Top Tweets for #lrms
Abofé que si! Vémonos o luns ás 21:45 h 😘
#LRMS
Canta razón ten esta muller! 🤪 No @LandRoberTVG, @Eddyfisterra e @violetaacrespo escoitan a sabedoría dos que levan décadas en parella 😍 https://t.co/nMrsvAn2iR
The illusion of thinking: large reasoning models (LRMs) collapse in the face of complicated tasks
https://t.co/RJ9OrnhZZJ
#km #kmers #knowledgemanagement #ai #artificialintelligence #llms #lrms #largelanguagemodels #largereasoningmodels #reasoning #thinking
#GreenAI #LRMs Beyond DeepSeek Big News on Nature. 🚀 Our MoPPS— “Model Predictive Prompt Selection” with @quyun52425662 @vtaohu for Fast RL finetuning of reasoning LLMs is Made Public on Codes now😊: https://t.co/3Og20qvDpI
🎯 Key idea: treat each prompt’s success rate as a latent variable with a Beta prior; update via Bayesian inference & posterior sampling to predict prompt difficulty — no costly LLM rollouts for every candidate.
🔍 Benefits:
• Accurately predicts which prompts are “just hard enough” (≈ success rate ~0.5) to give informative learning signal.
• Cuts down rollouts massively — only ~20–25% of rollouts compared to heavy evaluation methods like Dynamic Sampling (DS), while achieving similar or better performance.
• Gives ~1.6–1.8× speedup over uniform prompt sampling in several tasks (mathematics, planning, visual geometry).
🔄 Why Bayesian & Beta prior matter: the Beta(α,β) acts as a conjugate prior to Bernoulli outcomes → very efficient updates (α += successes, β += failures) & builds in uncertainty (“how confident are we about this prompt’s difficulty?”).
🛠️ Impact: If you’re using RL to finetune LLMs for reasoning, MoPPS offers a way to select prompts adaptively with low extra cost, maintain performance, and cut down computation. Ideal for large models / constrained budgets.
📚 Read more: arXiv:2507.04632 – MoPPS.
#LargeReasoningModel Can prompt difficulty be predicted online to accelerate RL finetuning of Reasoning Models? YES! with MoPPS:
✅Predict success rates without costly evaluation
🎯Select informative prompts online
⚡Faster training
🚀Better performance
📄https://t.co/HnTkEGnHAB
🤣🤣🤣
(Prof Subbarao Kambhampati invented the LLM-Modulo system, for solving LLM reasoning at inference time, a system that was later found in reasoning LLMs like o1 and R1)
#LLMs #LRMs #Reasoning #AI #AGI
.@Apple Reality check? #AI trajectory models: right or wrong track? With increasing hallucinations can they achieve accurate reasoning?
#ArtificialIntelligence #Innovation #technology #LRMs #LLMs
@mvollmer1 @Khulood_Almani @Shi4Tech @enilev @HaroldSinnott @IanLJones98 @antgrasso @mikeflache @efipm @Eli_Krumova @Nicochan33 @ahier @BetaMoroney @amalmerzouk @richardturrin @ralf_ladner @EliseQuevedo @SiddharthKS @arlenenewbigg @mary_gambara @NewsNeus @TAEVisionCEO @TheAdityaPatro @afigueiredo @gvalan @AnneLaureBEAUD2 @thomas_dettling @ipfconline1 @fogle_shane @pierrepinna @CurieuxExplorer @bamitav @TysonLester @AndrewinContact @AkwyZ @Analytics_699 @TerenceLeungSF @RLDI_Lamy @GlenGilmore @sonu_monika @sallyeaves @Hadel @bimedotcom @anand_narang @kalydeoo @KirkDBorne @kashthefuturist @YuHelenYu @smasked21
https://t.co/Y7VgpCcRtn
@MFarajtabar Thanks for sharing. Interestingly, our recent result also highlights the Mirage of #testtime scaling in #LRMs via thinking more? Performance improves initially and then degrades...
Link : https://t.co/SFHjZYX2h6
@amritsinghbedi3 @ghosal_suvra @furongh @MengdiWang10
🔥 Does test-time scaling in #reasoningmodels via thinking more always help?
🚫 Answer is No - Performance increases first and then drops due to #Overthinking
❓Why is this behaviour and how to mitigate
🚀 Check our recent findings #LLMReasoning
Link: https://t.co/V0IOoFqAgY
The Illusion of Thinking in LLMs
Apple researchers discuss the strengths and limitations of reasoning models.
Apparently, reasoning models "collapse" beyond certain task complexities.
Lots of important insights on this one. (bookmark it!)
Here are my notes:
(5/n) 🧪 We find that #LRMs act as efficient and stable optimizers delivering faster convergence and generating high-quality prompts.
Qualitatively, prompts optimized by DeepSeek-R1 often include extraction rules and exception handling—
crucial for Event Extraction.
