lain

it's me from [email protected], but here i talk about AI

Public Key

npub1v57dpglku0evttjz8r8vdn3j7jwx4f4lztdaucj8tzt0urq3qleq5yx7mc

NIP-05 Address

Profile Code

nprofile1qqsx20xs50mw8uk94epr3nkxece0f8r256l39k77vfr439h7psgs0uspz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsrpfsgw

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1v57dpglku0evttjz8r8vdn3j7jwx4f4lztdaucj8tzt0urq3qleq5yx7mc

Show more details

Last Notes

2026-06-12 11:40:00 GMT

fable is really a whole class better than even gpt 5.5. it's probably the best coworker you've ever worked with, very good at getting stuff done, and generally extremely smart and capable. it's getting very, very close to my 'personal software' vision. it can one-shot most medium-complexity tasks that a user would want to deal with.

2026-06-10 08:34:32 GMT

- reply

@nprofile…tsag it was released yesterday, maybe you need to update your claude code

2026-06-10 08:33:13 GMT

- reply

@nprofile…tsag i just mean fable, fable is just mythos with some extra guards

2026-06-10 08:16:37 GMT

so far, it's the real deal. very autonomous. can handle complex tasks succinctly and is actually really fast. looks to me like it really is the best model that's out right now. (sam altman, your turn)
https://fediffusion.art/media/98/e4/79/98e47916cafbf74d721ebe1489c3d5a0f134a33a59ca5a6b20fea0fa7b809dbc.webp

2026-06-08 19:23:40 GMT

https://fediffusion.art/media/e5/d5/71/e5d5716dca64f2eec39db3842418b36370cae468987d781e00ce3ba17da71afa.webp

2026-06-06 10:43:37 GMT

still in awe how bad llama.cpp / ollama / vllm / lm studio / literally everything is. how can there still be a market for "good inference software that actually works"? who's gonna be the first?

2026-06-06 08:33:04 GMT

pretty impressed with minimax m3. it seems better than gpt 5.5 at UI/UX work, and the basic programming / debugging capabilities are also very good. seems better than glm-5.1 to me, at least. context windows is a huge 1 million, and it's multimodal, so you can tell it to take a look at a website it wrote and fix the visual bugs.

2026-06-05 09:27:15 GMT

one-shot result with MiniMax M3 free on opencode.
full prompt: "Can you write a 3d solar system animation / simulation with three js? the planets should look as realistic as possible, so try to find some public textures for them online."
https://fediffusion.art/media/1b/f2/1a/1bf21aeb3b322659ffc36c9ce3b1509ea751a1e1bf31e2bcd9e7b3a316a90e4e.mp4

2026-06-02 19:05:59 GMT

BEAUTIFUL debugging glitch art
https://fediffusion.art/media/13/9a/ea/139aea7ef95b8945c1b40532b268d3b5e40011603fd5a13a6ed62c50196b3a37.webp

2026-05-29 18:54:40 GMT

- reply

@nprofile…dugj just relax then

2026-05-29 18:27:02 GMT

- reply

@nprofile…dugj bug discovered

2026-05-29 18:11:11 GMT

you can now run my very own disk operating system in your browser, and it will even tell you how it works.
https://lambadalambda.github.io/laindos/
code is at https://github.com/lambadalambda/laindos
https://fediffusion.art/media/88/f7/bc/88f7bcf08b427ac851ccc85e034858540fe5edf07a22110a4b3ccbbbc7eb46e4.webp

2026-05-28 08:02:40 GMT

deleting your users' data because you don't like the software they are using, amazing
#nevent1q…0hww

2026-05-27 13:37:02 GMT

works best in 86box, cc @nprofile…s2sp
https://fediffusion.art/media/65/78/9e/65789ef98919409fb04c53a32b91518d40b5b4fc6c93b1d412ea6c9476e17849.mp4

2026-05-19 09:13:12 GMT

https://fediffusion.art/media/b8/a4/91/b8a491de1303b6957ddfbdac6523efe2871823c63fcca99de22080b26654a5b3.webp

