specialization beats scale in ai procurement
a 3b specialized model outperformed frontier apis at lower cost, showing training alignment matters more than parameter count.
aisummaries filed under research
a 3b specialized model outperformed frontier apis at lower cost, showing training alignment matters more than parameter count.
aia small fraction of autoregressive model requests can loop into repetition, inflating batch inference time by over 40% and exposing a blind spot in standard benchmarks.
aia new study extends fastkan by using leave-one-out cross-validation to set kernel shape parameters and adding matern and wendland kernels for more flexible function approximation.
aia new method trains language models to predict which research ideas will work better without running experiments, using a dataset of 11,488 idea pairs from paperswithcode.
ainew bounds show how finite-width neural network outputs stay close to their infinite-width mean-field limit over all training time, without needing strong convexity or noise.
aia new framework helps ad platforms decide which reserve-price policies are worth testing, using logged data without overpromising gains.
aia new method selects covariates for causal effect estimation without requiring pretreatment or causal sufficiency assumptions, using local learning to avoid global structure search.
aia new proof shows that feature rankings cannot be simultaneously faithful, stable, and complete when features are correlated, and proposes an ensemble method as a solution.
aia curated look at top small language models under 7b parameters on hugging face, with benchmarks and code to get started.
aia neural estimator learns pairwise conditional mutual information from pretrained masked diffusion models, enabling faster parallel decoding by identifying independent variable subsets.
aia new governance layer called lbw-guard improves training stability and reduces perplexity in language models under stress conditions.
aia new transformer variant fixes training instability in looped models by distributing signals across layers and reusing attention states, enabling deeper iteration without extra parameters.
ainew research provides statistical learning bounds for using machine learning to predict lagrangian multipliers in mixed integer linear programming, showing generalization scales with problem size and sample count.
aihellora fine-tunes mixture-of-experts models by adding low-rank adapters only to the most active experts, cutting parameters and compute while boosting performance.
aia new neural network processes tensor data by splitting it into low-rank and sparse parts, with proven error bounds and automatic structure selection.
aia new framework uses beta distributions to measure calibration-conditional coverage in conformal prediction, providing bounds on coverage gaps and bad-calibration probabilities under non-i.i.d. settings.
aigoogle deepmind's co-scientist helps biologists find genetic factors that reverse cellular aging, cutting analysis time from months to days.
aistudy tests gemini 3.0 flash on 2,257 patient queries using personal health records, finding better accuracy with full clinical notes.
aia calibration-first router uses isotonic regression to map token uncertainty to error probability, enabling cost-optimal escalation thresholds without per-workload tuning.
aigoogle deepmind's project genie now uses street view imagery to create interactive virtual worlds based on real locations, available to google ai ultra subscribers.
ai