Sr. AI Engineer, George Strong explains how #Test_Time_Compute has disrupted AI:
- Enabling #LRMs from @deepseek_ai to answer questions with greater depth
- Revolutionized generative image synthesis
- Making Agentic AI possible
https://t.co/hH0WGChJRl
On the Stone Soup of LLM Reasoning #SundayHarangue
Stone soup is the European folk story where some clever travelers convince the gullible locals that they are making delicious soup with a stone--and they do need a few things to "improve its flavor"--such as carrots, potatoes, onions, butter etc..
There is nothing ipso facto wrong with Stone Soup--it is, after all, soup! It may even be delicious! The question instead is how much credit should the stone get for the soup's delciousness.
The version of Stone Soup in AGI/AI circles are claims that LLMs can reason and plan--with just a few things to "improve the flavor".
These needed things needed can range from external tools/verifiers etc (as in LLM-Modulo), to tacking on search on top of LLMs (the tree of thoughts), to tacking on a Mu_zero/Alpha_go like RL component that influences pretraining as well as inference stage (the 🍓 o1 model; see https://t.co/5nivQhAXFB). The question is whether the "augmented" systems are LLMs or some other qualitatively different beasts better called LRMs (cf. https://t.co/CvHuWhlKNj).
Although it has become fashionable to equate LLMs to AI, and ask "When will AI Reason?"--the fact of the matter is that we have always had AI systems--planning, RL etc.--capable of reasoning. It can be argued that much of AI before LLMs was in fact was deep and narrow System 2 approximators (and thus https://t.co/m0Ecnvnx6E and https://t.co/NcZaqR2QzH).
The appeal of LLMs is that they are the first effective System 1 approximators that AI managed to develop (c.f. https://t.co/UsnxSYaeKU).
There have been several attempts to get System 2 reasoning behaviors from LLMs while keeping their essential autoregressive System 1 nature intact. These early attempts--such as "fine tuning", "Chain of thought" etc.--have by now been shown to be deeply flawed (c.f. https://t.co/DEjF9gLR8q & https://t.co/NE0MizcWN7). In other words, no soup from them! #Seinfeld
In contrast, the approaches that tack on search/RL etc. at the inference stage seem to be more promising (c.f https://t.co/RqRf4fWjrU). But these "compound" LRM systems are no longer autoregressive LLMs and don't have any "start completing the prompt as soon you hit return" characteristic that is such a big part of LLM popularity!
In particular, if the LRM is taking indefinite and costly inference time compute, the right comparison will be to other System 2 AI approaches that incur such inference time costs. Such considerations bring some of the traditional CS analyses--all but forgotten when online computation was given up for dead (c.f. https://t.co/I9Ya3j2EER)--back into play (see https://t.co/RqRf4fWjrU & https://t.co/JeZ8L0ryia).
Interestingly, you can also err assuming that LRMs will inherit all the limitations of LLMs (something that the analyses that clump, for example, o1 with autoregressive LLMs mistakenly make).. They don't! Stone Soup is more than Stone--even if it may not be the best way to make soup!
Specifically, many of the common critiques of LLM reasoning capabilities don't directly apply to LRMs. Not recognizing this and adding LRMs like o1 as yet another entry in a table brimming with autoregressive LLMs confuses the message. (For example, if you look at the appendix of that latest Apple study on LLM reasoning, you will see that o1 does quite fine--accuracy-wise--on their instances with irrelevant information.) Another argument for separating their analysis--as we do (c.f. https://t.co/3BUmLS7kTr).
Student-teacher swap! One of my students dressed like me and even borrowed my sweater - the LRMS band shirt was a nice touch 🙌🏼
#GoLightning #LRMS
Last Seen Hashtags on Sotwe
lowlvl()***********************************
Seen from United States
omegle
Seen from United States
momson filter:videos
Seen from Turkey
CBT
Seen from Japan
summer
Seen from United States
altcoin
Seen from United States
finansalköle
Seen from Turkey
BsnlFailsAtOurPlace
Seen from South Africa
ควยบ้านๆ
Seen from Thailand
Trends for you
Most Popular Users
Elon Musk 
@elonmusk
240.8M followers
Barack Obama
@barackobama
119.2M followers
Donald J. Trump
@realdonaldtrump
111.7M followers
Cristiano Ronaldo
@cristiano
111.2M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
97.8M followers
NASA
@nasa
92.2M followers
Justin Bieber
@justinbieber
91M followers
KATY PERRY
@katyperry
87.9M followers
Taylor Swift
@taylorswift13
81.8M followers
Lady Gaga
@ladygaga
73.3M followers
Virat Kohli
@imvkohli
70.4M followers
Kim Kardashian
@kimkardashian
69.9M followers
YouTube
@youtube
68.7M followers
Bill Gates
@billgates
64.1M followers
Neymar Jr
@neymarjr
63.2M followers
The Ellen Show
@theellenshow
62.4M followers
CNN
@cnn
61.9M followers
Selena Gomez
@selenagomez
61M followers
X
@x
60.8M followers