sb-ai-lab / replay Goto Github PK

🚀 Feature Request

Add .count() just after cache methods in all place

Motivation

It is needed for more honest estimation of execution time

Checklist

• feature proposal description
• motivation

To reproduce:
Using python 3.8.10 kernel execute examples/09_sasrec_example.ipynb (or examples/10_bert4rec_example.ipynb) [release/v0.16.0 branch]
->
On the fitting stage the error occurs: "collections.OrderedDict" object has no attribute 'name'

will it work on Windows11 for example LightFM
LightFM
classreplay.models.LightFMWrap(no_components=128, loss='warp', random_state=None)
Wrapper for LightFM.
from
https://sb-ai-lab.github.io/RePlay/pages/modules/models.html#replay-recommenders

🚀 Feature Request

Add to recommenders UCB algorithm

Motivation

it is needed for covering RL area of algorithms

Proposal

Alternatives

Additional context

Checklist

• feature proposal description
• motivation
• additional context / proposal alternatives review

to ensure everything is up-to-date

🐛 Bug

Linux Mint 21.2_x64 - Replay-3.2.4.deb

Even if you have downloaded the models, it does not allow you to press the "continue" button

https://imgur.com/RJafAQr

greetings and thanks for everything

🚀 Feature Request

Implement metrics with normalisation: MAP, NDCG, HitRate, Roc Auc, Precision, Recall
https://arxiv.org/pdf/1801.07030.pdf

Motivation

For RS simulator

🐛 Bug

user_test_size in UserSplitter

🚀 Feature Request

Update model with new data in efficient way.
https://docs.google.com/spreadsheets/d/1oULhfyWgA3rqb3M9NzlBiWeY80xNYizUodyv0tQW9Js/edit#gid=888497155

Motivation

Work with bigger data volumes, Update model without re-fit.

🐛 Bug

I have 'user_id' column in my data, so when I try to use some of the methods or functions, I get this error:
pyspark.sql.utils.AnalysisException: cannot resolve 'user_idx'

To Reproduce

Examples where 'user_idx' column is required:
replay/models/base_rec.py", line 321, in _fit_wrap users = log.select("user_idx").distinct()

replay/filters.py", line 28, in min_entries entries_by_user = data_frame.groupBy("user_idx").count() # type: ignore

🚀 Feature Request

Find out how to preprocess and pass model weights/vectors to Kafka

Motivation

Models inference

🚀 Feature Request

Replace RDD-based operations in metric calculation with UDF to replace with Scala UDF in future

Motivation

Metrics calculation speed up

🐛 Bug

UserSplitter неисправно работает c параметром user_test_size

To Reproduce

data_frame = pd.DataFrame({"user_idx": [1,1,1,2,2,2],
"item_idx": [1,2,3,1,2,3],
"relevance": [1,2,3,4,5,6],
"timestamp": [1,2,3,3,2,1]})

data_frame = convert2spark(data_frame)

UserSplitter(2,1,seed=80083).split(data_frame)[0].toPandas()