Chase

6.3K posts

Chase banner
Chase

Chase

@Chasewhip

programming money @sphere by @sphere_labs | formerly @teleportxyz @seagullfi @jungledefi | views are my own, nfa

The Cloud Katılım Temmuz 2014
1.2K Takip Edilen859 Takipçiler
Chase
Chase@Chasewhip·
@tmuxvim You have to make sure they are writing correct tests and not bullshit though, once you get a good harness in place it makes life easy.
English
0
0
0
25
tmuxvim
tmuxvim@tmuxvim·
maybe my favorite part of clankers is how enthusiastically they will write unit tests for me as they fix and implement things. I feel more powerful every day with my massive test suite. more confident in shipping unreviewed code. it's amazing.
English
4
1
22
857
Chase
Chase@Chasewhip·
@HenronEldini I noticed it change on Thursday. The fact it doesn't take 30m for a task makes me trust it less lmfao.
English
0
0
1
8
HenryE
HenryE@HenronEldini·
The new GPT Pro model is around 3x to 4x faster than regular GPT Pro, which is great but it really needs to hit more like 30x speed up for normies on the regular plan to start feeling the benefit of how much better Gpt Pro is in terms of reducing hallucinations & instr following
English
1
0
0
87
Chase
Chase@Chasewhip·
@0xrwu Non flaky playwright tests? no such thing exists
English
0
0
1
34
Richard Wu
Richard Wu@0xrwu·
What's the best prompt/hack you use to fix flakey playwright tests?
English
3
0
1
856
tracy 💻 🦝
tracy 💻 🦝@tracy_codes·
if you're a Solana protocol and you explicitly don't support off-curve addresses, why?
English
2
0
10
1.5K
Chase
Chase@Chasewhip·
@tmuxvim @10x_er This feels like a training artifact where they might be able to get better average performance learning this end to end.
English
0
0
0
31
tmuxvim
tmuxvim@tmuxvim·
@10x_er high reasoning can actually make models perform worse at certain tasks.
English
1
1
16
951
10x Engineer
10x Engineer@10x_er·
This stuff is literally not real. I give this feature 6-12 months tops before they remove it, bookmark this and confim. Like what do you mean you want me to pick how hard my model wants to try? I want it to try its hardest, always, and I want it to be fast. Why would I ever need to consider tradeoffs when doing knowledge work with AI.
10x Engineer tweet media10x Engineer tweet media
English
298
19
1.6K
344.6K
Chase
Chase@Chasewhip·
@cavemanloverboy If hardware stays the same for running future models then that is definitely where you start investing in some tinybox greens
English
0
0
0
60
Mike MacCana
Mike MacCana@mikemaccana·
SOLANA PEOPLE: durable nonces are a terrible idea, absolutely like a blank cheque and we should get rid of them. Agreed. 👍 But: how are people handling recurring payments without them? @sphere_labs? Other knowledgable people? Thanks.🤔
English
14
2
39
5.7K
Jarred Sumner
Jarred Sumner@jarredsumner·
@thdxr we fixed a lot of stuff. we are also running a JavaScript engine fuzzer on bun APIs + many node APIs. When bugs are found, it automatically spawns claude to fix.
English
9
0
360
23.6K
dax
dax@thdxr·
bun related segfaults have decreased quite a bit
English
9
1
360
98.1K
NATE BAG 🎮
NATE BAG 🎮@NatebagCEO·
My dude said it’s gonna take me YEARS to decompile a GameCube game? Buddy you have 4 days, get to fucking work
NATE BAG 🎮 tweet media
English
7
0
47
3.6K
Chase
Chase@Chasewhip·
AI writes good enough Nix configurations now, no excuses not to use it.
English
0
0
3
70
Chase
Chase@Chasewhip·
@NADEAU__ Emails: never click the link, go to the official site directly Phone Calls: always hang up and call the official number directly and never trust the voice, voice models are too good now
English
0
0
0
16
S◎Lcruiter 🧊
S◎Lcruiter 🧊@NADEAU__·
I needed to go to a crypto ATM and deposit 5k and it would give me a QR code for the sheriff to scan. I then knew what was going on and called them out and went and actually reported it to the sheriff and when I told them they said its been happening non stop for a week.
English
1
0
5
336
S◎Lcruiter 🧊
S◎Lcruiter 🧊@NADEAU__·
Be CAREFUL out there, just got an elaborate attempt at a scam. Phone call said it was sheriffs from my county that I missed Grand Jury duty (never got a notice) told me to come into the office to to signature verification and will need 5k to post bail, on the way there they said
English
6
2
18
1.3K
neira
neira@borjaneira_·
Reading X, I'm realizing something: It's sad that posting on X is becoming a prompt for 20 people to reply with a copied and pasted response from an LLM model. Is this the new prompt engineering?
English
6
0
20
1.3K
Chase retweetledi
Noah 🎈
Noah 🎈@redacted_noah·
In 10 years it's going to be obvious who used AI to turn their brain off, and who used it to challenge themselves and maximize their learning. Tools aren't inherently good or evil, it's how you use them that matters. Don't let AI brainrot you.
English
15
3
40
1.4K
Chase
Chase@Chasewhip·
@alessandrod o3 was a beast, but maybe I’m just remembering it highly. The newer o3 revision you can still access changed the thinking level (or something) and was never the same.
English
0
0
0
49
munch
munch@munchPRMR·
I just tried to offramp 25k from Coinbase and they tried to charge me $500
English
119
32
858
160.5K
Chase
Chase@Chasewhip·
@SolPlay_jonas @redacted_noah Not even lying I forgot your name this year but know what you look like and it’s probably mutual 😂
English
1
0
2
20
Jonas Hahn
Jonas Hahn@SolPlay_jonas·
@redacted_noah Haha same :D After a few words i remember again what we talked about last time. For some my only way to talk to them again is wait for the next solana event because many tg messages are just „hi“ and the handle some gibberish. Its so fun!
English
1
0
2
133
Noah 🎈
Noah 🎈@redacted_noah·
There are dozens people I see at every breakpoint/solana event, have great conversations with, and have no idea their names. I probably learned their name 5 breakpoints ago. I consider these people friends, and will continue to refer to them as "hey, you..."
English
13
0
54
2.9K
Noah 🎈
Noah 🎈@redacted_noah·
Can someone IBRL my sinuses? Breakpoint plague strikes again, I've got more congestion than Eth
English
13
0
31
2.8K