Sebastian Nickels
@snickels.io
Data Science @ Delivery Hero building time estimation models.
Interested in DS/ML, engineering, open/responsible tech, tools for thought and learning in general.
Interested in DS/ML, engineering, open/responsible tech, tools for thought and learning in general.
With first time I’m referring to after reading your comment. In subsequent restarts it was a bit faster probably due to caching.
2s does not seem much, but when I‘m in a conversation and I want to quickly note something down, it‘s just too much and breaks my flow. Starting notes is faster than that
2s does not seem much, but when I‘m in a conversation and I want to quickly note something down, it‘s just too much and breaks my flow. Starting notes is faster than that
December 18, 2024 at 9:42 AM
With first time I’m referring to after reading your comment. In subsequent restarts it was a bit faster probably due to caching.
2s does not seem much, but when I‘m in a conversation and I want to quickly note something down, it‘s just too much and breaks my flow. Starting notes is faster than that
2s does not seem much, but when I‘m in a conversation and I want to quickly note something down, it‘s just too much and breaks my flow. Starting notes is faster than that
When I started it for the first time, it felt like 2-3s but it showed 1.4s in the results.
The majority of it is spent on total app startup and only 30% is spent on plugins as I use lazy loader to asynchronously load the slowest plugins such as Exaclidraw.
The majority of it is spent on total app startup and only 30% is spent on plugins as I use lazy loader to asynchronously load the slowest plugins such as Exaclidraw.
December 18, 2024 at 12:18 AM
When I started it for the first time, it felt like 2-3s but it showed 1.4s in the results.
The majority of it is spent on total app startup and only 30% is spent on plugins as I use lazy loader to asynchronously load the slowest plugins such as Exaclidraw.
The majority of it is spent on total app startup and only 30% is spent on plugins as I use lazy loader to asynchronously load the slowest plugins such as Exaclidraw.
I‘d agree with the others mentioning mobile performance - an alternative here would be to build some kind of quick add widget for the Home Screen to skip that loading time.
December 17, 2024 at 7:13 PM
I‘d agree with the others mentioning mobile performance - an alternative here would be to build some kind of quick add widget for the Home Screen to skip that loading time.
In this space there are a lot of techniques that are just assumed to be true but never questioned - this reminds me of the paper on SMOTE that showed that it doesn’t seem to do what people think it does (arxiv.org/abs/2201.08528)
To SMOTE, or not to SMOTE?
Balancing the data before training a classifier is a popular technique to address the challenges of imbalanced binary classification in tabular data. Balancing is commonly achieved by duplication of m...
arxiv.org
December 9, 2024 at 1:03 PM
In this space there are a lot of techniques that are just assumed to be true but never questioned - this reminds me of the paper on SMOTE that showed that it doesn’t seem to do what people think it does (arxiv.org/abs/2201.08528)
That‘s an interesting find! Assuming you properly sample your data from the population, one should refrain from using stratification. Although I can see edge cases e.g. when samples from different classes are harder to acquire or incomplete (e.g. on weekday x samples are incomplete)
December 9, 2024 at 1:01 PM
That‘s an interesting find! Assuming you properly sample your data from the population, one should refrain from using stratification. Although I can see edge cases e.g. when samples from different classes are harder to acquire or incomplete (e.g. on weekday x samples are incomplete)
How does it compare to Cursor? I found Cursor better than Copilot but haven‘t tried out Windsurf yet
November 17, 2024 at 11:53 AM
How does it compare to Cursor? I found Cursor better than Copilot but haven‘t tried out Windsurf yet