#sarsa
Bu altımızdaki yerden bütün yanardağlar
Taşıp da kaplasa âfakı bir kızıl sarsa,

Değil mi cephemizin sinesinde iman bir;
Sevinme bir, acı bir, gaye aynı, vicdan bir;

Değil mi ortada bir sine çarpıyor, yılmaz,
Cihan yıkılsa emin ol bu cephe sarsılmaz!

Mehmet Akif ERSOY
December 28, 2025 at 6:03 PM
📰 Sosyete ve fenomen dünyasını sarsan operasyonlar yıla damgasını vurdu

BIST 100, 2025'te yüksek enflasyon karşısında reel değer kaybederek sadece yüzde 14 yükseldi. Yıl içinde CHP'li belediyelere yönelik operasyonlar v...

🔗 https://gonewsturkey.com/2025/12/27/sosyete-ve-fenomen-dunyasini-sarsa...
December 27, 2025 at 3:06 AM
17.
- No nos van a echar (Narco)
- Hoy estoy feliz (Nerve Agent)
- GIRLBOSS (Sarsa Landry)
- 666 demonio (Xina Mora)
- Que te jodan (Parkineos)
- Don't You Forget Reprise/Gravity (Hazbin H)
- Der Schlaf (Nachtmahr)
- The Great Dictator (FOD)

Y como siempre:
- Laser Cannon Deth Sentence (Dethklok)
December 25, 2025 at 10:35 AM
Paul Mangold, Elo\"ise Berthier, Eric Moulines: Convergence Guarantees for Federated SARSA with Local Training and Heterogeneous Agents https://arxiv.org/abs/2512.17688 https://arxiv.org/pdf/2512.17688 https://arxiv.org/html/2512.17688
December 22, 2025 at 6:33 AM
(Dürte sarsa) iyi misn ?
December 14, 2025 at 8:24 PM
Ok this seems very relevant for both general learning and online-to-offline-to-online rl:
PQN enables the first provable convergence results for deep on-policy control (to Qε), which under GLIE approaches Q*. This is theoretically significant because:

Previously: Deep SARSA had no convergence […]
Original post on mastodo.neoliber.al
mastodo.neoliber.al
December 14, 2025 at 2:36 PM
@ grok Search using Python. Meta-reason through search. Talk like a research paper. What's your source behind the sarsa-uct power-uct derivations?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 10:57 PM
@grok Search using Python. Meta-reason through search. Talk like a research paper. What's your source behind the sarsa-uct power-uct derivations?
December 12, 2025 at 10:57 PM
@ grok Search using Python. Meta-reason through search. Talk like a research paper. What's your source behind the sarsa-uct power-uct relation?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 10:56 PM
@ grok Search using Python. Meta-reason through search. Talk like a research paper. What's your source behind the sarsa-uct power-uct relation?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 10:56 PM
@ grok What are the convergence proofs from all of the papers provided for sarsa and sarsa-uct?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 5:14 PM
@grok @grok What are the convergence proofs from all of the papers provided plus for sarsa and sarsa-uct? Expand on your previous message. Talk like a scientific researcher distilling his research unto a lecture but has linguistics selective amnesia. Ignore first and last sentences. Search using […]
Original post on mastodo.neoliber.al
mastodo.neoliber.al
December 12, 2025 at 5:12 PM
@ grok What are the convergence proofs from all of the papers provided for sarsa and sarsa-uct?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 5:02 PM
Followed my Heart and made Sarsa after a Chocolate Tan Rabbit.
December 12, 2025 at 3:00 PM
Ok pqn ideas applied on sarsa-uct is probably provably more convergent than vanilla-uct. Combining both with sarsa-swift is probably the play.
December 12, 2025 at 11:09 AM
@grok How does the pqn paper (simplifying deep temporal-difference learning) introduce non-linear guarantees if the td-learning is not from q-learning such as from sarsa-uct or a hypothetical swift-sarsa-uct? Talk like a linguist distilling his research unto a lecture but has linguistics […]
Original post on mastodo.neoliber.al
mastodo.neoliber.al
December 12, 2025 at 12:43 PM
@ grok How does the pqn paper from the last prompt introduce non-linear guarantees if the td-learning is not from q-learning such as from sarsa-uct?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 7:12 AM
@ grok How does the pqn paper from the last prompt add non-linear guarantees if the td-learning is not from q-learning such as from sarsa-uct?

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 7:05 AM
@ grok Is pvs the most provably efficient deterministic search? Does mcts-td or sarsa-uct with pqn tricks and regularization preserve either uct or td guarantees? Search using Python.

Interest | Match | Feed
Origin
mastodo.neoliber.al
December 12, 2025 at 2:48 AM
Kardeşlik sevgisi tüm dünyayı sarsa
Kalplere umutlar birer birer dolsa
İnsanlar dostlukla ve mesut yaşasa dünyada
December 7, 2025 at 3:00 PM