#AI Training
Latest news articles tagged with "AI Training". Follow the timeline of events, related topics, and entities.
Articles (17)
-
🇺🇸 Thermodynamics of Reinforcement Learning Curricula
[USA]
arXiv:2603.12324v1 Announce Type: cross Abstract: Connections between statistical mechanics and machine learning have repeatedly proven fruitful, providing insight into optimization, generalization, ...
Related: #Curriculum Design -
🇺🇸 Mastering Negation: Boosting Grounding Models via Grouped Opposition-Based Learning
[USA]
arXiv:2603.12606v1 Announce Type: cross Abstract: Current vision-language detection and grounding models predominantly focus on prompts with positive semantics and often struggle to accurately interp...
Related: #Negation Understanding -
🇺🇸 AI companies want to harvest improv actors’ skills to train AI on human emotion
[USA]
If you've got strong creative instincts, the ability to authentically portray emotion, and are capable of staying true to a character's voice throughout a scene, there's a job listing calling for your...
Related: #Creative Labor -
🇺🇸 Causal Prosody Mediation for Text-to-Speech:Counterfactual Training of Duration, Pitch, and Energy in FastSpeech2
[USA]
arXiv:2603.11683v1 Announce Type: cross Abstract: We propose a novel causal prosody mediation framework for expressive text-to-speech (TTS) synthesis. Our approach augments the FastSpeech2 architectu...
Related: #Speech Synthesis -
🇺🇸 Learning Transferable Sensor Models via Language-Informed Pretraining
[USA]
arXiv:2603.11950v1 Announce Type: new Abstract: Modern sensing systems generate large volumes of unlabeled multivariate time-series data. This abundance of unlabeled data makes self-supervised learni...
Related: #Sensor Models -
🇺🇸 Entropy-Preserving Reinforcement Learning
[USA]
arXiv:2603.11682v1 Announce Type: cross Abstract: Policy gradient algorithms have driven many recent advancements in language model reasoning. An appealing property is their ability to learn from exp...
Related: #Reinforcement Learning -
🇺🇸 Thousand-GPU Large-Scale Training and Optimization Recipe for AI-Native Cloud Embodied Intelligence Infrastructure
[USA]
arXiv:2603.11101v1 Announce Type: cross Abstract: Embodied intelligence is a key step towards Artificial General Intelligence (AGI), yet its development faces multiple challenges including data, fram...
Related: #Cloud Infrastructure -
🇺🇸 When Fine-Tuning Fails and when it Generalises: Role of Data Diversity and Mixed Training in LLM-based TTS
[USA]
arXiv:2603.10904v1 Announce Type: cross Abstract: Large language models are increasingly adopted as semantic backbones for neural text-to-speech systems. However, frozen LLM representations are insuf...
Related: #Speech Synthesis -
🇺🇸 The Curse and Blessing of Mean Bias in FP4-Quantized LLM Training
[USA]
arXiv:2603.10444v1 Announce Type: cross Abstract: Large language models trained on natural language exhibit pronounced anisotropy: a small number of directions concentrate disproportionate energy, wh...
Related: #Quantization -
🇺🇸 IH-Challenge: A Training Dataset to Improve Instruction Hierarchy on Frontier LLMs
[USA]
arXiv:2603.10521v1 Announce Type: new Abstract: Instruction hierarchy (IH) defines how LLMs prioritize system, developer, user, and tool instructions under conflict, providing a concrete, trust-order...
Related: #LLM Development -
🇺🇸 Training Language Models via Neural Cellular Automata
[USA]
arXiv:2603.10055v1 Announce Type: cross Abstract: Pre-training is crucial for large language models (LLMs), as it is when most representations and capabilities are acquired. However, natural language...
Related: #Computational Models -
🇺🇸 Grounding Synthetic Data Generation With Vision and Language Models
[USA]
arXiv:2603.09625v1 Announce Type: cross Abstract: Deep learning models benefit from increasing data diversity and volume, motivating synthetic data augmentation to improve existing datasets. However,...
Related: #Data Generation -
🇺🇸 ActiveUltraFeedback: Efficient Preference Data Generation using Active Learning
[USA]
arXiv:2603.09692v1 Announce Type: cross Abstract: Reinforcement Learning from Human Feedback (RLHF) has become the standard for aligning Large Language Models (LLMs), yet its efficacy is bottlenecked...
Related: #Data Efficiency -
🇺🇸 DexHiL: A Human-in-the-Loop Framework for Vision-Language-Action Model Post-Training in Dexterous Manipulation
[USA]
arXiv:2603.09121v1 Announce Type: cross Abstract: While Vision-Language-Action (VLA) models have demonstrated promising generalization capabilities in robotic manipulation, deploying them on specific...
Related: #Robotics, #Human-in-the-Loop -
🇺🇸 Grouter: Decoupling Routing from Representation for Accelerated MoE Training
[USA]
arXiv:2603.06626v1 Announce Type: cross Abstract: Traditional Mixture-of-Experts (MoE) training typically proceeds without any structural priors, effectively requiring the model to simultaneously tra...
Related: #Model Efficiency -
🇺🇸 Boosting deep Reinforcement Learning using pretraining with Logical Options
[USA]
arXiv:2603.06565v1 Announce Type: new Abstract: Deep reinforcement learning agents are often misaligned, as they over-exploit early reward signals. Recently, several symbolic approaches have addresse...
Related: #Reinforcement Learning -
🇺🇸 When Rubrics Fail: Error Enumeration as Reward in Reference-Free RL Post-Training for Virtual Try-On
[USA]
arXiv:2603.05659v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) and Rubrics as Rewards (RaR) have driven strong gains in domains with clear correctness signals...
Related: #Virtual Try-On
Key Entities (8)
- Machine learning (2 news)
- Mixture of experts (1 news)
- Reinforcement learning (1 news)
- Artificial intelligence (1 news)
- Graphics processing unit (1 news)
- OpenAI (1 news)
- TTS (1 news)
- Large language model (1 news)
About the topic: AI Training
The topic "AI Training" aggregates 17+ news articles from various countries.