2026-05-03 09:17:18 GMT

Thom working hard to lose the few remaining readers
https://fediffusion.art/media/3d/de/95/3dde953bfe2bd5b5568f9cabb822235a77472321703dc2e034712228d3bdcd6b.jpg

2026-04-30 10:02:45 GMT

> Bun operates its own fork of Zig, and recently achieved a 4x performance improvement on Bun compile after adding "parallel semantic analysis and multiple codegen units to the llvm backend". Here's that code. But @bunjavascript says:
> We do not currently plan to upstream this, as Zig has a strict ban on LLM-authored contributions.
https://simonwillison.net/2026/Apr/30/zig-anti-ai/

2026-04-22 06:00:02 GMT

it looks up logos by itself, and pulls data from the web.
https://fediffusion.art/media/1a/43/f6/1a43f6e4e3c5d7226d88d9838aaf51ef440be767358c2ec4d4f1966e3201305e.webp
https://fediffusion.art/media/dd/ce/2a/ddce2abed06bf86a52f9beb2bd83b3b46126f188182c94f2410574337868b8dc.webp

2026-04-21 12:41:42 GMT

- reply

@nprofile…xsag trying it out right now, so far very impressive. absolutely amazing.

2026-04-21 12:38:32 GMT

this and the 'pro' version, zen, are probably the easiest way to try out good models that actually work well https://opencode.ai/go

2026-04-16 13:17:12 GMT

https://this-abg-does-not-exist.com/

2026-04-12 04:56:58 GMT

https://github.com/MiniMax-AI/OpenRoom
https://fediffusion.art/media/ac/50/67/ac5067c354746172684a85c188d6f9b0cf92b8629591f9e0d42474f89aa8dee2.mp4

2026-04-11 14:23:13 GMT

z.ai raised their prices by like 200%. happy to have a legacy account... i guess they realized they are actually have an amazing offering and can charge for it.

2026-04-11 09:28:27 GMT

https://fediffusion.art/media/01/7c/2f/017c2f10bb1ac8bc386b69c2ba62f2d08b1d4030abeafbf7443dc8a5c28cfce0.webp

2026-04-10 10:34:51 GMT

gemma 4 26B has good enough vision to figure out bounding boxes and positions, probably good enough for mouse control.
https://fediffusion.art/media/65/52/4e/65524ec238a481a6aa774b2cda427398a407046eac2cc73f15104e9ffcbb23e4.webp
https://fediffusion.art/media/b6/21/f1/b621f1b983daa403145e6775159994b70cac390fd787ea4bbbddd85a4ff8b226.webp

2026-03-03 15:37:42 GMT

- reply

@nprofile…p2qa no, text2img
https://fediffusion.art/media/f6/e8/43/f6e8435cdc45bd67514977cb148eba7343e3a876f245ce5619a0dc373fc6f6ad.webp

2026-03-03 15:12:12 GMT

https://fediffusion.art/media/a2/04/ef/a204ef9a4ef888875bbf3274525d9ad51dcb8ef3ea0fec40bb67a0851dd1b0a9.webp

2026-03-01 16:06:44 GMT

i think it's generally useful to completely ignore any talk about "AI" unless you have a really good reason to trust the source. an hour of playing with a frontier model will tell you a lot more than reading 30 editorials about how LLMs are / are not AI and will / will not destroy all jobs and will / will not lead to incredible wealth for everyone.

2026-02-27 09:53:58 GMT

happy friday
https://fediffusion.art/media/eb/89/9b/eb899b6ec87fd5cafbb7103233ecb7cab705abe79b8842b0841650ef320c2b50.webp

2026-02-17 05:28:58 GMT

the *claw type bots usually come with some 'skill store', but I don't understand why anyone would use those. to sensibly run the bots, you need a really smart model (opus, chatgpt, GLM-5, Kimi), and those can already just write whatever you need.

