "author_name": "simonveitner", "author_url": "", "post_date": "11/04/2022 11:41:41", "content": "

Did you make it work on kaggle? If so, would you be so kind and share a short notebook :)

", "votes": null, "replies": [ { "id": 2016963, "author_name": "snnclsr", "author_url": "", "post_date": "11/04/2022 11:56:09", "content": "

With their examples, yes I managed it to run in Kaggle, but with the competition data still in progress. If I can manage, definitely I will share.

", "votes": null, "replies": [] }, { "id": 2016976, "author_name": "simonveitner", "author_url": "", "post_date": "11/04/2022 12:05:30", "content": "

would be nice if you could put your notebook where example works here.
\nthan i can try to make it run for comp data aswell during the weekend and share here if i get it running :)

", "votes": null, "replies": [] }, { "id": 2017017, "author_name": "snnclsr", "author_url": "", "post_date": "11/04/2022 12:37:55", "content": "

There you go, made it public.
\nhttps://www.kaggle.com/code/snnclsr/transformers4rec-synthetic-data-example

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2016356": "Hey everyone,\n\nI want to share this repository from NVIDIA that integrates with the HuggingFace transformers library and makes state-of-the-art Transformer architectures available in PyTorch for recommender systems.\n\nhttps://github.com/NVIDIA-Merlin/Transformers4Rec\n\n*From the repository:*\n\nTransformers4Rec is a flexible and efficient library for sequential and session-based recommendation and can work with PyTorch.\n\nIt works as a bridge between NLP and recommender systems by integrating with one the most popular NLP frameworks [HuggingFace Transformers](https://github.com/huggingface/transformers), making state-of-the-art Transformer architectures available for RecSys researchers and industry practitioners.\n\n\"Sequential
\n
\n
Sequential and Session-based recommendation with Transformers4Rec
\n
\n\nTransformers4Rec supports multiple input features and provides configurable building blocks that can be easily combined for custom architectures.\n\nYou can build a fully GPU-accelerated pipeline for sequential and session-based recommendation with Transformers4Rec and its smooth integration with other components of [NVIDIA Merlin](https://developer.nvidia.com/nvidia-merlin): [NVTabular](https://github.com/NVIDIA-Merlin/NVTabular) for preprocessing and [Triton Inference Server](https://github.com/triton-inference-server/server).\n\nAnd in their examples directory, you can find the following tutorials/examples:\n- [Getting started - Session-based recommendation with Synthetic Data](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/getting-started-session-based)\n- [End-to-end session-based recommendation](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/end-to-end-session-based)\n- [Tutorial - End-to-End Session-Based Recommendation on GPU](https://github.com/NVIDIA-Merlin/Transformers4Rec/tree/main/examples/tutorial)\n\n> I personally found it a bit hard to set up but once you have the momentum and understand the background more, it becomes easier.", "2016958": "Did you make it work on kaggle? If so, would you be so kind and share a short notebook :)", "2016963": "With their examples, yes I managed it to run in Kaggle, but with the competition data still in progress. If I can manage, definitely I will share.", "2016976": "would be nice if you could put your notebook where example works here.\nthan i can try to make it run for comp data aswell during the weekend and share here if i get it running :)", "2017017": "There you go, made it public. \nhttps://www.kaggle.com/code/snnclsr/transformers4rec-synthetic-data-example" }, "source": "meta" }