Top AI Infrastructure This Month

Looking for opinion of people in the industry. [D]

I am researching about AI infrastructure and would value someone's perspective who is close to enterprise AI deployment. At a high level, we are seeing more often: as enterprises move from copilots...

Project Aurelia — A 3-model architecture (80B + 13B + 9B) that physically reacts to my real-time heart rate via mmWave radar, spatial awareness via Lidar, and Vibration via Accelerometer. All on a Framework Desktop + eGPU

Hey everyone, I’ve been building a multi-agent system in my spare time, and I just open-sourced the repository. I was getting tired of the standard text-in/text-out chat paradigm and wanted to buil...

[2604.10423] Replicable Composition

Abstract page for arXiv paper 2604.10423: Replicable Composition

Siemens, NVIDIA hit chip verification milestone for AI

Google DeepMind just published the strongest argument I’ve read against AI consciousness. And they’re right on the core point, with one critical gap.

Their paper, The Abstraction Fallacy, shows that symbolic computation cannot instantiate consciousness because symbols require an external “mapmaker” to assign semantic content. No matter how compl...

How are they able to charge ~50% less than Lovable if they’re using the same models?

Hey everyone, I’ve been using tools like Lovable, Antigravity, and Claude Code for a while now, and after some time it all started to feel a bit repetitive (same kind of outputs, similar templates,...

[2604.11087] CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models

Abstract page for arXiv paper 2604.11087: CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models

[2604.11305] Beyond Fixed False Discovery Rates: Post-Hoc Conformal Selection with E-Variables

Abstract page for arXiv paper 2604.11305: Beyond Fixed False Discovery Rates: Post-Hoc Conformal Selection with E-Variables

[2604.10672] One-Step Score-Based Density Ratio Estimation

Abstract page for arXiv paper 2604.10672: One-Step Score-Based Density Ratio Estimation

Trained a Qwen2.5-0.5B-Instruct bf16 model on Reddit post summarization task with GRPO [P]

So, a few days back I shared a post where I trained a tiny Qwen2.5-0.5B-Instruct model on smoltldr (reddit post summarization dataset of 2k rows), to output summaries of about 64 max length using R...

[2601.08039] Riemannian Zeroth-Order Gradient Estimation with Structure-Preserving Metrics for Geodesically Incomplete Manifolds

Abstract page for arXiv paper 2601.08039: Riemannian Zeroth-Order Gradient Estimation with Structure-Preserving Metrics for Geodesically Incomplete Manifolds

[2601.15498] MARS: Unleashing the Power of Speculative Decoding via Margin-Aware Verification

Abstract page for arXiv paper 2601.15498: MARS: Unleashing the Power of Speculative Decoding via Margin-Aware Verification

[2603.18492] AIMER: Calibration-Free Task-Agnostic MoE Pruning

Abstract page for arXiv paper 2603.18492: AIMER: Calibration-Free Task-Agnostic MoE Pruning

[2604.21469] Cross-Domain Data Selection and Augmentation for Automatic Compliance Detection

Abstract page for arXiv paper 2604.21469: Cross-Domain Data Selection and Augmentation for Automatic Compliance Detection

[2412.11390] PAT: Privacy-Preserving Adversarial Transfer for Accurate, Robust and Privacy-Preserving EEG Decoding

Abstract page for arXiv paper 2412.11390: PAT: Privacy-Preserving Adversarial Transfer for Accurate, Robust and Privacy-Preserving EEG Decoding

[2507.07067] How to Bridge the Sim-to-Real Gap in Digital Twin-Aided Telecommunication Networks

Abstract page for arXiv paper 2507.07067: How to Bridge the Sim-to-Real Gap in Digital Twin-Aided Telecommunication Networks

[2604.21728] Ramen: Robust Test-Time Adaptation of Vision-Language Models with Active Sample Selection

Abstract page for arXiv paper 2604.21728: Ramen: Robust Test-Time Adaptation of Vision-Language Models with Active Sample Selection

[2605.07068] WiCER: Wiki-memory Compile, Evaluate, Refine Iterative Knowledge Compilation for LLM Wiki Systems

Abstract page for arXiv paper 2605.07068: WiCER: Wiki-memory Compile, Evaluate, Refine Iterative Knowledge Compilation for LLM Wiki Systems

[2604.21870] Locating acts of mechanistic reasoning in student team conversations with mechanistic machine learning

Abstract page for arXiv paper 2604.21870: Locating acts of mechanistic reasoning in student team conversations with mechanistic machine learning

[2605.07234] Reformulating KV Cache Eviction Problem for Long-Context LLM Inference

Abstract page for arXiv paper 2605.07234: Reformulating KV Cache Eviction Problem for Long-Context LLM Inference

[2509.20712] CE-GPPO: Coordinating Entropy via Gradient-Preserving Clipping Policy Optimization in Reinforcement Learning

Abstract page for arXiv paper 2509.20712: CE-GPPO: Coordinating Entropy via Gradient-Preserving Clipping Policy Optimization in Reinforcement Learning

[2510.20064] Not-a-Bandit: Provably No-Regret Drafter Selection in Speculative Decoding for LLMs

Abstract page for arXiv paper 2510.20064: Not-a-Bandit: Provably No-Regret Drafter Selection in Speculative Decoding for LLMs

[2411.14748] Cosmological Analysis with Calibrated Neural Quantile Estimation and Approximate Simulators

Abstract page for arXiv paper 2411.14748: Cosmological Analysis with Calibrated Neural Quantile Estimation and Approximate Simulators

If AI is about to get 10x smarter, how do we prevent the internet from collapsing under synthetic noise?

Im all for acceleration. I think the faster we hit AGI the better. but theres a bottleneck nobody here talks about enough-training data. right now we are quietly poisoning the well. More than half ...

[2512.08216] Tumor-anchored deep feature random forests for out-of-distribution detection in lung cancer segmentation

Abstract page for arXiv paper 2512.08216: Tumor-anchored deep feature random forests for out-of-distribution detection in lung cancer segmentation

[2605.07985] Dooly: Configuration-Agnostic, Redundancy-Aware Profiling for LLM Inference Simulation

Abstract page for arXiv paper 2605.07985: Dooly: Configuration-Agnostic, Redundancy-Aware Profiling for LLM Inference Simulation

[2512.05439] BEAVER: An Efficient Deterministic LLM Verifier

Abstract page for arXiv paper 2512.05439: BEAVER: An Efficient Deterministic LLM Verifier

[2604.27747] Position-Aware Drafting for Inference Acceleration in LLM-Based Generative List-Wise Recommendation

Abstract page for arXiv paper 2604.27747: Position-Aware Drafting for Inference Acceleration in LLM-Based Generative List-Wise Recommendation

[2605.04722] Exact Dual Geometry of SOC-ICNN Value Functions

Abstract page for arXiv paper 2605.04722: Exact Dual Geometry of SOC-ICNN Value Functions

[2605.07631] Inference Time Causal Probing in LLMs

Abstract page for arXiv paper 2605.07631: Inference Time Causal Probing in LLMs