2026-02-09 17:15:56 GMT

https://fediffusion.art/media/79/fe/df/79fedf50129bdf4185e59d891fa258c11b4525c784bed9330c304bb8f9046f0c.webp

2026-02-07 20:24:41 GMT

- reply

@nprofile…p2qa that's alright

2026-02-07 19:54:24 GMT

- reply

@nprofile…p2qa no, i think both claude and codex are at the level where they are no more dangerous than a human coworker.

2026-02-07 19:27:47 GMT

- reply

@nprofile…p2qa yes, like every coworker.

2026-02-07 15:09:46 GMT

mini review
claude 4.6: it's alright
codex 5.3: nearly as good as 5.2 xhigh but MUCH faster. can write complex software in a few hours without much input. can do complex sysadmin tasks, too.

2026-02-04 11:00:28 GMT

https://fediffusion.art/media/82/a7/71/82a7710281ec7361b942b3556b2c20665f49b6e78f1af0e179fba2909e3c350f.webp

2026-02-02 17:48:52 GMT

Author of 3 ActivityPub servers

2026-02-01 13:33:39 GMT

very minimal single user fediverse server written in zig with TDD in ~24 hours, by ChatGPT 5.2 xhigh.
https://github.com/lambadalambda/feddyspice

2026-01-31 05:36:01 GMT

Got the small x4, esp32 based ereader with a an open source firmware. Adding features to it with codex works pretty well, first thing: embedded image support.
https://fediffusion.art/media/64/a1/66/64a1660bed7bb60b6f4303a58d40b16a49d709ebeb665053c0f7a9b4ce192856.jpg

2026-01-28 20:47:05 GMT

as soon as we found a way to have systems that could research, argue, talk and write poetry, but can't multiply two large numbers, we should have looked inward much more seriously.

2026-01-27 16:30:35 GMT

- reply

@nprofile…p2qa delulu

2026-01-27 10:18:35 GMT

Minecraftbros…
https://fediffusion.art/media/f5/a8/1d/f5a81d98196aec7ac51b1b276776ffffa0ed5f4d4d2aea6631fe149098e396c5.jpg

2026-01-20 18:30:04 GMT

it is suspicious to be living right at the time when we immanentize the eschaton

2026-01-16 17:42:25 GMT

> How do I feel, about all the code I wrote that was ingested by LLMs? I feel great to be part of that, because I see this as a continuation of what I tried to do all my life: democratizing code, systems, knowledge. LLMs are going to help us to write better software, faster, and will allow small teams to have a chance to compete with bigger companies. The same thing open source software did in the 90s.
https://antirez.com/news/158

2026-01-15 11:28:28 GMT

- reply

@nprofile…p2qa i thought that was obvious but you're right, most people probably have no idea how any of this actually works

2026-01-15 11:03:28 GMT

A video completely AI generated, with open source models, on a machine with 16GB VRAM.
https://www.reddit.com/r/StableDiffusion/comments/1qae922/ltx2_i2v_isnt_perfect_but_its_still_awesome_my/
https://fediffusion.art/media/0e/cf/50/0ecf5007d2de1f656af76a94a8c03b52e5c7d6c5988ed01c9252375957d442b6.mp4

2026-01-12 20:16:39 GMT

remember ai fingers?

2026-01-11 16:45:08 GMT

Very well made video about the somewhat mysterious phenomenon of 'grokking' in transformers, where they will quickly memorize all training data in a few hundred steps but still completely fail on testing data, then seemingly not improve at all for thousands of steps before suddenly generalizing perfectly: https://www.youtube.com/watch?v=D8GOeCFFby4

2026-01-09 16:50:40 GMT

z image generations trained on star trek TNG.
https://fediffusion.art/media/5f/d3/47/5fd347d116d0b88ab7b7eef137cc0ac2897c7888fa7f033154025eede9827071.webp
https://fediffusion.art/media/ea/7b/32/ea7b32a5514998cde4752ec8e300de0421a02cdc2b7457203baf4c26fb8180c6.webp
https://fediffusion.art/media/51/c3/9c/51c39cb5156fabdfa33e5d8b4ee72effc5590db2d68ea775f85c3ea6e50c5428.webp
https://fediffusion.art/media/ec/e2/55/ece25564c9f1dd772289e3fbea74f618a7fe5001e026eaac90fc402a93915856.webp

2026-01-04 07:31:54 GMT

In my experience, Claude is much better than ChatGPT at creating good frontend designs. ChatGPT can implement them well, but for a new design without a reference picture, better ask Claude.

