luckyt

5 ms·

1.
by luckyt 2y ago
I went for NextAuth - the use case was relatively simple, and I wanted maximum control.
2.
by luckyt 2y ago
Clerk has quite a few dark patterns in their free tier, eg: if your app is on Clerk free tier, all your users will be forced to log out and re-login every 7 days (and they try to obfuscate this fact until you're locked in). For this re
3.
by luckyt 2y ago
There are several issues that make the KV cache as-is unsuitable for caching across requests. First, it requires the cached tokens to be in the exact same position in the sentence, this means it's mainly only useful for autoregressive
4.
by luckyt 3y ago
Yeah, I had a similar experience with Chroma DB. On paper, it checked all my boxes. But yea, it's alpha software with the first non-prerelease version only coming out in July 2023 (so it's 3 months old). I ran into some dumb issue
5.
by luckyt 3y ago
It would be great to see more innovation like AI in DAW tools, but there are some challenges. The main constraint is it needs to process in real time, allowing just a few ms to process a sample. Very few neural methods can work with that co
6.
by luckyt 3y ago
I had the same thought, but overall, there is probably an order of magnitude more people using LLMs in applications or fine-tuning them compared to those trying to pretrain LLMs from scratch.
7.
by luckyt 3y ago
I guess this goes to show how challenging it can be to implement transformer neural networks correctly. There are so many ways in which you can make mistakes at various steps, and there is no surefire way of knowing, you'll just have a
8.
by luckyt 3y ago
> I hope we find a path to at least fine-tuning medium sized models for prices that aren't outrageous It's not that bad; there are lots of things you can do with a hobbyist budget. For example, a consumer GPU with 12 or 24 GB V
9.
by luckyt 3y ago
Yea, ONNX runtime is mostly used for inference. The requirements for training and inference differ quite a lot: training requires a library that can calculate gradients for the back propagation, loop over large datasets, split the model acr
10.
by luckyt 3y ago
I found it helpful to start with CUDA on numba since it lets you write GPU kernels in python. Assuming you're like most ML engineers and you're more familiar with python than C++, this allows you to separately learn CUDA concepts
11.
by luckyt 3y ago
Yea, this was my experience too when I tried it out last week for my side project. It's easy to get started, but it's quite complex and disorganized and poorly documented. There are usually several ways to do things (which is by d
12.
by luckyt 3y ago
Hmm, so is your conclusion that dog walking startups are not venture scale? But there are several dog walking startups that have received VC funding, and at least one that's IPO'ed. I agree with all your points that makes these ty
13.
by luckyt 3y ago
Yeah, it's not clear how to estimate the market size or TAM of an idea even if there already is a market, like take random examples of a dog walking or podcasting app. You can easily find sources that claim the dog walking or podcast b
14.
by luckyt 3y ago
I think the author's point broadly still holds -- you can get further with more engineering resources and data, whether you're using 2015 era models or 2023 retrieval-augmented LLMs and fine-tuning. Just that now you can accomplis
15.
How to deploy your deep learning side project on a budget(luckytoilet.wordpress.com)
1 points by luckyt 3y ago|0 comments
16.
Lessons learned after 6 months of building a language learning startup(luckytoilet.wordpress.com)
5 points by luckyt 4y ago|0 comments
17.
by luckyt 7y ago
There's definitely a lot of opportunities for technical debt in machine learning projects that don't exist in usual software development, which makes careful design decisions even more important. Reminds me of this paper, which ta
18.
by luckyt 8y ago
It's really interesting that XGBoost failed when ran on a dataset that had no noise. I've also seen a similar thing occur with the Adam optimizer when training neural networks on perfect synthetic data [1]. Always interesting to t
19.
Deep Learning for NLP: SpaCy vs. PyTorch vs. AllenNLP(luckytoilet.wordpress.com)
6 points by luckyt 8y ago|0 comments
20.
The Ethics of (not) Tipping at Restaurants(luckytoilet.wordpress.com)
1 points by luckyt 8y ago|0 comments
21.
by luckyt 8y ago
Ah, I failed to consider that. The original post is correct.
22.
by luckyt 8y ago
This is incorrect: there's no reason to expect that X and Y will each appear 3 times in 6 trials if their probabilities are equal. If all 3 measurements of X are smaller than all 3 measurements of Y, then you have X < Y with confide
23.
by luckyt 8y ago
Yes, ideally you would carefully read the paper and evaluate its methodology, evaluations, and results, but this requires you to be knowledgeable in the field. That's why peer review exists -- some experts in the field read it and dete
24.
by luckyt 8y ago
You're absolutely right. A good way to tell if a paper is legit is search it on Google Scholar and see how many citations it has: this paper has none. If it's an arXiv preprint (not in a peer-reviewed journal) and also a low numbe
25.
How to read research papers for fun and profit(luckytoilet.wordpress.com)
1 points by luckyt 8y ago|0 comments
26.
What are the real world applications of automaton theory?(luckytoilet.wordpress.com)
2 points by luckyt 9y ago|0 comments
27.
Publishing negative results in ML is like proving dragons don't exist(luckytoilet.wordpress.com)
1 points by luckyt 9y ago|0 comments
28.
by luckyt 9y ago
There's the Transit app, which tells you how long you have to wait for the next train. If more people used it, people would know about delays before entering the platform, and only take the train if they absolutely have to.
29.
by luckyt 9y ago
Ladders sound hard for neural networks. IIRC the original AlphaGo had a ladder-solver hardcoded that would determine if a ladder position is winning or losing and feed this bit into the neural network.
30.
XGBoost learns the Canadian Flag(luckytoilet.wordpress.com)
2 points by luckyt 9y ago|0 comments

More ›