Topfi

54 ms·

721.
by Topfi 3y ago
Ipse se nihil scire id unum sciat.
722.
by Topfi 3y ago
I have refrained from commenting on this situation due to the persistent lack of outside transparency, not knowing enough to make a judgment on the specific motivations everyone involved in this had/has and will reserve making a full j
723.
by Topfi 3y ago
That is very true, but you had a simultaneous stream of amazed reactions that counteracted a lot of negative reporting in the beginning. Such a luxury wasn't afforded to Metas efforts at the time.
724.
by Topfi 3y ago
I feel that the article doesn't really cover what aspects made Galactica more open to criticism and led to its reception, outside of pointing to a lack of awareness for hallucinations in the wider public, which were equally present dur
725.
There is a new version of GPT-4 Turbo now live in ChatGPT(twitter.com)
30 points by Topfi 3y ago|88 comments
726.
by Topfi 3y ago
Sam Altman just posted the following on X: >>there is a new version of GPT-4 Turbo now live in ChatGPT; you should hopefully find it a lot better! please let us know what you think.<< ChatGPT in this case is most likely referrin
727.
by Topfi 3y ago
Thanks for the clarification. Opinions will differ, of course and I cannot say whether the time required is justified or calls for optimizations, but personally, I view that moving this to the first search query is less elegant than having
728.
by Topfi 3y ago
whisper-large-v2 was/is available via API from OpenAI, simply wasn't as much of a focus in their recent communications.
729.
by Topfi 3y ago
I am very much looking forward to, but also dreading, testing gpt-4-turbo as part of my workflow and projects. The lowered cost and much larger context window are very attractive; however, I cannot be the only one who remembers the differen
730.
MediaTeks New Flagship Dimensity 9300 – All Big Core Design(corp.mediatek.com)
2 points by Topfi 3y ago|1 comments
731.
by Topfi 3y ago
Titel was editorialized, original: MediaTek’s New All Big Core Design for Flagship Dimensity 9300 Chipset Maximizes Smartphone Performance and Efficiency
732.
by Topfi 3y ago
Here is a direct link to the motion for those interested: https://scribd.com/document/681174239/Order-on-motion-to-dis...
733.
by Topfi 3y ago
I slightly edited the original title due to its length: "Exclusive: Google confirms with Notebookcheck it blocked benchmarks during Pixel 8, Pixel 8 Pro review embargo period". Historically, I have witnessed a lot of odd, problema
734.
Google confirms it blocked benchmarks during Pixel 8 review embargo(notebookcheck.net)
2 points by Topfi 3y ago|1 comments
735.
by Topfi 3y ago
> But of course, on-device generative AI is really complex: 150 times more complex than the most complex model on Pixel 7 just a year ago. Tensor G3 is up for the task, with its efficient architecture co-designed with Google Research. Th
736.
by Topfi 3y ago
Ok, but we were talking about the default, the average for the majority of consumers. Did any of them care about the far more significant latency that their TVs processing imposes (or did impose, as now many TVs switch to "game mode&qu
737.
by Topfi 3y ago
> laws of physics Honest question, which of those is in any way applicable to this technology? Besides the fact that these services already exist and, provided you have decent infrastructure, which will only become more accessible as tim
738.
by Topfi 3y ago
To add some important context, as far as I can tell (I am not very knowledgeable in US filing, so someone please correct me if I am wrong), this was submitted in late 2019, and it stands to reason that a lot of the more recent developments,
739.
by Topfi 3y ago
I agree with you that there is a lot of room for ambiguity here. I'd imagine their target are e.g. architectural studios and advertisers, but how much interactivity is too much or whether this is separated by the platform a UE project
740.
Show HN: AI Detector.eu – Always Accurate(ai-detector.eu)
6 points by Topfi 3y ago|1 comments
741.
by Topfi 3y ago
Amazing to hear from someone so deeply involved in the design process. Thanks for clarifying that some parts of the article are not 100% accurate. Concerning: > [...] really pushing the bounds of what you can do with an in order core and
742.
by Topfi 3y ago
> the customer [... is] fined heavily. Sorry, but do you have a source for that claim? Cause I have yet to witness the Guardia di Finanza charging a customer in any way whatsoever.
743.
by Topfi 3y ago
Once again, my respect for the work of the Asahi team and especially Linas GPU related efforts grows further. Great to see that she was officially recognized[0] and received a bounty for her efforts. [0] https://support.apple.com
744.
by Topfi 3y ago
Heyo, I have spent some of my off-time in recent weeks working on a small alternate reality game focused around Large Language Models and startup culture. Partly satirical, partly educational. I look forward to hearing your comments!
745.
Show HN: Ethical AI – Alternate Reality Game on LLMs(ethical-ai.eu)
5 points by Topfi 3y ago|4 comments
746.
by Topfi 3y ago
Correct me if I am wrong on this, but to me, it would seem that something like VirGL would still serve a purpose with wider spreader full SR-IOV support on consumer GPUs, as VirGL could find application in many scenarios where a GPU vendor&
747.
by Topfi 3y ago
I have been thinking that PoE is seriously underutilized for what it's capable off. There are so many use cases where one needs both networking and power, yet if you want a one cable solution, you are forced to contend with wireless wh
748.
I used AI in a video. There was backlash. – AustinMcConnell [video](youtube.com)
2 points by Topfi 3y ago|0 comments
749.
by Topfi 3y ago
I really believe it depends on the tool and use case. A Zettelkasten (digital or physical) does create some reliance, yes, but considering the improved output commonly associated, I'd be very surprised if a Zettelkasten tool prevented
750.
by Topfi 3y ago
This post did cover many of the same ideas I have been ruminating on concerning model weights and the nomenclature of current efforts. That's also why I generally tend to stick with calling these[0] "local/self hosted models&

More ›