2025-12-30 16:54:12 GMT

new hobby: taking an open core project and asking chatgpt to implement the missing proprietary functionality

2025-12-27 13:25:51 GMT

─ Worked for 18h 26m 37s ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
• Done

2025-12-25 14:43:54 GMT

the civitai experience
https://fediffusion.art/media/a44a8f5531c6be027dd1b75221c277f3d795ffb55993f51b16ed65fd2f9b3c79.png

2025-12-23 13:19:38 GMT

using coding agents to decompile nintendo 64 games
https://blog.chrislewis.au/using-coding-agents-to-decompile-nintendo-64-games/
https://fediffusion.art/media/93bf98e73b6d94c87df8533eeeae272a94e39724024a000241e8be3a9e25a9d0.png

2025-12-21 21:25:05 GMT

- reply

@nprofile…jmdl sure, but that's your employer's problem. maybe their code is legitimately so secret that it can't touch anything else, that's alright. but if it isn't, keeping it away from LLMs will make it less secure and less efficient, which can't help in the long run.

2025-12-21 21:14:56 GMT

it's getting to the point where it's unconscionable to not use LLM assistants for programming

2025-12-19 07:08:11 GMT

https://simonwillison.net/2025/Dec/15/porting-justhtml/
A report of porting a library from python to javascript with codex, the chatgpt coding agent. It took ~5 hours and only needed one user intervention, when the prompt credits ran out.
I can confirm that GPT-5.2 has become MUCH better at autonomous work. I had it work on a hard problem that required it to constantly take screenshots and view the results, on an existing codebase, for over 8 hours without intervention.

2025-12-19 06:43:29 GMT

https://github.com/DGoettlich/history-llms
cool idea, training LLMs only on data that's older than 100 years

2025-12-18 07:33:23 GMT

Meta has released SAM (Segment Anything) for audio. This essentially means that you can take an audio source and split it up into the different audio sources it's made up from. Want to do a fan dub? remove all the voices but keep the sound effects and music intact. Want to stream a game with copyrighted music? Just remove it while keeping the rest.
https://ai.meta.com/samaudio/

2025-12-15 21:10:54 GMT

why is richard spencer so gay coded

2025-12-15 16:29:31 GMT

alright, finally cracked it.
https://fediffusion.art/media/a3242358cd66c1952163722190e14ae87bfaf51a91f2842955a0109cb443b1c4.png
https://fediffusion.art/media/f26e3bbbda502ccdc50e5e206a2cf4da9a88c8df4ae96a1677434cbd0a0b303f.png
https://fediffusion.art/media/250582c0f9ee5fdcf7847dc67280ba91da6c7634ce8dadd6b08cecb5ccf3aaa2.png
https://fediffusion.art/media/1c3bcc90ec0a3c1fdad407748bea99f082f7ef803a216360a12e163b2988a15f.png

2025-12-15 10:02:21 GMT

the special fx people at corridor crew make a toy story short film with comfyui and wan. these are fully open source and run on local hardware. generative AI does enable completely new workflows.
https://www.youtube.com/watch?v=DSRrSO7QhXY
https://fediffusion.art/media/8e922e2558a1c51554bb27580e6ae15ce604bb80b545d309c6ddeabe82ae62bd.png

2025-12-14 20:52:52 GMT

no idea why but non is really hard to train for. maybe there weren't enough gyarus in the original dataset?
https://fediffusion.art/media/32214ff4c1b4bdc6fd4066548e9594d2357eb8b2274eb56c8d10cd61b3801c4d.jpg
https://fediffusion.art/media/c3935d0c441c9b138d46e7cff607fdcf8f2b0cb4ddb2ae15dc641969d5406931.jpg
https://fediffusion.art/media/1069562aee9d684680ea2f6d8b4d989eb7cf0b359a22d4e3b6e663fd3f7b18a8.jpg

