SELAUR: Self Evolving LLM Agent via Uncertainty-aware Rewards
Dengjia Zhang, Xiaoou Liu, Lu Cheng, Yaqing Wang, Kenton Murray, Hua Wei
https://arxiv.org/abs/2602.21158 https://arxiv.org/pdf/2602.21158 https://arxiv.org/html/2602.21158
arXiv:2602.21158v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly deployed as multi-step decision-making agents, where effective reward design is essential for guiding learning. Although recent work explores various forms of reward shaping and step-level credit assignment, a key signal remains largely overlooked: the intrinsic uncertainty of LLMs. Uncertainty reflects model confidence, reveals where exploration is needed, and offers valuable learning cues even in failed trajectories. We introduce SELAUR: Self Evolving LLM Agent via Uncertainty-aware Rewards, a reinforcement learning framework that incorporates uncertainty directly into the reward design. SELAUR integrates entropy-, least-confidence-, and margin-based metrics into a combined token-level uncertainty estimate, providing dense confidence-aligned supervision, and employs a failure-aware reward reshaping mechanism that injects these uncertainty signals into step- and trajectory-level rewards to improve exploration efficiency and learning stability. Experiments on two benchmarks, ALFWorld and WebShop, show that our method consistently improves success rates over strong baselines. Ablation studies further demonstrate how uncertainty signals enhance exploration and robustness.
toXiv_bot_toot
A man accused of using thousands of bot accounts to stream AI songs, making millions in royalties, pleads guilty in US district court to wire fraud conspiracy (Murray Stassen/Music Business Worldwide)
https://www.
🇺🇦 #NowPlaying on BBCRadio3's #Breakfast
Wolfgang Amadeus Mozart, Murray Perahia & English Chamber Orchestra:
🎵 Piano Concerto No 26 in D major, K 537 (3rd mvt)
#WolfgangAmadeusMozart #MurrayPerahia #EnglishChamberOrchestra
Good Morning #Canada
#HappyBirthday to Benjamin Mink, born January 22, 1951, in Detroit, Michigan. He became a Canadian after he moved with his parents to Toronto at age 11. He is likely one of Canada’s least known musical talents despite his long career as a songwriter, multi-instrumentalist, and producer. His industry connections are extensive as he has produced and/or performed with K.D Lang, Barenaked Ladies, Anne Murray, Dan Hill, Mendelson Joe, Prairie Oyster, Raffi, Jane Siberry, Ian and Sylvia Tyson, Valdy, Bruce Cockburn, Murray McLauchlan, Mick Jagger, Susan Aglukark, Methodman, Alison Krauss, Feist, Daniel Lanois, Sarah McLachlan, Roy Orbison, Elton John, and Heart.
Despite releasing only one album under his own name, Mink has multiple Juno, Genie, Gemini, and Grammy Awards.
#CanadaIsAwesome #CanadianMusic
https://en.wikipedia.org/wiki/Ben_Mink
Crosslisted article(s) found for cs.GR. https://arxiv.org/list/cs.GR/new
[1/1]:
- PointSLAM : Robust Dense Neural Gaussian Point Cloud-based SLAM
Xu Wang, Boyao Han, Xiaojun Chen, Ying Liu, Ruihui Li
https://arxiv.org/abs/2601.11617 https://mastoxiv.page/@arXiv_csCV_bot/115932177919726523
- Soft Shadow Diffusion (SSD): Physics-inspired Learning for 3D Computational Periscopy
Fadlullah Raji, John Murray-Bruce
https://arxiv.org/abs/2601.12257 https://mastoxiv.page/@arXiv_csCV_bot/115933514032062891
- NeuralFur: Animal Fur Reconstruction From Multi-View Images
Sklyarova, Kabadayi, Yiannakidis, Becherini, Black, Thies
https://arxiv.org/abs/2601.12481 https://mastoxiv.page/@arXiv_csCV_bot/115933577604103858
- Deep Feature Deformation Weights
Richard Liu, Itai Lang, Rana Hanocka
https://arxiv.org/abs/2601.12527 https://mastoxiv.page/@arXiv_csCV_bot/115933596082529002
- Criminator: An Easy-to-Use XR "Crime Animator" for Rapid Reconstruction and Analysis of Dynamic C...
Pooryousef, Besan\c{c}on, Cordeil, Flight, Ross, Bassed, Dwyer
https://arxiv.org/abs/2601.13689 https://mastoxiv.page/@arXiv_csHC_bot/115933503545993846
- Rig-Aware 3D Reconstruction of Vehicle Undercarriages using Gaussian Splatting
Kulkarni, Devarashetti, Cluss, Forte, Buckmaster, Schneider, Qiao, Vereshchaka
https://arxiv.org/abs/2601.14208 https://mastoxiv.page/@arXiv_csCV_bot/115933812616301038
toXiv_bot_toot
The Federal Court of Appeal delivered two major environmental victories last month. The court found former Fisheries and Oceans Minister Joyce Murray was “procedurally fair and reasonable” when she did not renew licences for open-net salmon pens in the Discovery Islands in 2023. The same court found that the federal government acted well within its purview when it listed plastic manufactured items as “toxic” under the Canadian Environmental Protection Act in 2021.
One fascinating film trivia I learned recently is that Sofia Coppola and Spike Jonze’s marriage from 1999 to 2003 partially inspired Coppola’s 2003 film Lost In Translation and Jonze’s 2013 film Her.
Coincidentally, both films star Scarlett Johansson as the main female protagonist, both films were nominated for the Oscar Best Picture, and both films won the Oscar Best Original Screenplay for Coppola and Jonze.
The Washington Post told employees on Wednesday that
it was beginning a widespread round of layoffs that are expected to
decimate the organization’s sports, local news and international coverage.
Matt Murray, The Post’s executive editor, told newsroom employees on a call that
all sections would be affected in some way.
But he said that the end result would be a publication that focused even more on national news and politics,
as well as business and health…
Apple Music introduces Transparency Tags, a metadata system for music distributors and labels to disclose AI-generated creative elements like artwork or lyrics (Murray Stassen/Music Business Worldwide)
https://www.musicbusinesswor…
Muttered and crawled back in his hole after once again being misidentified as that varmint from that other Bill Murray movie.
#GroundHogSawAndHappened
#HashTagGames
🇺🇦 #NowPlaying on BBCRadio3's #PrivatePassions
Benjamin Britten, Stuart Skelton, Stuart Skelton, Bergen Philharmonic Orchestra, Marcus Farnsworth, Bergen Filharmoniske Kor, Erin Wall, Edward Gardner, Robert Murray, Susan Bickley, Catherine Wyn-Rogers, Roderick Williams & James Gilchrist:
🎵 Peter Grimes: Now the Great Bear...He's mad or drunk...Old Joe..
#Spotify
A coalition of music publishers, including UMG, filed a second lawsuit against Anthropic, seeking $3B over allegations of copyright infringement of 20k songs (Murray Stassen/Music Business Worldwide)
https://www.
🇺🇦 #NowPlaying on BBCRadio3's #EssentialClassics
Murray Perahia & Fryderyk Chopin:
🎵 Study in A flat major, Op.25 no.1
#MurrayPerahia #FryderykChopin
#Spotify
🇺🇦 #NowPlaying on BBCRadio3's #EssentialClassics
Franz Schubert & Murray Perahia:
🎵 Piano Sonata No 20 (3rd mvt)
#FranzSchubert #MurrayPerahia