Loading…
Topic
617 posts about Machine Learning. Every summary links to the original.
See the idea
A walkthrough of how a 3D CT model turns a voxel volume into a structured diagnostic report: isotropic resampling, 3D patch tokens, spatial encoding, and language-model reasoning.
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Newest first
Anthropic is committing $10 million CAD to Canadian research institutions to support beneficial and responsible AI applications, building on Canada’s contributions to neural networks, deep learning, and reinforcement learning. The announcement names partnerships with Amii, Mila, Vector, CHEO, CAMH, Université Laval, University of Toronto, and the University of Saskatchewan, with Claude credits supporting work spanning trust and safety, health, scientific discovery, mental health, robotics, sustainability, and public services. Amii, Mila, and Vector will also join Anthropic’s for Startups program, providing affiliated Canadian startups with at least $5,000 USD each in API credits. A Canadian country brief based on the March 2026 Anthropic Economic Index reports that Canada ranks eighth in Claude.ai use, accounts for 2.6% of global consumer use, and ranks second by per-capita adoption; usage is highest in British Columbia and closely follows professional, scientific, and technical employment.
Existing voice-AI benchmarks increasingly suggest near-human performance, while real-world conversations still expose failures involving speaker consistency, hesitation, accents, noise, emotion, and natural responses. Real World VoiceEQ measures human quality in voice interaction by testing whether systems recognize, produce, and respond to acoustic information omitted by transcripts, including tone, emotion, speaker identity, and background context. It evaluates more than 40 proprietary and open-source models across 15+ dimensions and 60+ metrics covering ASR, TTS, S2S, and Speech Understanding, using more than 1 million human ratings, including 785,000 TTS and 48,000 STS ratings. Findings indicate that capabilities are specialized, speech-to-speech systems often underuse paralinguistic cues, and traditional benchmarks can overestimate performance in noise and other conversational conditions. The research also reports that human agreement remains essential for subjective acoustic and social judgments because SLM evaluators were less reliable outside clear, verifiable tasks.
The NVIDIA Nemotron Model Reasoning Challenge on Kaggle tasked over 5,000 participants with improving reasoning accuracy on Nemotron-3-Nano-30B using restricted LoRA adapters on Google Cloud G4 VMs. Top-performing competitors approached reasoning as an end-to-end engineering workflow by auditing synthetic chain-of-thought traces with programmatic solvers before training. Successful teams also compressed reasoning representations into compact signatures and bit-manipulation formats to prevent long chains from exceeding runtime token generation limits. Additionally, leading approaches separated precomputed reusable structures from live computation and conducted granular per-category validation to monitor regression and non-determinism. These results show that reasoning improvements stem from verifiable training data, compact context encoding, and structured validation rather than unconstrained data scaling.
Airbnb built a four-layer LLM infrastructure framework to reduce iteration and evaluation turnaround from weeks to a single day. By using per-sample caching for generated references and judge scores, they established a deterministic evaluation foundation that separates genuine model drift from measurement noise. This setup enables rapid, bounded model hotfixes using micro LoRA adapters and comprehensive end-to-end validation across component boundaries.
Setting up reinforcement learning workflows and running repetitive experimentation often consumes significant engineering time before meaningful metrics emerge. To address this, an autonomous autoresearch workflow uses a frontier coding agent running Codex with GPT 5.5 to manage NVIDIA NeMo RL and NeMo Gym on an NVIDIA Brev GPU instance. The agent executes full-stack setup, creates a novel NeMo Gym visual counting environment, tunes Qwen3-VL-2B-Instruct accuracy from 25.0% to 96.9%, and implements an off-policy reinforcement learning algorithm directly from a research paper. Reusable skills, including brev-etiquette, session-memory, and autoresearch, enforce local conventions and prevent context drift during long executions. This setup offloads environment configuration, dependency resolution, and experiment tracking while preserving human oversight for goals and strategy.
Adapting vision reasoning models for specialized physical AI video tasks often demands days of manual container configuration, training script authoring, and hyperparameter tuning. NVIDIA Cosmos 3 Nano, an omnimodal mixture-of-transformers model, pairs an autoregressive Reasoner tower with an iterative diffusion Generator tower for video question answering. By integrating NVIDIA TAO agent skills with a coding agent like Codex, developers can automate data patching, evaluation, and Low-Rank Adaptation (LoRA) post-training. In experiments using the Woven Traffic Safety dataset, LoRA fine-tuning boosted zero-shot baseline accuracy from 54.41% to 87.14% while using approximately seven times fewer GPU hours than full-parameter supervised fine-tuning. Combining this agent workflow with TAO AutoML further elevated model accuracy to 93.35% before automated deployment via NVIDIA NIM.
