Skip Navigation
InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)KR
kraegar @lemmy.world
Posts 1
Comments 14
What tools do you use?
  • I know it's an old thread, but I want there to be more DS stuff in the Lemmy.

    Python is arguably the defacto standard. Jupyter gets used a lot (but shouldn't be used to frequently IMO).

    I haven't used R but it seems to be a distant second to Python. That being said, it really depends on the industry!

    Generally my setup is VSCode and Anaconda developing on remote VMs.

  • PSA: Mastodon is NOT Twitter and does not aim to be.
  • George Takei feels like half the reason I still even used twitter.

    A lot of the communities I interacted with because people shilling their products, but George remained.

    Swapping to Mastodon was something I had low expectations for, but it turned out great.

  • Is anybody else more active here then they were on Reddit?
  • I know exactly what you mean. I even find that lemmy has had a less argumentative comments section. It is partially that I have avoided the wild communities, but it is still alarmingly chill.

    Sorry for having a conversation. I didn't realize we came here to TALK. /s

    I haven't found a good "wall of text" community yet. I am doing AI/ML research and was hoping that would grow quickly on lemmy. The Reddit equivalents had lots of comments and content.

  • Machine Learning @lemmy.ml kraegar @lemmy.world

    Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift

    openreview.net Reversible Instance Normalization for Accurate Time-Series...

    Statistical properties such as mean and variance often change over time in time series, i.e., time-series data suffer from a distribution shift problem. This change in temporal distribution is one...

    I found this paper when implementing TSMixer in darts and thought it was worth sharing.

    As far as I know, it may not be relevant to non-timeseries problems. That being said, does anybody know of instance normalization outside of timeseries? It is obviously similar to batch normalization but is slightly different.

    I added RIN to an N-HiTS model using the Energy benchmark dataset and got 21% and 22% reduction in MSE and MAE, respectively. This required two lines of code to be added to the model definition and increased compute time in no noticeable manner.

    0
    TinyStories: How Small Can Language Models Be and Still Speak Coherent English?
  • NLP isn't an area I know much about, but I find it really interesting that they used GPT-4 as the scoring mechanism. I do have concerns about this after the MIT "100% OF QUESTIONS CORRECT" fiasco recently, but it will likely be as equally or less biased than a human marker.

  • What are your tokui wazas?
  • I can't pick just one technique because newaza and tachiwaza techniques both have their own allure.

    For tachiwaza, I love uchi mata and seoi nage. It is so satisfying to hit a big throw with a clean landing. In newaza, juji gatame is my go to.

  • How long have you been practicing judo? What got you into judo?
  • I started judo because there was a demo at my elementary school and I thought it was cool.

    17 years later, I have gotten to travel all over the world and meet a ton of really cool people. I haven't done judo the past few years due to COVID and grad school; however, I finally have time to get back into it again and I am really excited!

  • Dear ML users/researchers, what hardware do you use?
  • I have a mixed approach: I have a laptop with an RTX3060 (surprisingly good for small models and dev work). There are a few beefy servers I have access to through work/school which I leverage when I need more resources.

    If I didn't have access to work/school servers, I would likely go with cloud or build a new desktop for myself. Most of my work has been with time series forecasting and anomaly detection so the models tend to be smaller. If you need bigger models this wouldn't work well for you.

  • What research area currently excites you?
  • I am really excited about GNNs too. I just submitted my PhD thesis on time series forecasting and anomaly detection for LTE/5G networks and I really think the next big jumps in that space are going to be graph-based.

    My post-grad employment is in industry and I have a sneeking suspicion that the company will have me looking at them and I am pumped.