Top all-time · #2260
adept.ai
Company blog · adept.ai/blogaiprogramming
adept.ai is the company blog at adept.ai, writing about ai and programming. Since 2022, 34 different Hacker News readers have linked it 42 times, most often “Act-1: Transformer for Actions”. No feed was found, so read it on the site itself.
Too few sampled comments (25) to say how often HN recommends it.?
Why HN links it
HN readers link Adept's blog for its ACT-1 action model and its Fuyu-8B and Persimmon-8B model posts. Most links cite these posts as examples of AI agents that operate software and of multimodal models.
Summarised by Claude from a sample of 25 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Act-1: Transformer for Actions8 people · 9 mentions · first linked 2022-09-14
- Adept Open Sources 8B Multimodal Modal Fuyu4 people · 4 mentions · first linked 2023-10-18
- Adept is getting acqui-hired by Amazon AGI4 people · 4 mentions · first linked 2024-06-28
- Adept AI Labs3 people · 4 mentions · first linked 2022-04-26
- Releasing Persimmon-8B3 people · 3 mentions · first linked 2023-09-07
- Adept Fuyu-Heavy: A new multimodal model3 people · 3 mentions · first linked 2024-01-24
- /blog/act-13 people · 3 mentions · first linked 2023-03-01
- /2 people · 2 mentions · first linked 2023-02-22
- The Adventure of the Errant Hardware2 people · 2 mentions · first linked 2023-09-20
What Hacker News says?
https://www.adept.ai/post/introducing-adept pretty much right after writing the transformers paper two of the co authors formed this company
Comment linking /post/introducing-adept
on “DeepMind: A Generalist Agent”Submitted Act-1: Transformer for Actions as a story · 267 points
HN discussionWorth noting that current models like Google LaMDA appear to already have access to the live Internet. The LaMDA paper says it was trained to request arbitrary URLs from the live Internet to get text snippets to use in its chat contexts. Then you have everyone else, like Adept https://www.adept.ai/post/introducing-adept (Forget anything about how secure 'boxes' will be - will there be boxes at all?)
Comment linking /post/introducing-adept
on “John Carmack's new AGI company, Keen Technologies, has raised a $20M round”Submitted Silent Data Corruptions: The Boogeyman of LLM Training as a story · 29 points
HN discussionhttps://www.adept.ai/act You can combine traditional RPA with modern ML and do very interesting things.
Comment linking /act
on “Web automation: Don't use Selenium, use Playwright”Submitted Adept Open Sources 8B Multimodal Modal Fuyu as a story · 16 points
HN discussionA bit off topic but every time I see DTrOCR I remember that marketing is a good idea because DTrOCR[1] and Fuyu[2] are basically the same architecture[3]. [1] https://arxiv.org/pdf/2308.15996v1.pdf [2] https://www.adept.ai/blog/fuyu-8b [3] If you don't want to search for the figures I made a tiny post about it on my weblog: https://weblog.snats.xyz/posts/2024/02/16/
Comment linking /blog/fuyu-8b
on “RAGFlow is an open-source RAG engine based on OCR and document parsing”Submitted Releasing Persimmon-8B as a story · 16 points
HN discussionI believe we are almost at peak GUI. The endgame here is that all these crappy GUIs that are getting worse every year will be relegated to a role of being APIs for AI agents. Instead of clicking around and filling out forms and waiting for loading spinners all the time, we'll just tell a large language model what we want to do in English, and it will go off and screen-scrape a bunch of apps and websites, do all the…
Comment linking /
on “Despite faster broadband every year, web pages don't load any faster”Submitted Adept is getting acqui-hired by Amazon AGI as a story · 6 points
HN discussionThis dream of no-code web data extraction and manipulation is still alive. The best example I've seen is https://www.bardeen.ai/ : a local-first in-browser solid collection of automations commands and integrations workflows. The future, though, will more likely look like this https://www.adept.ai/act
Comment linking /act
on “Yahoo Pipes”Submitted Adept Fuyu-Heavy: A new multimodal model as a story · 5 points
HN discussion> If GPT-6 was only larger, it would still only know the things in its training set, whereas a human can take arbitrary time to think, can go out and perform experiments, can Google/GPT things themselves, etc. A GPT model already has many ways of thinking and self-distilling; for example, it can give itself a lot more time to think using 'inner monologue' techniques, which make it capable of calculating out step by…
Comment linking /post/introducing-adept
on “The Scaling Hypothesis (2021)”Submitted Introducing Adept Experiments – use AI workflows to delegate repetitive tasks as a story · 5 points
HN discussionI wrote an article about these affecting LLM training at https://www.adept.ai/blog/sherlock-sdc
Comment linking /blog/sherlock-sdc
on “Silent data corruptions at scale (2021)”Submitted The Adventure of the Errant Hardware as a story · 5 points
HN discussionThat would be Persimmon-8B, no? https://www.adept.ai/blog/persimmon-8b
Comment linking /blog/persimmon-8b
on “Mistral 7B, the most powerful language model for its size to date, Apache 2.0”Submitted Adept Raises 350M to build AI that learns how to use software as a story · 3 points
HN discussionPersimmon-8B from Adept did it first no https://www.adept.ai/blog/persimmon-8b
Comment linking /blog/persimmon-8b
on “Mistral 7B, the most powerful language model for its size to date, Apache 2.0”Submitted Adept AI Labs as a story · 3 points
HN discussionHere is a little GIF demo: https://twitter.com/timo_schick/status/1624058382142345216 The possibilities of this line of research are endless. When language models can call APIs and/or use UIs, they can become the unified natural language interface to any software, any website, any app. See also https://www.adept.ai/act . Siri and Alexa and Google Assistant are dead ends. Language models trained to use tools will fin…
Comment linking /act
on “Toolformer: Language Models Can Teach Themselves to Use Tools”Submitted An Update to Adept as a story · 2 points
HN discussionThe interesting point here is that the two founders had previously co-founded a similar AI company, Adept, that raised a much larger amount of funding ($415m vs $57m here) to build foundation models: https://www.theinformation.com/briefings/two-co-founders-of-... Why did they leave only to create a rival company in the same space? One could argue that Adept is more focused on native multimodality and agentic models…
Comment linking /blog/act-1
on “Essential AI”Submitted Adept.ai Founders Left for Amazon as a story · 2 points
HN discussionYou might be interested in this: https://www.adept.ai/blog/act-1
Comment linking /blog/act-1
on “Microsoft Kosmos-1: A Multimodal Large Language Model”Submitted An Update to Adept as a story · 2 points
HN discussionhttps://www.adept.ai/blog/fuyu-8b Doesn't seem like multimodal inference is as complex as you're making it out to be. It all hinges on what you claim "understanding" to be; are trained weights understanding?
Comment linking /blog/fuyu-8b
on “Fireside Chat with Ilya Sutskever and Jensen Huang”Submitted Building Powerful AI Agents with Adept and AWL as a story · 1 points
HN discussionIt's worth noting that at least one startup (Adept AI) is literally building exactly this as a full end-to-end system. See: https://www.adept.ai/act
Comment linking /act
on “Show HN: Help, I let GPTChat control my computer”Submitted Multimodal model from Adept world-strongest after GPT-V and Gemini as a story · 1 points
HN discussion
