Can Skills Learned in Games Transfer to Real-World Work?
Good Start Labs trained an AI on a railroad game — and one version improved at financial research. The difference was the training design.
#Data and training
#Data and training
0 todaySep 16
Bypassing inference bottlenecks: Accelerating complex AI search with Retrieve-for-Train
Google Research proposed Retrieve-for-Train, using offline reinforcement learning to find reward-aligned query fan-out and compile it into supervision, then distilling it into a 53.9M-parameter diffusion retriever. At inference, it performs one non-autoregressive query fan-out without generating CoT reasoning tokens.
AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories
Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, Tuesday spoke on AI factory efficiency at the AI Infra Summit, the Santa Clara Convention Center event that has morphed into a Coachella of infrastructure tech.
Sep 11
ToolGrad: Efficient tool-use dataset generation with textual "gradients"
At ACL 2026, Google Research proposed ToolGrad, an answer-first, question-later approach that generates real tool-call chains before deriving user queries, replacing traditional DFS trial-and-error annotation.
Sep 10
Now everyone can put data to work
Meet the Data agent in ChatGPT Work. Connect company data, uncover insights, and build interactive dashboards with AI using natural language.
Async GRPO with LoRA across HF Jobs: a bucket, a proxy, and no NCCL
TRL v1.14's AsyncGRPOTrainer now supports training and synchronising only LoRA adapters to vLLM. A rank-1 adapter is just a few MB and can transfer through a Storage Bucket mounted to each job, without NCCL.
Sep 8
AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome
AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.
Sep 4
Transfer learning for genomic prediction in underrepresented populations
Google Research evaluated cross-population transfer of polygenic risk scores (PRS) on eight clinical measures using European UK Biobank data and nearly 200,000 Japanese participants from Biobank Japan. At target-population samples of 15,000 or more, target-specific models outperformed mixed European-data training, while highly genetically correlated traits continued to benefit from European data until target samples reached 25,000–40,000 or more.
A connectomics milestone: Mapping the complete male fruit fly brain
Google Research partnered with HHMI Janelia, the University of Cambridge and others to publish in Cell a complete connectome of a male fruit fly’s brain and central nervous system. It contains over 166,000 neurons and 125 million synaptic connections, making it the largest brain map to date by neuron count.
Sep 3
Fine-tuning a 350M Model for Better Structured Outputs in 100 GRPO Steps
A public low-cost approach fine-tunes LFM2.5-350M with TRL GRPO using around 500 samples and 100 steps, raising IFStruct from 22.6% to 29.7%. Training fits free Colab or Kaggle GPUs, with local llama.cpp evaluation on a MacBook and code open on GitHub.
Sep 2
Mapping global methane emissions from space with deep learning
Google Research proposed MAPL-EMIT, a Swin-S Transformer-based deep-learning framework automatically detecting methane plumes, predicting enhancements and locating them in NASA EMIT hyperspectral satellite data, with 84% recall on expert-labelled plumes.
Sep 1
TimesFM-3: A zero-shot foundation model for multivariate forecasting
Google Research released TimesFM-3, its first time-series foundation model pretrained natively for multivariate forecasting, with 330 million parameters and over 1 trillion pretraining time points.
GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models
What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration.
Aug 28
Planetary prediction engine: Automating global models via Earth AI
Google Research introduced the experimental Planetary Prediction Engine (PPE) under Google Earth AI. From a natural language query, it autonomously discovers geospatial data, engineers features, trains and evaluates models and produces reports, compressing weeks of manual data engineering into minutes.
Aug 27
GlucoFM: Foundation model for continuous glucose monitoring
Google released self-supervised GlucoFM, separating slow glucose trends from short-term fluctuations. Pretrained on 109,066 hours of unlabelled CGM data, it averaged 5.8 percentage points higher PR-AUC than the best GluFormer variant across 14 evaluations on seven clinical tasks in four cohorts, and achieved the lowest post-meal response prediction MAE.
Aug 26
Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers
Sentence Transformers v6.0 adds a fourth model type, MultiVectorEncoder, for ColBERT-style late-interaction retrieval, with a complete training approach.
Aug 25
Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original
Multiverse Computing published a paper proposing Quantization-Aware Healing (QAH). After compressing GPT-OSS 120B to 60B parameters and quantising it to MXFP4, the method distils directly from the original uncompressed model rather than a reconstructed bfloat16 checkpoint.
Aug 24
Import AI 470: No rights for machines; automating environment generation with SPADE; and building better GPU kernels with Hawkeye
Differential acceleration of cyber, math, and AI
Aug 22
An AI tool for prioritizing candidate biomarkers from wearable sensor data
Google introduced Biomarker Discovery Framework, a human-supervised multi-agent system organising candidate biomarker prioritisation into iterative research cycles. Across 9,279 participant observations in three cohorts, it automatically identified 41 candidate digital mental health biomarkers and 25 metabolic candidates, including an association between sleep-duration variability and PHQ-8 severity (ρ = 0.252).
Aug 21
How mobility gives language models a deeper understanding of place
Google Research proposed Mobility-Embedded POIs (ME-POIs), combining anonymised mobility patterns with text descriptions to generate place embeddings encoding both identity and dynamic function.
Broadening access to Skala creates a faster path to predictive DFT
Skala 1.1, the updated deep-learning exchange-correlation functional from Microsoft Research, provides greater accuracy, expanded accessibility across the computational chemistry ecosystem, and a living benchmark to track computational performance.