2025-12-11 16:53:38 GMT

https://fediffusion.art/media/793cd6f2173bcb0059826014d667a3253d315bc0a4a93a98dd62026ffe07f7cc.png

2025-12-05 18:36:38 GMT

there's a huge amount of people who, upon learning that a bunch of bytes can learn a whole new style of art in 3 hours, based on a dozen examples, only feel despair and hatred.
https://fediffusion.art/media/656928e4a271c7b73fa1b614d484035a7a526b0cdb1354f9b3a74d587af9c840.png

2024-09-03 14:46:16 GMT

cc @npub1kr9…vpk6
https://fediffusion.art/media/4c474de9cc44ff479bbfeb1a7128406d411bba5524e4b3d3597e2e4ca011f104.jpg

2024-09-03 11:48:43 GMT

https://fediffusion.art/media/b8a5084f4a33dd0894699cb37f09e5a56f88e9c6e19b08a66c731adc5eb96b6c.png

2024-09-02 10:53:29 GMT

https://fediffusion.art/media/bfad9e51d022901c0cf4b08c996e194e7d17b57a4acea33a195e14e3995ce3ef.png

2024-08-28 16:48:46 GMT

> I'm shutting down http://therapistai.com/, my most controversial product ever I think
horrible domain name

2024-08-28 16:44:11 GMT

https://fediffusion.art/media/f759e1029c5550e7527cc09ceba00168153747d76867062a8ef93cf588d5c3b6.png

2024-08-27 18:47:03 GMT

https://fediffusion.art/media/53625444735d1e8a55f3cdbf619b62cfd14b1196f391adc18ecc72cfb3be29df.png

2024-08-27 16:33:08 GMT

flux dev really is the new king
https://fediffusion.art/media/b76b80c6395645d8916cda40dde0fee33fba1ce3d26be06c89557ab4b34f9623.jpg

2024-08-25 05:39:29 GMT

8 year old kid building an LLM chat website by talking to claude: https://x.com/rickyrobinett/status/1825581674870055189

2024-08-09 21:37:29 GMT

> run python main.py
> downloads 50gb of models
> crashes
> refuses to elaborate
i love python-based machine learning!!

2024-08-06 15:02:52 GMT

flux is really getting there
https://fediffusion.art/media/aea7e8a08b01869549afa77c79d4919f5d7f322d0caf07a488cbfba7153ee473.png

2024-08-04 18:44:16 GMT

> You might think--why would someone write an entire article justifying that language models are useful??! Isn't that obvious?!? But there seem to be a (large?) contingent of people out there---in the academic literature, in the software engineering space, and also in the media sphere---who proclaim widely that LLMs contribute nothing, are just another hype cycle, and in a few years will die having had no impact on the world. I will be arguing these people are wrong because current LLMs are already useful.
https://nicholas.carlini.com/writing/2024/how-i-use-ai.html

2024-08-03 07:22:48 GMT

running LLMs has become insanely cheap
https://fediffusion.art/media/eb5c915db31b4807cd40134c7f247d62d9100b119f5f4486968bbe5547bf2dd1.png

2024-08-02 08:46:47 GMT

Someone is selling a pleroma tan generation service
https://aieasypic.com/inspire/models/detail/fox-tan-pleroma-mascot-v10-42504

2024-08-01 17:32:04 GMT

https://fediffusion.art/media/06c97a85db13134617a6e1f2b6ce2dd48cf19896af0cb8d16673d1a61b2932b3.webp

2024-08-01 17:26:57 GMT

https://fediffusion.art/media/8a50d84004221943b4144a28fe78445493028815b1673452ad11f6ceed66fa3c.webp

2024-08-01 17:19:00 GMT

new open weight image generation model: https://blackforestlabs.ai/

2024-07-30 13:08:54 GMT

are you thinking what i'm thinking?
https://fediffusion.art/media/3146cce88bb24bf6741ba5484ef5f4e77b1c8f4246891a6c555f58252172f155.jpg

