lanthissa

5 ms·

31.
by lanthissa 4mo ago
yes? the future for any verifiable task is the model attempts to verify initial state and a goal then decomposes its tasks in to every smaller verifiable subtasks, with /memory being the persistence between runs and then /dreaming
32.
by lanthissa 4mo ago
i used to use opus for everything, thats not an option once you move to a multi agent system unless you're working on like high end research. I could easily spend 3k a day if i was using opus as just a normal dev. As we build a better
33.
by lanthissa 4mo ago
this is the finance team doing a fantastic job. keep in mind they're raising this cash right before 3 major ipos in their sector which people will need to raise money for and will fight against htem in the narrative. If i was a google
34.
by lanthissa 4mo ago
Just as a google shareholder, this company bought back shares hand over fist at a low p/e for a few years, issues 100 year debt at low rates, and is selling equity when its at a premium to its peers right before 2-3 major ipos of compe
35.
by lanthissa 4mo ago
flash 3.5 is the best price/performance model for what i'm doing. I had been using opus for everything but as we started running many agents at once, and then eventually agent managing sub agents frontier is not an option. we star
36.
by lanthissa 4mo ago
if you want to personally manage your risk you can by taking a small short position or buying long dated puts. It being in the public markets is something you can deal with if you want. It being in private markets means you cannot choose to
37.
by lanthissa 4mo ago
you cant have it both ways, the public can either have exposure and capture the upside or not. there are ways for you to manage your risk if it in public markets, theres nothing you can do if its in private.
38.
by lanthissa 4mo ago
my stance is coming from working at one of the soggiest banks and having access to 3 with 2 more coming, and knowing the same is true at 2 of our large competitors.
39.
by lanthissa 5mo ago
mistral was never competitive and is getting less so, but that doesn't matter they cant be allowed to fail and have a long time to find their lane. They're smart and have an audience of like 600m people and the largest governments
40.
by lanthissa 5mo ago
no it really not, even the soggiest bank has multiple api vendors atm.
41.
by lanthissa 5mo ago
switching models is insanely cheap compared to token cost on anything signficant, this is a take so cynical it misses the reality
42.
by lanthissa 5mo ago
shouldn't typed code that uses functional style be kinda the perfect end game for llms? You can parallelize generation at any granularity, easily ring fence changes, reproduce everything, types give clues to the llm.
43.
by lanthissa 5mo ago
yeah people have been talking about this for a few weeks. Spacex is the only unallocated nvidia gpu's on earth, anthropic is going through a crisis due to lack of compute. It would have been a huge black mark on the record of both if t
44.
by lanthissa 6mo ago
every tpu thats been made is in use and sold at a high margin, demand is not the issue.
45.
by lanthissa 6mo ago
there are literally not enough tpu's on earth for them to break out, every tpu thats been made is in use, the spike in demand is recent and google has heavy competition for foundry space.
46.
by lanthissa 6mo ago
Googles giving them something thats a lot more scares to them then dollars, large volumes of chips quickly. If you gave anthropic 10b cash they couldn't get chips in the 0-6mo timeframe at scale. Anthropic is suffering reputational dam
47.
by lanthissa 6mo ago
googles multiple businesses and gemini isn't the largest one. anthropic is the anchor external customer of tpu's and nvidia is worth more than all of google. If tpu's actually breakout as a viable alternative over the next fe
48.
by lanthissa 6mo ago
for all the drama, its pretty clear both openai, google, and anthropic have had to degrade some of their products because of a lack of supply. There's really no immediate solution to this other than letting the price float or limiting
49.
by lanthissa 6mo ago
not really, china has gone domestic for everything as soon as it could. its naive to think they would have stayed on a 'western' stack. Most of the time 'losing' isn't making a bad choice its being put in a situatio
50.
by lanthissa 6mo ago
meta has laid off 34,800 people in just the large scale rounds we know about in the past 5 years. they're growing at high teens % a year and have record profits and a centi-billionaire has complete control. whats going on there is gros
51.
by lanthissa 6mo ago
never ever forget theo's gpt 5 hype video and then him having to walk it back. its very clear that theres money or influence exchanging hands behind the scenes with certain content creators, the information, and openai.
52.
by lanthissa 6mo ago
in new york you're not remotely right. the suburbs around new york are some of the richest in the world. Scardsale, every town near the ct border, rye, huge parts of li, montclair nj and the towns around it. the average household net w
53.
by lanthissa 6mo ago
this isn't accurate. america has a wealth per adult of 551,350 germany has a wealth per adult of 256,180 if you exclude the top 10 highest wealth holders in each country its 543,385 vs 252,811. america's a rich country compared m
54.
by lanthissa 6mo ago
i dont think you need to make it malicious, this is what not having enough compute looks like in practice. its true of gpt5, its true of why antigravity rate limits, and its true of claude. there is no possible action thats not user hostile
55.
by lanthissa 6mo ago
people pay them to use it, they find that positive
56.
by lanthissa 6mo ago
theres a lot of things to critique about the us, but the f35 isn't one of them. Over the past few years we have seen it operate with impunity over multiple countries. It astounding to me that in the 12 day war and the iran conflict the
57.
by lanthissa 6mo ago
its not though if you're working in a massive codebase or on a distributed system that has many interconnected parts. skills that teach the agent how to pipe data, build requests, trace them through a system and datasources, then updat
58.
by lanthissa 7mo ago
theres a famous painting about this https://en.wikipedia.org/wiki/Saturn_Devouring_His_Son
59.
by lanthissa 7mo ago
in a competitive market they would have been unified a long time ago. google has been making slow steps at doing this apple wont until google does
60.
by lanthissa 8mo ago
for cerbras, can we call them chips? you're no longer breaking the wafer we should call them slabs

More ›