Rivers Have Wings

472 Followers
29 Following
52 Posts
AI/generative artist. Writes her own code. Absolute power is a door into dreaming.
EleutherAI / StabilityAI
Twitter: https://twitter.com/RiversHaveWings
Really we should be thinking about how to prepare students for life in the twenty-first century. I'm not convinced clever ways to prevent them from using machine learning are a big part of doing that.
Turned on federation again in light of recent Twitter implosion and resulting instances worth federating with.

Right now I federate with https://sigmoid.social/
Sigmoid Social

A social space for people researching, working with, or just interested in AI!

Mastodon hosted on sigmoid.social

`
That dread has lifted and beneath is only

This glinting gift, a grave for me and you.
`

#AIArt #art #poetry

Trained a super-resolution Stable Diffusion model (before SD2 introduced an "official" one): https://twitter.com/Buntworthy/status/1593266512474824704?s=20

(You can get that model here: https://huggingface.co/lambdalabs/stable-diffusion-super-res)
---
RT @Buntworthy
Comparing my SD upscaler and rivers':

- original
- @rivershavewings latent upscaler
- My SD upscaler https://twitter.com/RiversHaveWings/status/1589724378492592128
https://twitter.com/Buntworthy/status/1593266512474824704

Justin Pinkney on Twitter

“Comparing my SD upscaler and rivers': - original - @RiversHaveWings latent upscaler - My SD upscaler”

Twitter
In principle that’s ground for rapprochement. Critics could say “look, we were right about what was missing, and they’ve had to adjust their strategy—which is now better.” But that doesn’t seem to be happening; instead a critique that was originally “this won’t work” is hardening and turning into “this must not be allowed to work.” https://irisvanrooijcogsci.com/2023/01/14/stop-feeding-the-hype-and-start-resisting/
Stop feeding the hype and start resisting

Three weeks ago, I wrote a blogpost about how ChatGPT is a “stochastic parrot” (a term coined by Bender, Gebru, McMillan-Major, & Shmitchell, 2021; see also this video for an explan…

Iris van Rooij
If anyone ever needs to compute the integral of softplus (2-fold integral of sigmoid), it's -Li_2(-exp(x)) where Li_2 is the dilogarithm, and you can find a very good and fast numerical approximation to the dilogarithm of real argument in https://arxiv.org/abs/2201.01678.
Comparison of methods for the calculation of the real dilogarithm regarding instruction-level parallelism

We compare different methods for the computation of the real dilogarithm regarding their ability for using instruction-level parallelism when executed on appropriate CPUs. As a result we present an instruction-level-aware method and compare it to existing implementations.

arXiv.org

RT @[email protected]

A classic paper on Reinforcement Learning for Human Feedback (RLHF) is @[email protected]'s "Learning to summarize from human feedback".

Our talented engineer @[email protected] replicated this paper using our trlX library!

Read our report (w/ a code walk-through) here: http://wandb.me/summarize-rlhf-trlx

🐦🔗: https://twitter.com/carperai/status/1613645352514768897

Implementing RLHF: Learning to Summarize with trlX

Implementation of Reinforcement Learning with Human Feedback for text summarization task using CarperAI's trlX framework. Made by Ayush Thakur using W&B

W&B

The Covid bivalent vaccine booster works better than expected.
I've reviewed the cumulative data
https://erictopol.substack.com/p/the-bivalent-vaccine-booster-outperforms

That's in contrast to a
NEJM essay today and 2 lab studies that used a pseudovirus assay without assessing the XBB variants

The bivalent vaccine booster outperforms

A review of the cumulative body of evidence

Ground Truths
I tell you what, having Semantic Scholar send you mail saying there are articles matching your filters "occult" and "folklore" and not worky stuff is a giant mood boost. #search