hailey @hailey.at · 30/09/2026sorry boss but the weather outside is just way too nice to not be drinking a claw while on the clock 400
hailey @hailey.at · 30/09/2026you dont have to like the website, but you all are just as bad about spreading misinformation as you claim the opponent to be and its gross. screenshot taken just now.@angrystaffer.bsky.social@angrystaffer.bsky.socialI think we all saw that coming 000
hailey @hailey.at · 29/09/2026more interesting imo is that jev is far more agressive in its responses than all the LLMs! most of the LLMs land extremely centrist 000
hailey @hailey.at · 29/09/2026sure why not. Jev takes the 12 axes quiz@minormobius.bsky.social@minormobius.bsky.socialJev 12axis political test 200
hailey @hailey.at · 29/09/2026the initial run did _not_ include image context at all, so these numbers are going to change drastically. however, particularly on high-confidence jev decisions agreement is very high. disagreement comes generally in the form of posts with low context, where modernbert actually performs better 200
hailey @hailey.at · 29/09/2026should have something exciting soon, doing some more reworking of the pipeline. however these costs are super exciting! have said this before about jev, but i see a lot of potential here. used 445M input tokens for ~180k posts, which cost about $18.70hailey@hailey.atjev distillation experiments are proving quite interesting 200
hailey @hailey.at · 29/09/2026ed conveniently forgetting this part. lol.Ed Zitron@edzitron.comReuters got some of the IPO docs. Anthropic is a total dog of a company. Spent $12.6bn to make $4.6bn in revenue in 2025, $7.33bn of which was compute costs. Operating loss of $8bn. Losses getting worse year over year. Amazing stuff www.reuters.com/business/fin... 200
hailey @hailey.at · 29/09/2026this cat also loves to do weird things but a different type of weird 000
hailey @hailey.at · 25/09/2026lol.@timfernholz.com@timfernholz.com🚨OpenAI friday news dump 🚨 techcrunch.com/2026/09/25/u...techcrunch.comUnsecured OpenAI agents posted 53 user images on the internet without the lab's knowledge | TechCrunchAI agents operating in OpenAI's research environment posted user images on public image-hosting sites without the lab's knowledge. 200
hailey @hailey.at · 25/09/2026no. white monster is garbage. please drink the blue logo can@sickoscommittee.org@sickoscommittee.orgBetter now? 000
hailey @hailey.at · 22/09/2026lol, and opus 5.5 is still not priced at a point that makes it worth using. incredible 400
hailey @hailey.at · 19/09/2026discourse overhailey@hailey.atone of the stranger things i’ve seen coming from the tilly norwood discourse is the insistence you must use “it/its” for her lol. i’m sure this won’t lead to problems. 000
hailey @hailey.at · 18/09/2026i’m sure this will work out well and will not be implemented in bad faith www.gov.ca.gov/wp-content/u... 000
hailey @hailey.at · 18/09/2026exhibit 205 of people making using AI something that necessitates revoking your trans card (and in this instance also killing you!!) normal website with normal posters 200
hailey @hailey.at · 17/09/2026latency is where things are really great though. particularly when batching, we can achieve _significantly_ lower latency, even with the larger prompts 200
hailey @hailey.at · 17/09/2026the “gate” results here take uncertain results from the zero shots then make another request nthat is substantially more input tokens but includes examples of what we consider for each label obviously the gate is more expensive, but it is still cheaper than frontier 200
hailey @hailey.at · 17/09/2026Here are some results from @typesafeai.bsky.social’s Jev on a common task we have, comparing results with Jev to top frontier models. with zero shots we can achieve nearly frontier levels of accuracy, but if we do a bit extra work we can beat frontier accuracy 200
hailey @hailey.at · 17/09/2026damn going to get back home and immediately be stuck playing this game huh 000
hailey @hailey.at · 17/09/2026btwhailey@hailey.atoh also. cost is nuts. the performance i’m getting is comparable AND is costing significantly less. this is a massive fucking game changer and should not be slept on 000