Serving Qwen3.5-397B, a 397-billion-parameter sparse Mixture-of-Experts model, on Ironwood (TPU v7x) requires fitting roughly 400 GB of weights into HBM while sustaining utilization across contrasting workloads. The report applies a modular, model-agnostic playbook built from reusable JAX/Pallas kernels and hardware-aware cost models, focusing new work on Gated DeltaNet linear attention and Attention Data Parallelism while integrating changes into vLLM and SGLang. The architecture activates exactly 17 billion parameters per token and combines 15 repeating blocks of three GDN layers and one GQA layer. Benchmarks use a single host with four Ironwood chips, eight logical devices, 8K/1K prefill-heavy and 1K/8K decode-heavy workloads, and concurrency tiers from 64 to 512. Between April and June 2026, stated optimizations improved inference performance by approximately 4.7x for 512-concurrency prefill-heavy workloads and 3.1x for decode-heavy workloads.
Yelp developed Training Orchestrator to standardize machine learning pipeline execution and eliminate divergent training scripts across its applied engineering teams. Previously, monolithic Spark scripts coupled training logic directly to cluster runtimes, preventing local testing, slowing debug cycles, and causing inconsistent configuration drift across projects. The new framework introduces a declarative configuration layer backed by Pydantic models, enabling static type checking and step schema validation before runtime execution starts. Training workflows are defined as modular step hierarchies that the orchestration engine compiles into directed acyclic graphs for execution in topological order. By injecting Spark and MLflow contexts into steps, the architecture ensures reproducible tracking, automated Slack alerts, and parity across local development and production environments.
Fault-tolerant quantum computing requires effective quantum error correction, but color codes have historically been constrained by difficult decoding demands. NVIDIA introduced the Ising Decoder ColorCode 1 Fast, a pre-decoder designed to accelerate and improve color code logical error rates. The framework uses a 3D Convolutional Neural Network trained via cuStabilizer in NVIDIA cuQuantum alongside PyTorch to handle localized error syndromes. In benchmark testing at code distance d=31 with a 0.3% physical error rate, the decoder achieved over 347.7x lower logical error rates and 7.3x faster runtime compared to the Chromobius decoder. These local, space-time predictions scale to arbitrary distances, supporting parallel blockwise decoding for real-time quantum algorithm execution.
Estimating the likelihood of extreme, high-impact events using brute-force Monte Carlo simulations requires excessive model iterations for expensive physics-based climate models. While guided diffusion models steer generation toward rare events like tropical cyclones, oversampling distorts the true probability distribution. To address this, researchers guide the diffusion-based climate emulator NVIDIA cBottle toward specific events and compute log-odds ratios between guided and unguided distributions. Reweighting the guided samples via these odds ratios enables importance sampling under the original climate distribution. Implemented in NVIDIA Earth2Studio, this technique reduced standard error by 25% compared to simple Monte Carlo sampling when evaluating tropical cyclone states.
Evaluating robotics foundation models accurately in simulation remains challenging due to visual domain overlap, benchmark saturation, diagnostic gaps, and small sample sizes. RoboLab addresses these deficiencies by providing a robot-agnostic simulation platform that supports rapid task creation and diagnostic evaluation suites. Tasks in RoboLab can be compiled across varying robot embodiments and target visual, procedural, and relational competencies. Statistical analysis using the Clopper-Pearson method reveals that narrowing the confidence interval around a 90 percent success rate from ten to two percentage points demands increasing rollouts from 70 to 1,030. Developed by NVIDIA Research, RoboLab powers the open-source Isaac Lab-Arena simulation framework to establish scalable evaluation for real-world robotic policies.
Large language model training workloads frequently face GPU high-bandwidth memory (HBM) capacity constraints before compute limits are reached. To alleviate this memory pressure in JAX training workloads, host offloading transfers selected forward-pass activations to pinned host memory over high-speed NVLink-C2C interconnects and reloads them during the backward pass instead of recomputing them. Benchmarks executed on 128 GPUs within NVIDIA GB200 NVL72 systems using MaxText demonstrated significant improvements. For DeepSeek-V3 671B, host offloading combined with the Latency Hiding Scheduler and pipelined offloading reached 908.2 TFLOPs/s/device, surpassing activation rematerialization throughput by 57% and unlocking larger batch configurations. Similarly, QKV activation offloading on Llama 3.1 405B achieved a 2.9% throughput increase when transfer latency was hidden behind computation.
amazon.science
Robotic manipulation in contact-rich settings requires accurate tactile force perception, but existing physics simulators either run too slowly for reinforcement learning or oversimplify shear forces. HydroShear addresses this tactile reality gap by incorporating path-dependent force tracking into hydroelastic contact models. By remembering object motion history across soft elastomer sensors, the GPU-parallelizable simulator computes full 3-D contact deformations, friction, and slipping dynamics. Reinforcement learning policies trained entirely in HydroShear transferred directly to a real Franka robot equipped with GelSight Mini sensors without fine-tuning. Across four challenging manipulation tasks, HydroShear achieved a 93 percent average success rate, outperforming baseline models TacSL and FOTS.