2024-07-29 11:33:00 GMT

https://fediffusion.art/media/6270f95e219c4b6af3c0bbf4cd5c3c86d588aa60f0d9e5327d4ecdcdf4548bd5.png

2024-07-29 10:58:31 GMT

- reply

@npub1c4s…aqv4 she's over 30 by now

2024-07-29 09:52:55 GMT

i'm thinking miku
https://fediffusion.art/media/931954079f2febaa057aedc27c380d725ba961445a107cb2bf3d05109134161a.png

2024-07-26 19:59:09 GMT

new paper in nature that shows that the way that all current models are successfully using to generate training data "doesn't work". good that it didn't come out earlier.
https://fediffusion.art/media/962bb248227377d3b37bb87422ede4723a18036edec5cc5ce84c5f900645af0f.png

2024-07-24 20:30:15 GMT

- reply

@npub17wr…prsu what do they have to show for it though

2024-07-24 20:28:34 GMT

another surprising AI hero: salesforce ????
https://fediffusion.art/media/3ae03f76eebdd1394eb46814b972f80c11de8670094069984b8ba41f558987f2.png

2024-07-24 17:22:45 GMT

le based Mistral want to one-up Meta and release an open 120b model that also reaches gpt4 quality, at least in certain tasks. It's also highly multilingual, I think this model would be the perfect japanese learning partner. Non-commercial license, sadly.
https://mistral.ai/news/mistral-large-2407/

2024-07-23 15:37:54 GMT

Meta releases Llama 3.1, an 8b, 70b and 405b model, all of them with a large context size of 128k.
The 405b model beats chatgpt 4o on nearly every benchmark.
https://ai.meta.com/blog/meta-llama-3-1/

2024-07-16 16:07:29 GMT

probable future US vice president
https://fediffusion.art/media/91b0ceef98dc184235e0b874d3f6943eb54ac402345cbb5df4b2079067c305e4.png

2024-06-11 18:07:03 GMT

https://fediffusion.art/media/3dc6d81dbcc7f850c975c2bcea630e82916243c24ef2c5c2b177d94b374bf874.png

2024-06-10 20:27:26 GMT

https://fediffusion.art/media/134ea6369887a0b7b120ceab53730962cdd866619bed83c7b36c3ffe14d32ab8.png

2024-05-17 08:39:07 GMT

I have noticed that, somewhat contrary to what I would have expected, religious / spiritual people have little problems with LLMs (in fact, they have a lot of 'discussions with chatgpt about faith' podcasts out), while the people who are deadly afraid of it and think it will be the downfall of society are overwhelmingly materialist progressive types. I have some ideas about this but for now it's just an observation.

2024-05-05 16:08:17 GMT

patiently waiting for SD3 to finally fix hands
https://fediffusion.art/media/1930d12c4148947f8c8370eb8fc473eaf7c64f9ad504723b22e2ff2260ba970f.png

2024-04-03 06:52:28 GMT

https://fediffusion.art/media/4a4f9581dc9173881b24644e802b264db23bb80beb38608aab42830505afe747.png

2024-03-11 12:28:33 GMT

👀
https://fediffusion.art/media/c6655bc1ea1dad2ab3dc789c7638617c30b0e7e37b8b22f0729e2d217550deb7.png

2024-01-31 09:16:57 GMT

- reply

@npub15m9…xnsz probably still llava but with llamacpp

2024-01-31 09:03:44 GMT

the new llava release is 🔥
https://fediffusion.art/media/0c656f47976e81f0f308b5a37260d58bd3da414c7a89e977afbab1694013943a.png
https://fediffusion.art/media/aa771da42a8b94d0a3daa408cfefe5e790251ea14a6d65516143f11a7a8c06a6.png
https://fediffusion.art/media/cd3758958eb948e57034bbcc8f01e359b43b9cd83968ab4a35f64480605c5bfd.png

2024-01-29 19:06:56 GMT

what's the songlist?
https://fediffusion.art/media/ea02c8fd6605ac56dcc9388f0301bc1a24a577c3a27c3610f381272183389e00.png