1/5 Launching Kepler Systems—an AI research initiative where I build open models and share datasets. Starting with datasets I’ve curated to preserve Indo-Pak Urdu poetry’s richness. No corporate labs here—just a solo researcher and a lot of love for language and art.
@ObinoWilliam@HSVSphere Most of this is correct except it was Morphe who squashed commit history. You can see it on their own first commit. ReVanced gave proper co-author attributions to every code originating from Morphe. Basically fight got intensified when Morphe added least legally required credits
@HSVSphere Former ReVanced dev here. It's still active and being updated. There is some drama involved between team members (I am close to all of them) but it's nothing devastating. I still use it as my main (though I am working on my own patcher in rust). Lmk if you have any questions
@menhguin I forgot to attach the code: https://t.co/5To0DrHRnd. It's very simple, I just send the song and artist names. Gemini works the best because of strong vision though Kimi has the best taste. I got like 79% compression ratio
There you have it. Anthropic's CEO said it: The murder of more than 100 schoolgirls in Minab targeted by Anthropic's CLAUDE "is a use case that doesn't even violate our red lines." Time to rise up against these technofeudal war criminals.
Ironic because Nex is also not a base model. And model merging has been a common thing since inception of oss llms. Literally most of the early popular models were merges of multiple models (especially in RP use cases)
The Rio 3.5 model broke the internet this week. The plot twist? It’s essentially our open-source model, Nex N2 Pro, wearing a different hat.
🤯 We analyzed the weights, and the recipe is exact: Rio 3.5 ≈ 0.6 * Nex N2 Pro + 0.4 * Qwen 3.5
It even literally introduces itself as "Nex N2 Pro" if you ask it without initial system prompt!
😂 We are flattered that the City of Rio used our work to achieve SOTA performance. Thanks for the ultimate benchmark validation.
🤝 But in the open-source world, attribution matters.
👇 Full mathematical proof & verify script in the first reply!
Back to Codex/Opus. Fable flags my patcher.
Prompt: how do we make this memory efficient and faster? this needs to run on phone and it is very important for it to be efficient
@giffmana Definitely Immich for photos. For dropbox/gdrive alternate checkout Cloudreve.
I didn't have very good results for face recognition though. Nowhere near Google photos (as expected) but smart search and OCR is very nice. My library is ~32k photos and ~2.2k videos
@badlogicgames I ported Pi (only agent core) to Go without any tests. It's sloppy but I'll clean it up overtime. Sucks that Pi is TS only but majority of my new projects are in Go.
https://t.co/ZuXSCpwylO
@norpadon It's also very very good at audio understanding. Competes fairly with Gemini 2.5 Pro (beating flash!). Unfortunate that it's limited to only 30 seconds.