HackerTrans
TopNewTrendsCommentsPastAskShowJobs

GaggiX

4,812 karmajoined 5 anni fa

Submissions

Explaining Attention with Program Synthesis

arxiv.org
2 points·by GaggiX·8 giorni fa·0 comments

FlashMemory-DeepSeek-V4

arxiv.org
1 points·by GaggiX·mese scorso·1 comments

Ideogram 4.0 Technical Details: Open model at the forefront of design

ideogram.ai
3 points·by GaggiX·mese scorso·0 comments

[untitled]

1 points·by GaggiX·mese scorso·0 comments

Snowboard Kids 2 is 100% Decompiled

blog.chrislewis.au
284 points·by GaggiX·2 mesi fa·109 comments

Nvidia: Nemotron Labs Diffusion 14B

huggingface.co
3 points·by GaggiX·2 mesi fa·0 comments

Cerebras Brings Trillion Parameter Inference to Enterprises with Kimi K2.6

cerebras.ai
2 points·by GaggiX·2 mesi fa·0 comments

Generalized Discrete Diffusion from Snapshots

arxiv.org
1 points·by GaggiX·4 mesi fa·0 comments

Attention Residuals

github.com
241 points·by GaggiX·4 mesi fa·34 comments

Eval awareness in Claude Opus 4.6's BrowseComp performance

anthropic.com
3 points·by GaggiX·4 mesi fa·0 comments

MathArena: Evaluating LLMs on uncontaminated math questions

matharena.ai
2 points·by GaggiX·5 mesi fa·0 comments

Step 3.5 Flash

static.stepfun.com
2 points·by GaggiX·5 mesi fa·0 comments

FLUX.2 [Klein]: Towards Interactive Visual Intelligence

bfl.ai
227 points·by GaggiX·6 mesi fa·58 comments

OmniPSD: Layered PSD Generation with Diffusion Transformer

showlab.github.io
1 points·by GaggiX·7 mesi fa·0 comments

NaTex: Seamless Texture Generation as Latent Color Diffusion

natex-ldm.github.io
2 points·by GaggiX·8 mesi fa·0 comments

Back to Basics: Let Denoising Generative Models Denoise

arxiv.org
3 points·by GaggiX·8 mesi fa·0 comments

Veo 3.1 and new creative capabilities in the Gemini API

developers.googleblog.com
4 points·by GaggiX·9 mesi fa·0 comments

Linus Learns Analog Circuits

github.com
171 points·by GaggiX·10 mesi fa·42 comments

UnifoLM-WMA-0: A World-Model-Action (WMA) Framework Under UnifoLM Family

unigen-x.github.io
2 points·by GaggiX·10 mesi fa·0 comments

comments

GaggiX
·14 ore fa·discuss
I really like Gemini 2.5 Flash Lite because it's a dirt cheap model that support every input modalities.

At least now MiMo v2.5 exists and can be used as another dirt cheap multimodal model.
GaggiX
·14 ore fa·discuss
Not at all, even in the Borno state where Boko Haram and ISWAP are present, towns are controlled by the government, I don't think there is a single town that is controlled by Boko Haram/ISWAP right now.
GaggiX
·16 ore fa·discuss
I would be more interested about terrorists organization like Al-Shabaab that at least control many towns.

Does Boko Haram and ISWAP even control a single town or they just control a few villages in Lake Chad and in the Sambisa forest?

Also reading the report they seem quite clueless.
GaggiX
·l’altro ieri·discuss
Grok did not render anything, they had to prompt it again.
GaggiX
·11 giorni fa·discuss
Someone should try it while sleeping and see if anything is related to a dream.
GaggiX
·15 giorni fa·discuss
Popular open models on Openrouter have dozens of providers.
GaggiX
·27 giorni fa·discuss
I guess we can start a chain of comment listing their favorite ones, I haven't tried many but "wobble stack" is pretty cool and simple, "Starfall Defense" is a much more complex game but I love tower defense.
GaggiX
·28 giorni fa·discuss
One of the many therapies that are being developed so that you can survive longer even with the most lethal tumours.
GaggiX
·28 giorni fa·discuss
Chinese models have become really good and cheap. MiMo V2.5 Pro, Kimi K2.7-code, Minimax M3 etc
GaggiX
·29 giorni fa·discuss
This is just an Ad, I thought it was a new product from Kagi.
GaggiX
·29 giorni fa·discuss
I know nothing about this field, but I imagine the actual problem is how do you deliver the Cas12a2 protein to each individual cancer cell compare to a viral gene therapy?
GaggiX
·mese scorso·discuss
Codex also works, before Opus 4.8 and Fable it wasn't very clear who had the best agentic model.
GaggiX
·mese scorso·discuss
Well with a standard autoregressive model you can generate for example 256 tokens at once if you have 256 users, with this approach you can generate 256 tokens for a single user but you need several forward steps.

So the diffusion process takes more GFLOPs, if you have enough users you can already balance memory and compute.
GaggiX
·mese scorso·discuss
$0.435/$0.87 for the standard speed, this one should be 3 times that.
GaggiX
·mese scorso·discuss
Not to be confused with Flash Attention.

What's novel here is the extremely small KV cache memory usage per long context windows, like 0.77GB with 512K, a 90% memory usage reduction compare to the already really small KV cache memory usage of Deepseek V4 Flash.
GaggiX
·mese scorso·discuss
>I’ll take a few f bombs and the truth.

Don't want to ruin it but go read some old posts from the author about AI, the tone is the same and he is very much wrong.
GaggiX
·mese scorso·discuss
If MiMo v2.5 Pro can run at >1000tk/s on GPUs then I will soon expect the same from OpenAI/Anthropic/Google.
GaggiX
·mese scorso·discuss
> That's technically encoding

Isn't that just projecting the patches into the d_model size vectors that the models takes?

>I am assuming that involves of quantization

12B model in 16GB seems very reasonable to me, int8 is top quality for running models.
GaggiX
·mese scorso·discuss
>Even though it feels vibed

Where does it feel vibed?
GaggiX
·mese scorso·discuss
Is opus being used for the audio or it's not the solution for extreme lossy compression?