🎉 Excited to share that our paper "OptiMer: Optimal Distribution Vector Merging Is Better than Data Mixing for Continual Pre-Training" has been accepted to EMNLP 2026 Main! See you in Budapest!
📄 Paper: arxiv.org/abs/2603.28858
#EMNLP2026 #ModelMerging #ContinualPretraining #BayesianOptimization
📄 Paper: arxiv.org/abs/2603.28858
#EMNLP2026 #ModelMerging #ContinualPretraining #BayesianOptimization
OptiMer: Optimal Distribution Vector Merging Is Better than Data Mixing for Continual Pre-Training
Continual pre-training is widely used to adapt LLMs to target languages and domains, yet the mixture ratio of training data remains a sensitive hyperparameter that is expensive to tune: they must be f...
arxiv.org
September 15, 2026 at 5:24 AM
🎉 Excited to share that our paper "OptiMer: Optimal Distribution Vector Merging Is Better than Data Mixing for Continual Pre-Training" has been accepted to EMNLP 2026 Main! See you in Budapest!
📄 Paper: arxiv.org/abs/2603.28858
#EMNLP2026 #ModelMerging #ContinualPretraining #BayesianOptimization
📄 Paper: arxiv.org/abs/2603.28858
#EMNLP2026 #ModelMerging #ContinualPretraining #BayesianOptimization
Checkout Dreams Die - Growth Hacking Stops
https://marketgrid.xyz/checkout-dreams-die-growth-hacking-stops/?utm_source=bluesky&utm_medium=social
#bayesianoptimization #abtesting #checkoutoptimization
https://marketgrid.xyz/checkout-dreams-die-growth-hacking-stops/?utm_source=bluesky&utm_medium=social
#bayesianoptimization #abtesting #checkoutoptimization
August 28, 2026 at 5:20 AM
Checkout Dreams Die - Growth Hacking Stops
https://marketgrid.xyz/checkout-dreams-die-growth-hacking-stops/?utm_source=bluesky&utm_medium=social
#bayesianoptimization #abtesting #checkoutoptimization
https://marketgrid.xyz/checkout-dreams-die-growth-hacking-stops/?utm_source=bluesky&utm_medium=social
#bayesianoptimization #abtesting #checkoutoptimization
Our new Hot Paper 🔥 is out in @angewandtechemie.bsky.social: co-first authors Lucas G. Balzat and Robert Calaminus use #BayesianOptimization to boost Li7SiPS8 ionic conductivity while lowering synthesis temperature and time. #SolidElectrolytes doi.org/10.1002/anie...
Optimize Before You Synthesize—Enhancing the Ionic Conductivity of Li7SiPS8 Using Bayesian Optimization
Bayesian optimization was used to optimize the synthesis parameters of the solid electrolyte Li7SiPS8$\mathrm {Li}_7\mathrm {SiPS}_8$ in order to increase the ionic conductivity. After only 32 iterat...
doi.org
June 1, 2026 at 9:53 AM
Our new Hot Paper 🔥 is out in @angewandtechemie.bsky.social: co-first authors Lucas G. Balzat and Robert Calaminus use #BayesianOptimization to boost Li7SiPS8 ionic conductivity while lowering synthesis temperature and time. #SolidElectrolytes doi.org/10.1002/anie...
25% Cut Speed Breeding Cycle With Process Optimization
https://opscore.site/25-cut-speed-breeding-cycle-process-optimization/?utm_source=bluesky&utm_medium=social
#speedbreeding #bayesianoptimization #photoperiodcontrol
https://opscore.site/25-cut-speed-breeding-cycle-process-optimization/?utm_source=bluesky&utm_medium=social
#speedbreeding #bayesianoptimization #photoperiodcontrol
May 24, 2026 at 3:57 PM
25% Cut Speed Breeding Cycle With Process Optimization
https://opscore.site/25-cut-speed-breeding-cycle-process-optimization/?utm_source=bluesky&utm_medium=social
#speedbreeding #bayesianoptimization #photoperiodcontrol
https://opscore.site/25-cut-speed-breeding-cycle-process-optimization/?utm_source=bluesky&utm_medium=social
#speedbreeding #bayesianoptimization #photoperiodcontrol
Hybrid BOCS-Gaussian Process method with adaptive acquisition function selection resolves search stagnation in discrete optimization. Achieves 49-98% performance improvement over baselines with 49-55% reduction in required evaluations.
#QuantumAnnealing #BayesianOptimization #Research
#QuantumAnnealing #BayesianOptimization #Research
Adaptive Acquisition Functions in Discrete Black-Box Optimization
arxiv.org
May 12, 2026 at 10:19 AM
Hybrid BOCS-Gaussian Process method with adaptive acquisition function selection resolves search stagnation in discrete optimization. Achieves 49-98% performance improvement over baselines with 49-55% reduction in required evaluations.
