authorfly

22 ms·

241.
by authorfly 2y ago
Caveat that being from 2022, the Tesseract version used was almost certainly v4 (if Linux), rather than v5 which is much better (and widely available on Windows in 2022, but not Linux yet). However Tesseract is quite behind still as you not
242.
Ask HN: What at first great things were crutches for your startup?
3 points by authorfly 2y ago|0 comments
243.
by authorfly 2y ago
Because they enable you to involve things typically unrelated (i.e. two types of embeddings), think and understand and manipulate them in a simpler mathematical model (a single vector) and see how you can then apply the most fundamental mat
244.
by authorfly 2y ago
And the key takeaway of the T5 paper was also how "instructions" could vary massively. But the main instructions were two types - summarise x, and translate y to z. Translation in many ways was the root of seeing that multiple tas
245.
by authorfly 2y ago
Yes - This book alongside one on AJAX (oh the days of ready state). I still had databases named mismatch_<x> up until about 2014 IIRC! The book was a perfect example of frictionless starting to code, jumping over hurdles, skipping who
246.
by authorfly 3y ago
I drafted a few longer responses about the feeling that I had after reading your post after that headline but they were a bit unavoidably asshole-y! So really, just something like "with Tesseract.js" in the headline is all I would
247.
by authorfly 3y ago
This is useful. Why the step of involving LLMs though? I note you cluster the tabs and then GPT-4 is involved to name them. But my tab groups don't usually need names - just the icons tells me what I mostly need to know. Could this wor
248.
by authorfly 3y ago
In practice, beam search doesn't seem to work well for generative models. Temperature and top_k (two very similar parameters) were both introduced to account for the fact that human text is unpredictable stochastically for each sentenc
249.
by authorfly 3y ago
Simon - hope you don't mind me commenting on you in third person in relation to the above. Simon is a great explainer, but I wish he would credit the underlying technology or library (like tesseract.js) a bit more upfront, like you. It
250.
by authorfly 3y ago
Nice demo, I like this. Will this work for programming languages? I often would like to type code by talking and especially, to talk into AI coding tools that help you complete code. Will this allow me to add links to text based on context?
251.
by authorfly 3y ago
Not only anthropoids... https://www.historyhit.com/the-extraordinary-daddy-long-legs...
252.
by authorfly 3y ago
I agree that the dataset will absolutely make it. But for the use case I'm looking at, the LLM has to be available for batch concurrent inference to be useful. Perhaps unlike most research on datasets, time-to-inference-completion is i
253.
Ask HN: Will I get burned relying on OpenAI for finetuning?
2 points by authorfly 3y ago|2 comments
254.
by authorfly 3y ago
Thank you. This is exactly it, perfect. I will now try and detect the speaker, I guess for podcasts where one side asks questions and the other tends to respond it might be easier, but perhaps not!
255.
Ask HN: Is there a whisper-like speech-to-text that detects the speaker?
13 points by authorfly 3y ago|3 comments
256.
by authorfly 3y ago
A great experiment to do would be to sit people down and have a quiz application like Duolingo randomly choose answers, them seeing if the answers are right/wrong. I suspect, much like experiments showing how people prefer/like (F