Efficient large language model serving requires co-designing architectures to balance datacenter token throughput and user interactivity within hardware constraints. Roofline analysis shows that linear layers become memory-bound when hidden or intermediate projection dimensions are small, leaving arithmetic hardware underutilized even at large token counts. For prefill workloads, Chunked Pipeline Parallelism reduces first-token latency on long contexts when models use regular, repeatable layer patterns that split into balanced stages. For latency-critical decoding, decoupling attention and feed-forward network parallelization enables independent optimization, applying expert parallelism for feed-forward networks and Helix parallelism to shard the key-value cache across sequences. Following architectural design rules—such as dimension alignment, width prioritization, and low-precision NVFP4 compatibility—allows models to achieve higher utilization on modern accelerator hardware.
Biomolecular structure prediction and co-folding pipelines encounter latency and memory ceilings during sequence alignment and multi-residue inference. NVIDIA addresses these constraints through the BioNeMo Agent Toolkit, which connects AI agents to accelerated biological computing components. The pipeline offloads multiple sequence alignment to MMseqs2-GPU, while cuEquivariance and the OpenFold3 NIM accelerate core geometric attention kernels to cut inference latency up to fourfold. To handle large molecular assemblies beyond single-GPU memory limits, the Fold-CP framework distributes context across multiple GPUs with O(N²/P) memory scaling. Together, these tools enable the prediction of complexes up to 32,000 tokens on 64 NVIDIA B300 GPUs, making large-scale virtual screening and ribosome-scale modeling computationally feasible.
Golden Gate Claude presents a research demonstration of interpretability work on Claude 3 Sonnet. Researchers identified millions of activated concepts, or “features,” including a specific combination of neurons associated with the Golden Gate Bridge, and adjusted that feature’s activation strength. Increasing it caused responses to mention the bridge across unrelated prompts, such as spending $10 or writing a love story; the demonstration was available online for 24 hours and is no longer available. The post distinguishes this intervention from verbal role-play, system prompts, and traditional fine-tuning, describing it as a precise change to internal activations. It also says similar techniques could modify safety-related features connected to dangerous code, criminal activity, or deception, potentially supporting safer models.
Real-world financial natural language processing datasets often overrepresent common events like earnings while underrepresenting rarer occurrences such as credit-rating changes. To address this imbalance, an iterative generation pipeline produced over 500,000 unique headlines across 13 categories using NVIDIA NeMo Data Designer, NeMo Curator, and Nemotron 3 Nano. Instead of relying on a single large batch that yielded 65 percent near-duplicates, the workflow executes iterative cycles of category-weighted generation, rule-based filtering, global semantic deduplication, and dynamic category reweighting. Farthest-from-centroid few-shot example selection guided subsequent rounds toward novel semantic spaces across 82 total iterations. Executed over approximately six days on an eight-way NVIDIA B200 node, the resulting corpus enabled fine-tuning compact student language models to achieve 95 percent of teacher model performance.
amazon.science
Reinforcement learning for agentic tasks requires updating model parameters against the precise token sequences generated during rollouts. Standard agent harnesses frequently modify conversation transcripts through message compaction and tool formatting, causing retokenization drift that degrades policy-gradient training. Turnstile addresses this issue as a Rust-based proxy situated between agent harnesses and inference backends such as SGLang. By intercepting OpenAI-compatible Chat Completions traffic, the proxy records exact token IDs, log probabilities, loss masks, weight versions, and mixture-of-experts routing traces at the moment of generation. Validation across text-based coding and multimodal computer-use agents demonstrated steady reinforcement learning improvements without requiring modifications to agent harnesses.
Anthropic argues that AI could have an impact comparable to the industrial and scientific revolutions, potentially within the coming decade, because scaling laws link increased training computation with predictable capability improvements. It says progress may continue toward systems matching or exceeding human performance across many intellectual tasks, while society faces disruption and competitive pressure to deploy systems that are not reliably helpful, honest, and harmless. The company’s safety strategy is empirically driven and multifaceted, emphasizing scalable supervision, mechanistic interpretability, process-oriented learning, and research into how systems learn and generalize. It also studies capability evaluation, red teaming, harmful behavior, economic effects, and policy implications. The stated goal is a broad portfolio of safety work that can adapt as evidence clarifies which future scenario is unfolding, including less optimistic possibilities.
Uber describes how it uses early chargeback signals, especially TC40 reports, to identify payment-fraud patterns before conventional chargebacks arrive. These notifications come from payment service providers through fraud reports and are ingested from external vendors via SFTP, third-party APIs, or webhooks, then validated, normalized, published to Apache Kafka, and stored in Apache Hive. A streaming pipeline maps signals to internal transactions and users, derives near-real-time features, and sends them to machine-learning models and a rule engine that determines risk actions. Redis TTL caching filters duplicate references, while Kafka propagates signals to downstream systems and Hive retains them for batch processing. Since launch, the project has processed more than 2 million additional signals, and TC40 signals arrive on average 4–5 days earlier than chargeback signals, although the source notes that TC40 data alone does not prevent chargebacks.
The best engineering reads, once a week.
