cmauck10

11 ms·

31.
by cmauck10 3y ago
Having accurate product listings is vital for E-commerce business for various reasons, such as obtaining accurate analytics, enhancing customer experience, improving product discoverability, and maximizing the effectiveness of SEO and adver
32.
Enhancing Product Analytics and E-commerce with Data-centric AI(cleanlab.ai)
2 points by cmauck10 3y ago|1 comments
33.
by cmauck10 3y ago
Most LLMs allow you to specify the temperature parameter that governs the randomness and thus the creativity of the responses. For this experiment I used a very low temperature to ensure consistency for a given prompt.
34.
by cmauck10 3y ago
It's pretty common now for data scientists and ML engineers to validate the quality of their training data being fed into these LLMs, but what about their test data used to evaluate them? I spent some time playing around with the FLAN-
35.
Beware of Unreliable Data in Model Evaluation: A LLM Prompt Selection Case Study(cleanlab.ai)
66 points by cmauck10 3y ago|9 comments
36.
by cmauck10 3y ago
Before modeling a dataset, do you remember to check if it seems IID? Distribution drift and interactions between datapoints (autocorrelation) are common violations of the Independent and Identically Distributed (IID) assumption which make d
37.
Detecting Dataset Drift and Non-IID Sampling: A K-Nearest Neighbors Approach(cleanlab.ai)
4 points by cmauck10 3y ago|1 comments
38.
Datalab: A Linter for ML Datasets(cleanlab.ai)
2 points by cmauck10 3y ago|0 comments
39.
by cmauck10 4y ago
Have you ever wanted to use handy scikit-learn functionalities with your neural networks, but couldn’t because TensorFlow models are not compatible with the scikit-learn API? Now you can with one-line wrappers for TensorFlow/Keras mod
40.
Training Transformer Networks in Scikit-Learn?(cleanlab.ai)
4 points by cmauck10 4y ago|1 comments
41.
ActiveLab: Active Learning with Data Re-Labeling(cleanlab.ai)
4 points by cmauck10 4y ago|1 comments
42.
by cmauck10 4y ago
Labeled data is key to train models, but data annotators often make mistakes. One can collect multiple annotations per datapoint to get a more reliable consensus label, but this is expensive! To train the best ML model with the least data l
43.
Reduce Prediction Error Rate by 70% in Your XGBoost Model(cleanlab.ai)
2 points by cmauck10 4y ago|1 comments
44.
by cmauck10 4y ago
Data scientists and machine learning engineers are able to improve their model performance without having to change the model itself (think hyperparameter tuning). By finding and correcting errors within their training data to improve the q
45.
by cmauck10 4y ago
Following the success of Andrew Ng's previous competition, we wanted to host another one to share the importance of data-centric ai. In this two-part competition competitors will take on text and image modalities, and are tasked with b
46.
Data-centric AI Competition 2023(machinehack.com)
2 points by cmauck10 4y ago|1 comments
47.
[on GitHub trending] Cleanlab: Data-Centric AI Package(github.com)
2 points by cmauck10 4y ago|1 comments
48.
by cmauck10 4y ago
cleanlab is trending on github daily python packages --- https://github.com/trending/python?since=daily It is a data-centric AI package for data quality and machine learning with messy, real-world data and labels.
49.
70% error reduction with tabular data and XGBoost(pub.towardsai.net)
2 points by cmauck10 4y ago|1 comments
50.
by cmauck10 4y ago
Using data-centric techniques with cleanlab produces a 70% reduction in error on tabular data using an XGBoost classifier.
51.
Automatically Detect Annotation Errors in Image/Text Tagging Datasets(cleanlab.ai)
1 points by cmauck10 4y ago|1 comments
52.
by cmauck10 4y ago
When image/document data has multiple labels, many label errors and/or missing tags make it hard to train highly accurate ML models. We're excited to share with the HN community our newest research on algorithms to automatica
53.
Measuring Perception in AI Models(deepmind.com)
2 points by cmauck10 4y ago|0 comments
54.
by cmauck10 4y ago
I think this strikes a common theme throughout much of the gaming and production industry.
55.
by cmauck10 4y ago
We all know that reliability is the Achilles’ heel of modern ML, as predictions are often wrong for out-of-distribution (OOD) inputs that do not resemble previous training data. While scientists have proposed many complex OOD detection a