←Ben’s Bites Waybackbensbites.beehiiv.com/p/stable-diffusion-phone10 Nov 2022sent to 3,783opened 61.0%clicks 2,757unsubs 8← 9 Nov · 11 Nov → · random

Hey friends! Welcome to the 108 new folks joining us since yesterday. I'm heading to Cardiff today to visit family (hi dad - he reads this!), relax and eat, of course.

Layoffs, crypto imploding, bad things 3, 4, 5, etc. Social media is tough these days. But the good news is Ben's Bites is here to serve you up a tasty dish of all the optimistic cool stuff going on that transports us into the future.

Let's tuck in and enjoy scrolling some positivity today.


Prompt: List the topics in today’s email:

🚛 Scale's big launches📑 OpenAI in your docs🦾 Robots rearranging objects🤔 AI for creative writing


🫦 BEN'S BITES

Scale has been busy! 2 big things were released:Scale Spellbook871 - an effortless way to create and deploy LLM apps: a language model + prompt or fine-tune. It’s got a bunch of helpful tools baked in. Nipping at the heels of Hugging Face a bit here. waitlist 871🙄

Scale Forge285 - launched on Product Hunt today184 (I hunted it) - marketers and brand managers at eCommerce marketplaces can create stunning visual content for ad creatives, campaigns, and social media. waitlist 556🙄

Coda now has an OpenAI pack390 for its documents. Summarise meeting notes or Slack threads, get new ideas, create tweetable content and much more109. Hector, from Coda, is a reader and created a way to summarise Ben’s Bites emails in seconds - check it out202.

The a16z Podcast has re-launched with Steph Smith as host. In this episode, they talk about “AI and the Creator Economy with Karen X Cheng”308. Cheng argues that AI will play an increasingly important role in this economy, helping creators to produce content more efficiently and effectively.

👋 Too many links?! I created a database for all links mentioned in these emails. Refer 1 friend using this link and I'll send over the link database.

Draw Things: AI-assisted Image Generation in Your Pocket573. Basically, Stable Diffusion on your phone. You can read the blog post behind this app here - it’s a good’un.

An easy way to compare multiple models206 (flan t5, cohere, GPT-3).

Luma AI has launched its NeRF Point Clouds193. And now you can signup and try Luma on the Web617 even if you do not have an iPhone.

I'm so effing excited for Adept to be released - their app that works across your browser looks pretty powerful. Check out their new demo274 - probably overkill for that task but still cool!

A plugin for GPT-3 AI396-110assisted note-taking in Logseq396, now with the ability to generate images from DALL-E.

A model trained on images taken by the James Webb Space Telescope563.

Samsung announced its ‘Samsung AI Researcher of the Year130’ - Simon Du, for exploring deep questions surrounding deep learning

An algorithm that overcomes the problem of AI being too curious176.

Tracker-NeRF - learning 4D reconstruction from a dataset of videos of cats and dogs216. The internet really is full of cat pics.

The first open-source Chinese & English Bilingual Stable Diffusion212, trained on 20M filtered Chinese image-text pairs.

DreamBooth Isometric cities model.341 AI Sims has gotta be soon?

Schmidt Futures announced the first cohort151 of AI Early Career Fellows.

BLOOM: A 176B-Parameter open-access multilingual language model.512 

Remove and suppress inappropriate image parts319 during the diffusion process.

A model for rearranging objects346.

Given a portrait image, this tool generates plausible renditions511 had it been captured in different decades between the 19th and 21st centuries.

This model can learn to identify languages by extracting features from the data167 that are specific to each language and make it more accurate.

Efficiently scaling Transformer inference.307

Dreambooth Voxel Art Model508. I love the look of these! Now to animate them 👀

This is a hot topic: should we think about AI replacing jobs or helping with our jobs? This research217 looks at the potential of natural language generation (NLG) to impact the creative writing process, based on a study of professional writers.

Key takeaways:1. NLG can be helpful for brainstorming, generating story details, and world-building. 2. NLG technologies struggle to preserve the style and authorial voice. 3. In order for AI-powered writing assistants to be more effective, they need to take into account the diverse goals and expertise of human writers.

Counter arguments:1. NLG technologies are not yet advanced enough to be truly helpful for creative writing. 2. AI-powered writing assistants could never replace human writers.

This approach can help to better understand the decision-making process of AI models239 and improve their performance in various tasks.

An AI platform that analyses massive streams of freeform text358 (support tickets, reddit, Zoom transcripts) and spits out a key insights report.

The ERNIE-UniX2 pre-training model integrates multiple learning paradigms, including contrastive learning and language modelling, based on an encoder-decoder architecture. This allows it to learn better joint representations across languages and modalities.112

Who will decide the fate of AI-generated content online108, the moderators or the courts?


🖼 AI IMAGES OF THE DAY

More voxel art because I think it's cool!


👋 SEE YA


Stable Diffusion on your phone
🚛 Scale's big launches 📑 OpenAI in your docs 🦾 Robots rearranging objects 🤔 AI for creative writing

less more clicks · number = clicks

Most clicked in this issue

871×2scale.com/spellbook
617captures.lumalabs.ai/me
573draw.nnc.ai
563huggingface.co/dallinmackay/JWST-Deep-Space-d…
556scale.com/catalog
512arxiv.org/abs/2211.05100
511facesthroughtime.github.io
508huggingface.co/Fictiverse/Stable_Diffusion_Vo…
396×2github.com/briansunter/logseq-plugin-gpt3-ope…
390coda.io/@coda/openai-for-coda
358askviable.com
346arxiv.org/abs/2211.04604

786 readers clicked at least once. 36 links, 36 with clicks. Heat is relative to the most clicked link in this issue.

Beehiiv · digest