@teortaxesTex Do you have any skills/rules/memories? It uses different files than the other agents so it wouldn’t inherit the same configuration every other agent does and could also have it’s own old configuration making it act weird if you’ve ever used Claude before
@AdamHoltererer Looking at the quoted thread you’re taking this too far.
Can you imagine how demoralizing it would be if you spent months handcrafting something then the moment you share your progress someone else can replicate it in 5 minutes?
If you can’t admit you made a mistake people have
@AdamHoltererer Looking at the quoted thread you’re taking this too far.
Can you imagine how demoralizing it would be if you spent months handcrafting something then the moment you share your progress someone else can replicate it in 5 minutes?
If you can’t admit you made a mistake people have
Looking at the quoted thread you’re taking this too far.
Can you imagine how demoralizing it would be if you spent months handcrafting something then the moment you share your progress someone else can replicate it in 5 minutes?
If you can’t admit you made a mistake people have the right to block you, to avoid this happening to them.
I’m all for AI accelerating what indie game devs can make, but stealing people’s ideas to test this isn’t the way.
You should come up with an original idea you can mock up and have the AI build if you actually care about indie game development
@qldTowers@synthwavedd I agree but error margins would be useful context, I still don’t know what percentage change would be statistically significant for this specific benchmark
@jimmythej1@felpix_ Curious, what evidence have you seen that makes the minimum goal AI has to reach to not be regurgitating training data much harder to reach than it was a few years ago?
@TankP338517@theo I agree, but it’s also helpful to have specific ways to measure these qualities
I believe when Theo says smart he’s referring to a combination of intelligence and knowledge, and when he says dumb he means comprehension
It makes more sense to me if these are specifically named
@mattpocockuk This is NOT something Jev should be used for, maybe Luna could lint the ENTIRE changed diff w/context from read files but 32K isn’t enough for this use case
I like to think of intelligence, knowledge, taste, and comprehension as 4 axes.
For example, Astra has lots of knowledge and intelligence but mid taste and bad comprehension.
Smart is a combination of all 4 traits.
With LLMs, we like to think of "smart" and "dumb" as one axis (because we think of humans this way).
I'd like to argue against this framing. Instead, try to think of "smart" and "dumb" as two different axes. A model can be incredibly smart AND dumb at the same time.
For a good
@NavDoesTech@theo Good points, I think this could be useful for public facing agents/chatbots within apps or bulk data processing with lots of short data points of varying levels of complexity, not necessarily for personal coding/frontier agent work
@NavDoesTech@theo Sorry if I came off that way, I like Theo and agree with almost all his takes, just disagreed that Jev here gets no context outside of the prompt and believe in some cases it can be useful for model routing
I get where you’re coming from though, there’s so many anti-Theo trolls
Good points, I think that the system prompt, tool definitions, and initial prompt don’t typically exceed 32k but in any cases where they do, more context than that is needed, or images are involved this wouldn’t be a good way to use Jev
I 100% agree Jev isn’t perfect here but this isn’t a “bad idea”, especially for non-coding tasks where it most likely has the context needed to make a smart decision
1K Followers 220 FollowingI am a software engineer crafting Mooligan - a deck-building and portfolio tracking app for TCG enthusiasts.
MTG, mainly modern.
413 Followers 289 Following18 • SWE • browser use guy
intern @asciidotdev (YC F26)
https://t.co/Bq4HsCOTRw
wanna have a call with me?
https://t.co/pZCC6GRGDp
5K Followers 101 Following20yr+ programmer, sharing on youtube, love talking AI, Host on Rate Limited podcast, Host on Automated Brand Podcast and Building https://t.co/mEanyLBqsA
230K Followers 224 FollowingWhere AI meets the real world. We measure and advance the frontier of AI through community-driven evaluation. We’re hiring → https://t.co/XBZCrsdD77
24K Followers 1 FollowingPrivacy-first browser with unbiased ad-blocking and a clean, minimal interface. Browse the web without noise.
Fully open source, by @imputnet
77K Followers 3K FollowingWe're in a race. It's not USA vs China but humans and AGIs vs ape power centralization.
@deepseek_ai stan #1, 2023–Deep Time
«C’est la guerre.» ®1
186K Followers 5 FollowingMakers of Devin, the first AI software engineer. We are an applied AI lab building end-to-end software agents. Join us: https://t.co/4Ss9hvpjRG
1.8M Followers 2 FollowingWe're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant @claudeai on https://t.co/FhDI3KQh0n.