Raptor
PulseAugur coverage of Raptor — every cluster mentioning Raptor across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New RAPTOR framework enhances private training for MoE AI models
Researchers have developed RAPTOR, a novel framework for differentially private training of Mixture-of-Experts (MoE) models. Existing methods treat these sparse models as dense blocks, leading to issues like gradient su…
-
d-Matrix Inc. unveils Raptor AI accelerator with integrated 3D DRAM
d-Matrix Inc. has developed a new AI inference accelerator called Raptor. This system aims to overcome the limitations of current AI memory architectures by integrating compute capabilities directly with custom 3D DRAM.…
-
Elon Musk's SpaceX to build gas turbine blade foundry for AI power needs
Elon Musk is reportedly establishing a gas turbine blade foundry within SpaceX in Texas to address the significant power shortages anticipated for AI data centers. Musk acknowledged the initiative, stating that the lack…
-
Zhipu AI releases GLM-5.3 Flash, a multimodal model with competitive pricing · 1 source tracked
Zhipu AI has released GLM-5.3 Flash, an open-source, natively multimodal model that has gained significant attention under the alias "Ox Alpha." This new model boasts impressive capabilities, including advanced multimod…
-
d-Matrix unveils 3D DRAM AI accelerator with 100 TB/s bandwidth
d-Matrix has unveiled its Raptor AI accelerator, which features a novel 3D stacking architecture. This design places a TSMC 4nm compute die directly on top of a custom-designed DRAM die, achieving an unprecedented 100 T…
-
AI infrastructure sees memory innovation, price hikes, and cognitive cost concerns
New developments in AI infrastructure are emerging, with d-Matrix announcing Raptor, a 3D DRAM stack promising high bandwidth at significantly lower power consumption than current High Bandwidth Memory (HBM) standards. …
-
AI framework GridCodex enhances power grid code reasoning and compliance
Researchers have developed GridCodex, a novel framework designed to assist the electricity industry with complex grid code reasoning and compliance. This system utilizes large language models and retrieval-augmented gen…
-
New SAPER framework prunes Vision Transformer attention heads for efficiency
Researchers have developed SAPER, a novel framework for pruning attention heads in Vision Transformers. This method uses spectral analysis and visualization techniques based on the Laplacian eigenvectors of attention ma…
-
RAG techniques measured: Corpus-specific testing reveals competition over composition
A recent measurement of retrieval-augmented generation (RAG) techniques on a corpus of 46,000 chunks revealed that many advertised methods compete rather than compose, and their effectiveness is highly dependent on the …
-
New RAG frameworks tackle reliability and accuracy challenges · 2 sources tracked
Two new research papers introduce advanced frameworks for retrieval-augmented generation (RAG) systems, aiming to improve reliability and accuracy in agentic AI. LayerRAG-Bench proposes a benchmark to evaluate RAG syste…
-
AI Rocket Engine Startup LegendSpace Secures $29.5M Angel Round
LegendSpace, a Chinese startup focused on AI-driven rocket engine development, has secured $29.5 million in angel funding. The company utilizes its proprietary "Prime" platform, which integrates AI design, simulation, a…
-
SpaceX aborts Starship test flight due to engine ignition failure
SpaceX's 13th Starship test flight was aborted just before liftoff due to a failure in multiple Raptor engines. CEO Elon Musk confirmed the issue, stating that two engines did not ignite as expected, triggering an autom…
-
New RAPTOR method enhances temporal knowledge graph reasoning efficiency
Researchers have developed RAPTOR, a novel pretraining method designed to improve the efficiency of temporal knowledge graph (TKG) reasoning. This self-supervised approach injects a reachability-aware inductive bias int…
-
DeepMind poker AI trio's startup EquiLibre raises Series A for trading AI
EquiLibre Technologies, a startup founded by three former DeepMind researchers, has secured Series A funding led by Creandum. The company leverages AI, specifically reinforcement learning techniques similar to those use…
-
LLM-guided planning system boosts accuracy on nuclear regulatory documents
Researchers have developed an LLM-guided planning system designed to improve multi-hop reasoning over complex nuclear regulatory documents. This system frames the task as a planning problem, where an LLM agent navigates…
-
AI Engineering Evolves: Harness and Loop Concepts Emerge
The article discusses the evolution of AI engineering terminology, moving beyond prompt and context engineering to include "harness engineering" and "loop engineering" as coined by human developers in 2026. Harness engi…
-
Developer open-sources SEQUOIA RAG system outperforming 7 pipelines
A developer has developed and open-sourced a Retrieval-Augmented Generation (RAG) system called SEQUOIA, which consistently outperformed seven other RAG configurations in benchmarks using real-world banking documents an…
-
New pruning technique speeds up public transport routing by 57%
Researchers have developed a new technique called Early Pruning to significantly speed up public transport routing algorithms like RAPTOR. This method optimizes transfer calculations by pre-sorting connections and disca…
-
New VLM Framework Enhances Clinical Cancer Referral Processing
Researchers have developed RAPTOR+, a multimodal framework utilizing Vision-Language Models (VLMs) to enhance the processing of clinical cancer referrals. This system aims to improve trust and auditability by directly l…
-
New Dijkstra algorithm optimizes public transit routing
Researchers have developed a new algorithm called Transfer Aware Dijkstra (TAD) to improve pathfinding in public transit systems. This algorithm correctly accounts for buffer times at stops, which is crucial for transfe…