ScoutOrgo

9 ms·

31.
by ScoutOrgo 7y ago
I hope the gestures are better than when they first came out with the "pill". I tried it for a week and it was an awful experience. Maybe it's because I use the double tap on app switcher to quickly get to last app a lot and
32.
by ScoutOrgo 7y ago
And what if you had the battery replaced?
33.
by ScoutOrgo 7y ago
I've gone through all previous iterations of their courses, so have familiarity with using pytorch and the fastai library, but haven't done the best at keeping up with their rapid pace. This course helped me understand the buildin
34.
by ScoutOrgo 7y ago
It's more data for making better predictions on default. Better predictions = better returns, kinda like the stock market.
35.
by ScoutOrgo 7y ago
fast.ai has some pretty good defaults. The main HP to choose in their tabular learner are the neurons in the linear layers and the number of embeddings for each category.
36.
by ScoutOrgo 7y ago
I think it is actually preferable to start by converting categorical variables to numeric most of the time, even if they are not ordinal. The RF algo can separate off individual classes with 2 splits (e.g. <=7 then >=7) if a single cl
37.
by ScoutOrgo 7y ago
I have been wanting to do this, as I set up the Chase email notifications, but never got motivated to tackle what you used zapier for. One thing that it seems you could do is use Google sheets for your constants vs coded in the repo. That w
38.
by ScoutOrgo 8y ago
Feels worse to me than what Google was heavily fined for.
39.
by ScoutOrgo 8y ago
Fair enough, thanks for answering.
40.
by ScoutOrgo 8y ago
What is your business model if your platform is free? Either that price has to change, or you plan on making money on the same thing all other free services run on: data. The site isn't very upfront about it, which is the sketchy part.
41.
by ScoutOrgo 8y ago
So we should just ban all videos of children then?
42.
by ScoutOrgo 8y ago
I've gone through each iteration of the fast.ai courses and must say that it is worth it every time. The latest 2019 courses are again excellent. Even though I've covered the concepts before, I always learn a ton from every lesson
43.
by ScoutOrgo 8y ago
No mention of ULMFiT? http://nlp.fast.ai/classification/2018/05/15/introducting-ul...
44.
by ScoutOrgo 8y ago
I had a similar experience with a personal project. I scrapped NFL player/game stats dating back past 2000 and spent a bunch of time on the data, feature engineering, and modeling. Although my original goal was to beat vegas models (g
45.
by ScoutOrgo 8y ago
For a site design blog, the "What to read next" covering the bottom corner of text isn't too great.
46.
by ScoutOrgo 8y ago
Definitely not true. Transfer learning a model trained on a corpus then fine tuned on labeled classes is current SotA approach [1]. [1] http://nlp.fast.ai/category/classification.html
47.
by ScoutOrgo 8y ago
This notebook from fast.ai is more in line with what op mentioned. It starts from nothing but Python and slowly swaps out chunks of code with functions/classes from Torch/fastai: https://github.com/fastai/fast
48.
by ScoutOrgo 8y ago
What podcast?
49.
by ScoutOrgo 8y ago
Add a marketplace for resells to same place face value tickets are sold. Would be same site, but could be different pages or w/e. Cap resells at face value, removing incentive to flip. Then to prevent resells, would need an app with ti
50.
by ScoutOrgo 8y ago
> Joins should be indented to the other side of the river and grouped with a new line where necessary Don't agree here. Having all tables down the left makes it very easy to quickly scan all the tables that are being joined. With in
51.
by ScoutOrgo 8y ago
It is 64 hyper-parameters, not variables, which is a huge amount. Stopping rules are also a requirement of tuning. Can't just have a while True:. If an objective function doesn't improve after 100 (or any #), there needs to be log
52.
by ScoutOrgo 8y ago
I already do something like this, but with a lot of success: anything that requires an action on my end stays in my inbox and anything that I get done or is purely informational goes to a folders (typically project related). All this seems
53.
by ScoutOrgo 8y ago
Just reminding me of someones name would be nice... I'm pretty bad at that.
54.
by ScoutOrgo 8y ago
It really is frustrating using new software for work that has only icons with ~3 second hover delay of showing the action label.
55.
by ScoutOrgo 9y ago
Fast.ai's (fast.ai) deep learning and machine learning courses. No ads, good notes/forum, and very approachable material for anyone that knows basic coding.
56.
by ScoutOrgo 9y ago
I would disagree. I track on google flights 3+ months before I want to go and feel that I get a reasonable price every time. 3 months in advanced obviously doesn't work for everyone, but using google flights (or any other tracking site
57.
by ScoutOrgo 9y ago
fast.ai courses are the best I've taken. You start with implementing things immediately then get into the details later. It makes it less like a grind to learn everything beforehand.
58.
by ScoutOrgo 9y ago
In my experience, audiobooks cost too much. They are more expensive than digital or hard copies, and the "subscription" is just monthly book credits. I get netflix for $11/month for unlimited use. Wish amazon would get some
59.
by ScoutOrgo 9y ago
I'd suggest fast.ai mooc before keras docs. I took Hintons course and tried to learn through tf/keras docs, but wasn't able to really get going until I found fast.ai. Some of the best "classes" I've ever watche
60.
by ScoutOrgo 9y ago
So your competitive advantage is doing the exact same thing these companies are doing, except with a different demographic?

More ›