benob

10 ms·

121.
by benob 2y ago
Been using EndevourOS after stability issues with Manjaro...
122.
by benob 2y ago
This is the next step for SEO: be able to game ChatGPT prompts trying to filter out SEO crap...
123.
by benob 2y ago
Try it here: https://huggingface.co/spaces/stabilityai/stable-diffusion-3...
124.
by benob 2y ago
I would like more sites asking consent when passing info to a 3rd party becomes necessary (in relation to a requested service, like when you use location services in a phone app, for example) instead of asking for bulk consent for the whole
125.
by benob 2y ago
Difflogic should lead to very efficient hardware implementations of inference for a wide range of problems. But is it going to allow to generate the logical chains that elude current LLMs? --- edit --- After reading the paper more thoroughl
126.
by benob 2y ago
LLMs are high-order Markov chains with a factored estimator. N-gram LMs are also Markov chains with a different estimator.
127.
by benob 2y ago
Why is it iOS17+ only?
128.
by benob 2y ago
Do they make anechoic swimming pools that would absorb waves like acoustic anechoic chambers?
129.
by benob 2y ago
What are uses of LD_PRELOAD that make us keep it?
130.
by benob 2y ago
How about the same thing on a 68K emulated with an FPGA? (ok it will probably cost more than 5 pounds)
131.
by benob 2y ago
Is anyone running 9 as a daily driver? What hardware is suggested?
132.
by benob 2y ago
> OpenELM: An Efficient Language Model Family with Open-source Training and Inference Framework https://arxiv.org/abs/2404.14619 Apple is pushing for open information on LLM training? World is changing...
133.
by benob 3y ago
What makes you think that the accounts were not compromised only recently?
134.
by benob 3y ago
It doesn't make sense to show the progress of AI processing for featured songs which have been seen by dozens of visitors and surely are cached. Is this trick related to copyright issues?
135.
by benob 3y ago
Can Parquet be read/parsed in almost every programming language with very little effort?
136.
by benob 3y ago
There are efforts to port pico8 to microcontrollers, but the real problem with lua is memory (easily requires 4MB of memory which is only available on high-end microcontrollers). https://github.com/DavidVentura/PicoPico
137.
by benob 3y ago
Is the rpi rated for environments with high humidity? Shouldn't it be in a water resistant enclosure?
138.
by benob 3y ago
I had to check the date, I though it was April 1st.
139.
by benob 3y ago
This reminds me of bias tuning, a LoRA competitor. One can get decent adapters by only finetuning a vector added to each linear layer activations. I think I saw it first while reading [1] but there are other instances. [1] https:/&#x2
140.
by benob 3y ago
ollama
141.
by benob 3y ago
Look at how RingAttention is implemented: it's blockwise attention distributed among many GPUs, in other words bruteforce parallelization. For inference they use TPUs v4-128, not running this at home any time soon.
142.
by benob 3y ago
Google had it in 2015 [1] [1] https://arxiv.org/abs/1506.05869
143.
by benob 3y ago
I wonder how an ethical board was convinced to allow this kind of research
144.
by benob 3y ago
Could it be that they are mandated by law, like the French banking sector for credit scoring, to use an interpretable method?
145.
by benob 3y ago
For some reason I though the goal was to print (with a 3d printer) a 3d projection of the 4d content of the video. That would be cool...
146.
by benob 3y ago
There is also Whisperx, a modification of whisper with accurate word timing and confidence scores. It gives pretty good subtitles.
147.
by benob 3y ago
I think the point of the parent was that you can build expertise in all of those in a relatively short time frame (esp. when more developers will start building up experience), compared to acquiring a large customer base or a large dataset.
148.
by benob 3y ago
What explains the huge difference in perf with the approach from 2019 mentioned at the end? For image classification, couldn't one use the residual from applying a jpeg codebook learned on a training example as metric?
149.
by benob 3y ago
This is so true.
150.
by benob 3y ago
Do they make a teenager-sized model?

More ›