Topfi

21 ms·

361.
by Topfi 7mo ago
In fairness, Office is as generic a term as one can come up with for such a software suite. On top of that, I wouldn't be surprised if that fell under Genericide like Lego or Google and, lest we forget, the Microsoft Office brand does
362.
Coffeezilla – Investigating AI Deepfakes [video](youtube.com)
2 points by Topfi 7mo ago|0 comments
363.
Hunter Alpha – Stealth Model on OpenRouter(openrouter.ai)
4 points by Topfi 7mo ago|0 comments
364.
OpenUI – The Open Standard for Generative UI(github.com)
2 points by Topfi 7mo ago|0 comments
365.
by Topfi 7mo ago
> "The Moltbook team has given agents a way to verify their identity and connect with one another on their human's behalf," Shah says. "This establishes a registry where agents are verified and tethered to human owner
366.
Full Interview: Anthropic CEO Dario Amodei on Pentagon Feud [video](youtube.com)
27 points by Topfi 8mo ago|2 comments
367.
Krazam Presents: Paradise (Trailer) [video](youtube.com)
4 points by Topfi 8mo ago|0 comments
368.
by Topfi 8mo ago
Have been following your models and semi-regularly ran them through evals since early summer. With the existing Coder and Mercury models, I always found that the trade-offs were not worth it, especially as providers with custom inference ha
369.
Confluence Labs(confluence.sh)
1 points by Topfi 8mo ago|0 comments
370.
by Topfi 8mo ago
Minor version bumps are good and I want model providers to communicate changes. The issue I am having is that Gemini "preview" class models have different deprecation timelines and rate limits, making them impossible to rely on fo
371.
by Topfi 8mo ago
Should have clarified initially what I meant by stable, especially because it isn't that known how these terms are defined for Gemini models. Not talking about getting consistent output from a not-deterministic model, but stable from a
372.
by Topfi 8mo ago
My issue is that we haven't even gotten the release version of 3.0, that is also still in Preview, so may stick with 3.0 till that has been deemed stable. Basically, what does the word "Preview" mean, if newer releases happen
373.
by Topfi 8mo ago
Appears the only difference to 3.0 Pro Preview is Medium reasoning. Model naming has long gone from even trying to make sense, but considering 3.0 is still in preview itself, increasing the number for such a minor change is not a move in th
374.
by Topfi 8mo ago
In my evals, I was able to rather reliably reproduce an increase in output token amount of roughly 15-45% compared to 4.5, but in large part this was limited to task inference and task evaluation benchmarks. These are made up of prompts tha
375.
by Topfi 8mo ago
Unless I am mistaken, that is all plain old markdown, arguably the easiest to migrate format for such data there can possible be. Heck, that was half the pitch behind Obsidian, even if the project someday ended, markdown would remain. And s
376.
by Topfi 8mo ago
Sober, fair analysis, covers what a large contingent on here was commenting on the original release. Like the concise summary he did, though as is so often the case with claims on LLM capability that are beyond what is real, this won't
377.
Backlash over decision to retire GPT-4o shows dangers of AI companions(techcrunch.com)
3 points by Topfi 8mo ago|0 comments
378.
I Thought Game Publishers Cared About Protecting Their IP? – Unpacked [video](youtube.com)
1 points by Topfi 8mo ago|0 comments
379.
by Topfi 8mo ago
I understand the confusion, this was done by Anthropics internal Red team as part of model testing prior to release.
380.
by Topfi 8mo ago
The official release by Anthropic is very light on concrete information [0], only contains a select and very brief number of examples and lacks history, context, etc. making it very hard to gleam any reliably information from this. I hope t
381.
by Topfi 8mo ago
I am by no means an expert, but as far as I am aware FATCA violations carry slightly higher penalties then what you suggest and are very much not limited to "you can't do business with the US government anymore". Also, even i
382.
by Topfi 8mo ago
But neither of your described scenarios applies to either of the two. Both FATCA and GDPR apply to entities/companies that deal with citizens from their respective jurisdiction. FATCA applies e.g. to foreign banks handling US customers
383.
by Topfi 8mo ago
> Unless they signed a treaty agreeing to abide by it, their they're own sovereign entities and their businesses don't have to comply with remote EU laws. What is your opinion concerning laws such as FATCA and other such laws t
384.
Update on Irish GDPR Decision about TikToks Transfers of EEA User Data to China(tiktok.com)
2 points by Topfi 8mo ago|0 comments
385.
by Topfi 9mo ago
I have yet to experience any degradation in coding tasks I use to evaluate Opus 4.5, but I did see a rather strange and reproducible worsening in prompt adherence as part of none coding tasks since the third week of January. Very simple que
386.
by Topfi 9mo ago
If you are referencing Nvidia, note that Vertex, Groq, Cerebras, etc. all do not rely on GPUs for much or all of their inference and are all the better for it. Kind of why Nvidia see themselves forced to make such deals, to lock labs in bef
387.
by Topfi 9mo ago
I don’t really understand your point. A company offering only inference will inherently always have lower costs and require less hardware than one that both provides inference and does training of new models.
388.
by Topfi 9mo ago
Yes, it's in the repo.
389.
by Topfi 9mo ago
There are many providers (Fireworks, Groq, Cerebras, Google Vertex), some using rather common hardware from Nvidia, etc., others their own solutions focused solely on high throughput inference. They often tend to be faster, cheaper and&#x2F
390.
by Topfi 9mo ago
I'm still testing myself and cannot make a confident statement yet, but Artifical Analysis is a solid and independent, though also to be fair somewhat imperfect source for a general overview: https://artificialanalysis.ai&#x

More ›