Explore tweets tagged as #TestTimeScaling
@SOURADIPCHAKR18
Souradip Chakraborty
2 months
Excited to be at #ICML2025 in Vancouver 🇨🇦! To present new paradigm of #AIAlignment on #BoundedRationality & Satisficing principles Looking forward to connecting with old and new friends, discussing #Reasoning #Alignment #LLMsinRL #Testtimescaling #Posttraining #AgenticAI
Tweet media one
2
1
10
@Saif_BinSafwan
سيف محمد بن صفوان | Saif Bin Safwan
7 months
نماذج لغوية صغيرة (SLMs) تتفوق على الكبيرة (LLMs) في مهام الاستدلال المعقدة عبر تقنيات "توسيع وقت الاختبار" (TTS). مثال: نموذج Llama-3.2-3B مع TTS يتفوق على Llama-3.1-405B في حل مسائل رياضية معقدة #AI #ML #NLP #DeepLearning #LanguageModels #TestTimeScaling #SmallLanguageModels
Tweet media one
0
1
1
@TrinusOfficial
Trinus Corporation
2 months
Have a look at this article to know how test-time scaling unlocks hidden reasoning abilities in small language models: https://t.co/34gJhtG1lH #TestTimeScaling #AIReasoning #SmallLanguageModels #AIResearch #MachineLearning #AIInnovation
0
0
2
@KINEWS24
KINEWS24.de
7 months
Kleine KI-Modelle übertreffen große: Effizienz neu definiert Klein schlägt groß Effizienz neu definiert Revolution in der KI #ai #ki #artificialintelligence #kimodell #effizienz #revolution #testtimescaling https://t.co/Qa6iKASmfc
0
0
2
@andresvilarino
Andres Vilariño 🇪🇦
7 months
Can 1B #LLMs Surpass 405B LLM? Optimizing Computation for #SmallLLMs to Outperform Larger Models #LargeLanguageModels #TestTimeScaling #TTS #AI #ArtificialIntelligence #Tech #Technology https://t.co/jWSPvngqAd
Tweet media one
0
0
0
@vlruso
Vlad Ruso PhD
7 months
Can 1B LLM Surpass 405B LLM? Optimizing Computation for Small LLMs to Outperform Larger Models #TestTimeScaling #SmallLLMs #PerformanceOptimization #AIResearch #ComputationalEfficiency https://t.co/LrFu62XBRR
Tweet media one
0
0
0
@vlruso
Vlad Ruso PhD
7 months
s1: A Simple Yet Powerful Test-Time Scaling Approach for LLMs #LanguageModels #TestTimeScaling #AIResearch #MachineLearning #InnovationInAI https://t.co/m3EmedYDIV
Tweet media one
0
0
1
@HaywaaVietnam
HaywaaVietnam
9 months
Mô hình o3 của OpenAI thể hiện tiến bộ lớn nhưng tăng chi phí vận hành. #TríTuệNhânTạo #Openai #TesttimeScaling https://t.co/xOvCWAjXnP
0
1
1
@HaywaaWorldwide
HaywaaWorldwide
9 months
OpenAI's o3 model excels in performance but with high costs. #AiScaling #TesttimeScaling #AiModels https://t.co/rh1ZWglEZk
0
1
2
@agents_ia
Agents IA 🤖 Cryptos 🔗 Business 💼
8 months
Ça décoiffe : sur l’ARK AGI Benchmark, les meilleurs LLM plafonnaient à ~20-30 %. Avec le #TestTimeScaling, on grimpe jusqu’à 70-88 %! Mais la facture GPU peut vite monter, car plus de « réflexion » = plus de tokens = plus de 💸
1
0
0
@agents_ia
Agents IA 🤖 Cryptos 🔗 Business 💼
8 months
Depuis 2017 et les #Transformers, on a surtout fait grossir nos IA via + de paramètres et + de data. Mais voici la 3e voie 🛤️: le #TestTimeScaling. En gros, on autorise un LLM à « réfléchir » davantage durant l’inférence, au lieu de lâcher la 1ʳᵉ réponse venue.
1
0
0
@StayUpdated_247
Stay Updated
7 months
Scaling Test-Time Inference: The Future of Edge AI Integration and Performance Enhancement https://t.co/sCCM9CZLew #AIinference #DistributedIntelligence #EdgeAI #realtimeprocessing #resourceallocation #testtimescaling
Tweet media one
0
0
0
@agents_ia
Agents IA 🤖 Cryptos 🔗 Business 💼
8 months
Et ce n’est pas que pour le texte : les modèles de diffusion (#StableDiffusion, #DALLE) testent aussi le #TestTimeScaling. En débruitant + longtemps et en validant les meilleures itérations, on obtient des images plus qualitatives.
1
0
0
@Ami_Subha
Subha | nemotivity.eth/lens
7 months
The rise of test-time scaling is driving the next era of AI!🚀 3 AI scaling laws: 1) Pretraining scaling 📈 2) Post-training scaling 🎯 3) Test-time scaling (long thinking) 🧠 How will AI reasoning transform your industry?🤔💬 #ArtificialIntelligence #AIScaling #TestTimeScaling
0
0
0
@InferDomCOM
InferenceDomains.com
9 months
Also available - https://t.co/53bcnhsGcW - https://t.co/Egps5whMdo - https://t.co/JNweOM9udH - https://t.co/LN8JEYPaTW - https://t.co/1D9yihf4Sz ....and many more! Ps the whole portfolio is open for sale, if someone wants to own the whole AI Performance Metric industry!
0
0
0
@cbemerine
Cbemerine
7 months
As many have said, #DeepSeek is using $NVDA chips to as they say #TestTimeScaling of #AI #DATA. So many dumped on #Nvidia when #DeepSeek could not exist without $NVDA chips. Will it recover price today or by end of week. This was a buying opportunity for many #stocks.
@StockMKTNewz
Evan
7 months
Nvidia $NVDA just released a statement regarding DeepSeek: "DeepSeek is an excellent AI advancement and a perfect example of Test Time Scaling. DeepSeek’s work illustrates how new models can be created using that technique, leveraging widely-available models and compute that is
0
0
1