Aug 19
How Much Memory Does Your Agent Actually Need?
IBM Research published ALTK-Evolve research on the Hugging Face blog. Tests of eight models on 585 multi-step AppWorld tasks found agent memory is not an on/off switch but a dosage requiring model-specific calibration.
Aug 18
Same Cluster, 33 Points More Utilization: What Changed Was the Order
A Hugging Face blog describes a constraint-aware GPU allocator compared with FIFO under identical hardware and workloads. Across five real contention scenarios, GPU utilisation rose from 52–85% to 72–88%, while priority-weighted output improved 24.6–105.1%, averaging 52%.
Aug 17
Import AI 469: Science AI; RSI simulator; and Zuck's technological pessimism
The new frontier of AI is developing capable autonomous researchers
Aug 14
State of Open Models: Summer 2026 Observations
Hugging Face published its open-source model observatory report for January–August 2026. Hub data shows Chinese labs released the largest open-source models by parameter count in most months, with Chinese monthly peaks between 754 billion and 2.78 trillion parameters, while US models stayed below 130 billion in five of seven months.
Aug 13
Introducing OlmoEarth embeddings: Custom embedding exports from OlmoEarth Studio for downstream analysis
Hugging Face's OlmoEarth Studio now computes and exports embeddings. Users choose an area, period, encoder variant and resolution through UI or API and receive Cloud-Optimized GeoTIFF (COG) results.
Aug 12
Empty shelves or lost keys? Recall is the bottleneck for parametric factuality
Google Research's knowledge profiling framework evaluated 13 LLMs on WikiProfile's 2,150 Wikipedia facts. Gemini-3-Pro and GPT-5 encoded 95–98% but failed direct recall of 26–34%, and still failed 11–12% with thinking enabled. It argues frontier factual errors arise more from knowledge accessibility than absence, shifting the bottleneck from acquisition to use.
Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
Radiology AI is evolving beyond report generation. CARE-X explores a unified approach that combines flexible reasoning, calibrated predictions, and measurement-based tools for chest X-ray interpretation.
Aug 10
Making Knowledge Distillation Cheap Enough to Run at Scale
Multiverse Computing published the paper Efficient Knowledge Distillation for LLMs: Offline Top-K Logits and a Fused Chunked KL Loss.
Aug 6
WeatherNext: AI model achieves breakthrough in forecasting cyclones
Google DeepMind released the WeatherNext AI model, achieving state-of-the-art cyclone track, intensity and wind-field structure forecasts and adding an average extra day of warning, equivalent to roughly a decade of meteorological progress.
Jul 31
Echoverse: Deep, evolving environments for computer-use agents
Computer-use AI agents struggle with multi-step workflows like email and customer support. Echoverse trains agents in realistic environments rather than simply providing more training tasks, helping them improve as the tasks, tests, and environments evolve.
Jul 30
GPU Management: Why Idle GPUs Are the New Grounded Aircraft
Hugging Face's blog argues AI's next real constraint is GPU utilisation, rather than intelligence. GPUs are billed by calendar hours but produce only in compute hours, mirroring the cost structure of grounded aircraft.
Jul 26
Teaching LLMs to Update Beliefs for Efficient Long-Horizon Interaction
Overview of ABBEL compared to traditional recursive summarization. Beliefs replace the full interaction history as the agent’s working context, and belief grading improves performance by supervising the contents of each belief state.. As task horizons grow, LLM contexts can’t scale forever.
Jul 22
Accelerating the frontiers of scientific discovery: Google’s $40M commitment to the Genesis Mission
Google commits $40M in AI tokens and credits for the Genesis Mission
Jul 16
Newer Models, Same Advantage
DharmaOCR scored 0.925 on a Portuguese OCR benchmark, ahead of Mistral OCR4's 0.798 and Unlimited-OCR's 0.7587. It specialised through two-stage training: supervised fine-tuning on Portuguese corpora, then DPO to stabilise inference. The author argues that concentrating parameters on one language remains a structural advantage despite emerging architectures.
Towards demystifying the creativity of diffusion models
A Google Research ICLR 2026 paper traces diffusion model creativity to score smoothing from regularisation such as weight decay. Smoother learned score functions let denoising interpolate between training points, generating new samples rather than simply memorising them.
Jul 9
SensorFM: Towards a general intelligence and interface for wearable health data
Google Research and Google DeepMind proposed SensorFM, a large sensor foundation model learning directly from unlabelled wearable data. Pretraining uses over 1 trillion minutes of multimodal sensor signals from 5 million consenting participants across more than 100 countries and over 20 Fitbit and Pixel Watch devices.
Jul 7
Intelligence is Free, Now What? Data Systems for, of, and by Agents
... government of the people, by the people, for the people ... — Abraham Lincoln, Gettysburg Address (1863) The cost of AI is dropping rapidly. GPT-4-class capabilities cost roughly $30 per million tokens in early 2023; today the same runs under $1 , and some providers are pushing costs below $0.10 .
Jul 6
PRX Part 4: Our Data Strategy
Hugging Face's fourth PRX instalment details its data strategy: mixing public and internal pretraining datasets, regenerating long image captions with a VLM and converting them for streaming training. It uses Lance for construction and filtering and MDS for streaming.
Jul 1
Expanding our Heat Resilience data to 50+ global cities
Google Research expanded building-level roof reflectivity (albedo) data from 14 cities in 2024 to over 50 cities in nine countries, accessible through the new Heat Resilience Earth Engine App.