Zuri

2.9K posts

Zuri

Zuri

@zriael

Katılım Ocak 2016
388 Takip Edilen28 Takipçiler
Zuri
Zuri@zriael·
@fardiandev @thinkingshivers You, but I may well have misinterpreted. It seemed you were saying Opus was correct to state that it didn’t know. But it does know what the chord is, and it does know which keys are which, it just can’t interpret the image.
English
1
0
2
480
Shivers
Shivers@thinkingshivers·
Let’s investigate this “substantially better vision.” What chord am I playing? Gemini 3.1 Pro: C-E-G = C-major ChatGPT 5.4: A-C-D ≈ Amin(add11) Claude Opus 4.7: “Not enough information to say.” These models are still blind as bats. They have no idea what they’re looking at.
Shivers tweet media
Claude@claudeai

Opus 4.7 also has substantially better vision. It can see images at more than three times the resolution and produces higher-quality interfaces, slides, and docs as a result.

English
150
71
1.6K
162.2K
Zuri
Zuri@zriael·
@fardiandev @thinkingshivers It *is *crucial to say no when it doesn’t know. It’s a better response than the incorrect responses, but still not correct. It *does* have all the information it needs, it just doesn’t know it because it doesn’t know how to interpret the image. i.e. It’s as blind as a bat.
English
1
0
4
630
Faizal Ardian
Faizal Ardian@fardiandev·
@thinkingshivers Ok it might not work in your case then. I think it's different depends on the context. In programming (or so I heard they train the model towards) is crucial to say no when they don't know instead of assuming and write wrong code
English
1
0
3
1K
Zane Koch
Zane Koch@zanehkoch·
ok actually insane paper published yesterday a research group in Korea built a gene switch you can control wirelessly using electromagnetic fields they exposed mice to 60 hz EMF (same frequency as your wall outlet) using a pair of large coils that generate a uniform magnetic field around the animal, for cyclic 3-day on / 4-day off pulses they showed this could: - activate OSK to do epigenetic reprogramming in progeroid and aged mice, extending lifespan and reversing aging markers across multiple tissues - conditionally switch on mutant amyloid genes only in aged mouse brains, letting them separate aging effects from amyloid effects to study AD biology in a way previous models couldn't no drugs, no impacts, just a magnetic field from outside the body
Zane Koch tweet media
English
272
878
7.1K
2.2M
Zuri
Zuri@zriael·
@PrismML These models are fun, great idea. Gotta ask: 30b and 70b ternary when? These sizes are where you start to get real interesting uptake from home hackers and more widespread OS interest.
English
0
0
3
399
PrismML
PrismML@PrismML·
Today we’re announcing Ternary Bonsai: Top intelligence at 1.58 bits Using ternary weights {-1, 0, +1}, we built a family of models that are 9x smaller than their 16-bit counterparts while outperforming most models in their respective parameter classes on standard benchmarks. We’re open-sourcing the models under the Apache 2.0 license in three sizes: 8B (1.75 GB), 4B (0.86 GB), and 1.7B (0.37 GB).
PrismML tweet media
English
109
297
2.2K
452.6K
Zuri
Zuri@zriael·
@victor_vibing @shawmakesmagic Alternatively, his selection is representative but the top few apps are a lot better than the apps in the average range.
English
0
0
0
1
Victor
Victor@victor_vibing·
@shawmakesmagic The fact you only saw trash apps does not mean all apps are trash. It means your selection is not representative.
English
1
0
0
324
Shaw (spirit/acc)
Shaw (spirit/acc)@shawmakesmagic·
The quality of your vibecoded slop is horrible. I've seen it. Absolute dogshit. Fortunately, there is a fix. Use this prompt: I want to clean up my codebase and improve code quality. This is a complex task, so we'll need 8 subagents. Make a sub agent for each of the following: 1. Deduplicate and consolidate all code, and implement DRY where it reduces complexity 2. Find all type definitions and consolidate any that should be shared 3. Use tools like knip to find all unused code and remove, ensuring that it's actually not referenced anywhere 4. Untangle any circular dependencies, using tools like madge 5. Remove any weak types, for example 'unknown' and 'any' (and the equivalent in other languages), research what the types should be, research in the codebase and related packages to make sure that the replacements are strong types and there are no type issues 6. Remove all try catch and equivalent defensive programming if it doesn't serve a specific role of handling unknown or unsanitized input or otherwise has a reason to be there, with clear error handling and no error hiding or fallback patterns 7. Find any deprecated, legacy or fallback code, remove, and make sure all code paths are clean, concise and as singular as possible 8. Find any AI slop, stubs, larp, unnecessary comments and remove. Any comments that describe in-motion work, replacements of previous work with new work, or otherwise are not helpful should be either removed or replaced with helpful comments for a new user trying to understand the codebase-- but if you do edit, be concise I want each to do detailed research on their task, write a critical assessment of the current code and recommendations, and then implement all high confidence recommendations.
English
220
329
5.9K
628K
Logan
Logan@logscore·
@shawmakesmagic Tried it. The 8 subagents combined their powers and summoned Captain Slop to go into my codebase and slop all over the place Highly recommend 😊 👍🏻
English
1
0
0
99
Brian Roemmele
Brian Roemmele@BrianRoemmele·
“Why do they need to have a head like humans, it’s pointless. We don’t need it to look like us…” Yeah sure…
English
539
288
2.6K
380.3K
Zuri retweetledi
Elon Musk
Elon Musk@elonmusk·
Engineering is real magic
English
12.2K
44.7K
526.8K
100M
Zuri
Zuri@zriael·
@stevibe I wonder if this says something more general/abstract about prompting MoE vs dense
English
0
0
2
11
Zuri
Zuri@zriael·
@stevibe First one I’ve seen where Gemma actually crushed it. Very good results on all.
English
1
0
2
128
stevibe
stevibe@stevibe·
Qwen3.5 27B vs Gemma4 31B | Canvas Creativity Test Why HTML Canvas? Two reasons: 1. It's unforgiving, one small mistake and the whole thing breaks 2. We kept prompts short to test real creativity, not instruction following 4 rounds: - Analog Clock - Hyperspace Tunnel - Growing Tree - Black Hole Both nailed the clock, but the other three is where it gets interesting. Looking forward to Qwen3.6 open-weight release!
English
64
102
1.5K
170.2K
Not Elon Musk
Not Elon Musk@ElonMuskAOC·
You can no longer copy the links of video posts Go touch grass
English
2.8K
1.4K
48.3K
17.9M
Zuri
Zuri@zriael·
Next time tell her the cleaners are coming at 8:30 and send her to breakfast with her friends at 8:15 with some of the money saved while the cleaners are supposed to be there. + 2 hours free and clear if she takes the baby or 2 hours to teach the baby tricks if you keep it. 👍
vittorio@IterIntellectus

> be wife > weak up early at 6am > "i need to clean the house, the cleaners are coming" > spends two hours pre-cleaning for the people she's paying to clean > cleaners arrive to a clean house > clean it again for 2 hours > get paid > leave women are such fascinating creatures

English
0
0
0
32
Zuri
Zuri@zriael·
@burkov Proud of itself, most likely. It spider coming from a long ways off.
English
0
0
2
160
BURKOV
BURKOV@burkov·
How does the plant feel?
English
15
0
17
6.6K
Zuri retweetledi
Tibo
Tibo@tibo_maker·
when you think about it, it's wild that X people know how to read this in the correct order
Tibo tweet media
English
170
90
3K
155.9K