jerpint

10 ms·

121.
by jerpint 2y ago
The biggest problem these days is that it’s very easy to hack together a solution for a problem that, at first glance, seems to work just fine. Understanding the limits of the system is the hard part, especially since LLMs can’t know when t
122.
by jerpint 2y ago
I made a clock out of MNIST digits that are obtained travelling the latent space of an autoencoder: https://www.jerpint.io/blog/mnist-clock/
123.
by jerpint 2y ago
Could also work for social media which can be hard to scrape
124.
by jerpint 2y ago
I found grounding-dino better than Florence and faster
125.
by jerpint 2y ago
Funny, I was actually thinking of micromouse when I made my comment, thank you for clarifying
126.
by jerpint 2y ago
I’m not sure I agree, text is full of compressed information but lacks all of the visual cues we all use to navigate and understand our world. Video data also has temporal components which text is really bad at.
127.
by jerpint 2y ago
They mention that the input grid contains the layout of the maze, but isn’t the point of solving a maze that you don’t have a global view of it?
128.
by jerpint 2y ago
Video data is still very much untapped and likely to unlock a step function worth of data. Current image-language models are trained mostly on {image, caption} pairs with a bit of extra fine tuning
129.
by jerpint 2y ago
If these models could run offline, 100% locally, with expected privacy, Im up for it, i want to use a product, not be training data
130.
by jerpint 2y ago
Just a few seconds of speech gets you pretty far with a lot of models
131.
by jerpint 2y ago
Im assuming there are many open source alternatives to do this? Why pay to run edge devices?
132.
by jerpint 2y ago
Using a colab with printed outputs could be a good option to at the very least hint to reproducing results independently
133.
by jerpint 2y ago
I recommend VScode for that use case (other editors probably support this too) where you can ssh to your cluster, access all your files in the tree, and run Jupyter notebooks from VScode directly using HPC resources, very nice workflow
134.
by jerpint 2y ago
I wonder if this could be combined with fastHTML to have a full python html rendered client side
135.
by jerpint 2y ago
There’s an added objective in some of these models to make them more aesthetically pleasing based on subjective crowdsourced data that very likely contributes to this
136.
by jerpint 2y ago
One thing I find particularly interesting is that SOTA video understanding requires significantly less parameters than SOTA language understanding. Are we just overfitting way too much to language? Also how long until SAM gets aligned to an
137.
by jerpint 2y ago
Ive used p5js to build my own Tetris game with vim bindings :) https://www.jerpint.io/blog/tetris/
138.
by jerpint 2y ago
Models can easily regurgitate back training data verbatim, so anything private can be in theory accessed by anyone without proper access to that file
139.
by jerpint 2y ago
Awesome, thanks for sharing
140.
by jerpint 2y ago
Why do I need to supply an API key to try their demo? Seems absurd
141.
by jerpint 2y ago
I go back regligiously to this post whenever I need a quick visual refresh on how transformers work, I can’t overstate how fantastic it is
142.
by jerpint 2y ago
I have my own small blog , and am more and more tempted to “poison” it with made up facts about me to see if/when it gets scraped by LLMs
143.
by jerpint 2y ago
Still blazing fast on iPhone
144.
by jerpint 2y ago
I implemented a version of game of life using 2D convolutions from scipy, I’d be curious to see how it compares to GPU crunching
145.
by jerpint 2y ago
Nice! I did a game of life using stable diffusion recently, check it out :) https://www.jerpint.io/blog/diffusion-gol/
146.
by jerpint 2y ago
Why are llama 3 weights listed as partially available ? Also Olmo didn’t release a paper per se but a very detailed blog post / cookbook for reproduction which should count to a perfect score imo
147.
ControlNet Game of Life(jerpint.io)
72 points by jerpint 2y ago|6 comments
148.
by jerpint 2y ago
Not yet, some models in that direction are LTU-AS
149.
by jerpint 2y ago
Most “serious” companies would probably rewrite it from scratch anyway
150.
by jerpint 2y ago
It could be both. In my case my keys were used to call OpenAI, almost certain they were leaked from my Spaces secrets

More ›