Decstar
54 posts
Decstar
@obsolete_dev
Software guy optimizing my own obsolescence. https://t.co/ppZhHLxT9J
Entrou em Ocak 2026
65 Seguindo1 Seguidores
Vibe coded my own AI sprite tool because I kept doing the same workflow for walkcycles:
> Images 2.0 for reference sprite
> Seedance 2.0 to create walkcycle
> Split walkcycle into frames
> Select start end frames
> Auto-select in betweens (tweens)
> Generate spritesheet
Chong-U@chongdashu
More animated pixel art using Images 2.0 and Seedance 2.0. They work really well even without clear legs (as we see with the robe here)
English
If you still don't hate recruiters enough, or if you are still not convinced that hr is the most incompetent, harmful, and entitled industry in tech, just watch this video.
But TLDR is "I have 20 seconds to find 5 keywords in your CV, but nobody taught me how to use Ctrl+F"
English
@tautologer Hahah I felt so bad just idling a 5090 I rented out while I was figuring out how to transfer files lol
English
this is my first time spinning up a fat GPU node and man
idling 8xH100 while waiting for weights to download feels like a crime
it's also very expensive 😬
English
I trained an MLP and CNN from scratch to compare them in digit recognition. Here is the live demonstration.
blog.decstar77.com/project-mnist.…
English
M2.7 w/ hermes cli is replacing ~75% of my claude code / opus usage now, but we need clarity for using it as a coding agent @ work.
We're truly blessed to have the weights of this one, looking forward to seeing the license change. Definitely a model worth checking out.
RyanLee@RyanLeeMiniMax
Self-hosted M2.7 for code writing is absolutely allowed and free of charge!!!😁 I think this license is not detailed enough, so I will update it.
English
Можно повыебываюсь?!!!
Собрал сыну комп в честь его днюхи…
Пиздатый я батя да!?🤓
Русский
I’ve been diving into the math behind diffusion models and wrote a step-by-step breakdown of the backward process. Most resources felt disjointed or assumed too much prior knowledge. Mine is more gradual. Hope this helps!
blog.decstar77.com/diffusion-back…
English
Implemented both a MLP and Conv2D on the MNIST dataset.
You can play around with it live here:
blog.decstar77.com/project-mnist.…
English
I wonder if anyone in Warsaw would like to purchase this preconfigured, preoptimised and heavily benchmarked rig.
I can give a full spec sheet, including all the models it can run, as well as all the thousands of benchmarks I’ve ran.
I can also custom configure it for you
English
I’ve been diving into how diffusion models actually work and put together a blog post breaking down the math step by step for the forward pass.
blog.decstar77.com/diffusion-forw…
English
Nous officially cooked with Hermes Agent.
First time using a local-model agent that replaces a large segment of my Claude Code workflow.
Using Qwen3-Coder-Next 80B 4bit, plus a swarm skill for ensemble on harder stuff a model is getting stuck on.
Harrison Kinsley@Sentdex
This feels so random in the universe of possible agent skills, but now I'm going to have to setup Hermes just to try this out. LLMs historically are pretty darn bad @ manim, but it's an exceptionally useful animation engine.
English
