π”π”žπ”―π”¦π”¬! (@ratpenant) header
π”π”žπ”―π”¦π”¬! (@ratpenant) profile picture

Bioinformatics BSc. @ UMA AI MSc. @ UAM Math enthusiast

Madrid, EspaΓ±a
Joined December 2014
estoy dando volteretas
71
50
9
1,208
227,538
no me puedo creer que existiera este banger de quote y no la haya usado mΓ‘s en.wikiquote.org/wiki/Fritz_…
36
π”π”žπ”―π”¦π”¬! retweeted
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
596
1,622
982
17,334
3,003,620
nya.soy/loving-controls de μ•ˆκ°œκ΅¬λ¦„ (Mercuriusdream)
Distillation can reproduce powerful AI capabilities without the safeguards built around them. We’re sharing new research on our work to combat distillation of our models, how these capabilities transfer, and how we detect and disrupt coordinated attempts. These findings are helping us sharpen our defenses and support safeguards across the industry. openai.com/index/combatting-…
34
π”π”žπ”―π”¦π”¬! retweeted
esto es asi de hecho
134
1
609
8,258
no puede ser no puede ser cΓ³mo que hay una propuesta de GRH
1
70
seminario de pln para clasificar comentarios de la manosfera increΓ­ble
1
45
π”π”žπ”―π”¦π”¬! retweeted
What do diffusion models, representation learning, reinforcement learning, and test-time reasoning have in common? Each can be seen as learning an energy function or using one to guide sampling and search. Excited to share our tutorial on energy-based models with @daibond_alpha!
1/5 EBMs are usually taught as one model family among many. In our new tutorial with @du_yilun, we use them as a single lens on machine learning. "An Energy-Based Perspective on Machine Learning" arxiv.org/abs/2611.14383
4
56
4
443
24,999
π”π”žπ”―π”¦π”¬! retweeted
"Si el cielo de Castilla es tan alto, es porque lo levantaron los campesinos de tanto mirarlo" - Miguel Delibes.
7
1,253
14
9,664
114,670
NOOO
Press the button. Press. Press it. Perish it.
1
59
π”π”žπ”―π”¦π”¬! retweeted
Esta noche, 00:00. β€œAl Cielo, Devil May Cry”, β€œCuesta Arriba Sin Frenos”, β€œMi Manera De Vivir”. Tres canciones que formarΓ‘n parte d nuestro prΓ³ximo Γ‘lbum.
47
289
256
2,397
97,590
π”π”žπ”―π”¦π”¬! retweeted
LLMs are powerful, so why do we still make them call tools to handle 3D molecules? Introducing Nahual: a decoder-only model for any sequence of text and 3D molecules. #NeurIPS2026
2
20
2
105
5,161
π”π”žπ”―π”¦π”¬! retweeted
#deadlock the ivy going 2/12 on yellow
3
719
1
6,440
24,413
π”π”žπ”―π”¦π”¬! retweeted
omi gets new teeth πŸŽ™οΈ
32
1,116
7
14,042
169,360
hermano que esto saliΓ³ como hace 2 dΓ­as??? quΓ© bonito!
Made some Claude Desktop and Claude Code mods
1
53
realmente la investigaciΓ³n en ia es prueba y verΓ‘s
1
73
π”π”žπ”―π”¦π”¬! retweeted
πŸ†• cuPotential A CUDA-X library for computing nonbonded interactions in molecular dynamics. The OpenMM plugin replaces the nonbonded force, integrator, and barostat. In NPT benchmarks on systems from 24K to 400K atoms, it ran over 3x faster than OpenMM on RTX PRO 6000 Blackwell and about 2x faster on B200, H100, A100, L40S, A10, L4, and T4. Accuracy holds up too, with a few hundred times less energy drift than default PME. πŸ’» Github nvda.ws/3WIwogJ πŸ“„ Technical Blog nvda.ws/4AyIpDX πŸ“ Find us at #SC26 πŸ‘‡
5
80
21
432
38,185
tercera vez que me equivoco de plantilla al subir un artΓ­culo al ieee tal vez el problema soy yo
2
232
π”π”žπ”―π”¦π”¬! retweeted
ζ˜¨ε€©δΈ€ε£ζ°”ηŽ©δΊ†ε…«ε°ζ—Άivyβ€¦ζˆ‘ε₯½ε–œζ¬’ε₯Ή #deadlock #ivy
19
1,647
10
14,123
67,919