#QuantumAnnealing #BayesianOptimization #Research
#QuantumAnnealing #BayesianOptimization #Research
ARAIM Protection Level Calculation Method Based on Bayesian Optimization
www.mdpi.com/2673-4591/12...
By Sitong Liu et al.
From the European Navigation Conference 2025
#ARAIM #GNSS #Integrity #BayesianOptimization
www.mdpi.com/2673-4591/12...
By Sitong Liu et al.
From the European Navigation Conference 2025
#ARAIM #GNSS #Integrity #BayesianOptimization
www.mdpi.com
February 24, 2026 at 11:45 AM
ARAIM Protection Level Calculation Method Based on Bayesian Optimization
www.mdpi.com/2673-4591/12...
By Sitong Liu et al.
From the European Navigation Conference 2025
#ARAIM #GNSS #Integrity #BayesianOptimization
www.mdpi.com/2673-4591/12...
By Sitong Liu et al.
From the European Navigation Conference 2025
#ARAIM #GNSS #Integrity #BayesianOptimization
The real trick of SMOG isn’t just learning faster—it’s turning each optimization into an opportunity to teach the next one. What could this philosophy mean for next-gen AI systems?
BayesianOptimization ML metaknowledge
BayesianOptimization ML metaknowledge
SMOG: Scalable Meta-Learning for Multi-Objective Bayesian Optimization
arxiv.org
February 19, 2026 at 3:16 PM
The real trick of SMOG isn’t just learning faster—it’s turning each optimization into an opportunity to teach the next one. What could this philosophy mean for next-gen AI systems?
BayesianOptimization ML metaknowledge
BayesianOptimization ML metaknowledge
Just crunched 64.59 s of Optuna trials and hit 0.9617 validation accuracy. The Bayesian TPE magic shows how fast smart hyperparameter tuning can be. Curious? Dive into the details! #HyperparameterTuning #Optuna #BayesianOptimization
🔗 aidailypost.com/news/hyperpa...
🔗 aidailypost.com/news/hyperpa...
January 19, 2026 at 11:42 PM
Just crunched 64.59 s of Optuna trials and hit 0.9617 validation accuracy. The Bayesian TPE magic shows how fast smart hyperparameter tuning can be. Curious? Dive into the details! #HyperparameterTuning #Optuna #BayesianOptimization
🔗 aidailypost.com/news/hyperpa...
🔗 aidailypost.com/news/hyperpa...
We combine ML + #DigitalTwins to autonomously design photoreactors with high light capture and uniform irradiation. A virtual photoreactor guides #BayesianOptimization, allowing numerous designs in silico and fabrication only of the best candidates.
🔗 @chemrxiv.org : chemrxiv.org/engage/chemr...
🔗 @chemrxiv.org : chemrxiv.org/engage/chemr...
January 5, 2026 at 9:29 AM
We combine ML + #DigitalTwins to autonomously design photoreactors with high light capture and uniform irradiation. A virtual photoreactor guides #BayesianOptimization, allowing numerous designs in silico and fabrication only of the best candidates.
🔗 @chemrxiv.org : chemrxiv.org/engage/chemr...
🔗 @chemrxiv.org : chemrxiv.org/engage/chemr...
Meta releases Ax 1.0 for automated machine learning optimization #Meta #MachineLearning #ArtificialIntelligence #BayesianOptimization #OpenSource
Meta releases Ax 1.0 for automated machine learning optimization
Meta launches Ax 1.0, an open-source platform using Bayesian optimization to automate complex experimentation across AI development, infrastructure tuning, and hardware design.
ppc.land
November 18, 2025 at 6:39 PM
Meta releases Ax 1.0 for automated machine learning optimization #Meta #MachineLearning #ArtificialIntelligence #BayesianOptimization #OpenSource
Meta releases Ax 1.0 for automated machine learning optimization #Meta #MachineLearning #ArtificialIntelligence #BayesianOptimization #OpenSource
Meta releases Ax 1.0 for automated machine learning optimization
Meta launches Ax 1.0, an open-source platform using Bayesian optimization to automate complex experimentation across AI development, infrastructure tuning, and hardware design.
ppc.land
November 18, 2025 at 6:38 PM
Meta releases Ax 1.0 for automated machine learning optimization #Meta #MachineLearning #ArtificialIntelligence #BayesianOptimization #OpenSource
Quantile‑Scaled Bayesian Optimization (QS‑BO) lets BO use rank‑only feedback by turning ranks into quantile pseudo‑observations, achieving gains at p < 1 % on synthetic benchmarks. https://getnews.me/quantile-scaled-bayesian-optimization-for-rank-only-feedback/ #bayesianoptimization #rankfeedback
October 8, 2025 at 7:23 AM
Quantile‑Scaled Bayesian Optimization (QS‑BO) lets BO use rank‑only feedback by turning ranks into quantile pseudo‑observations, achieving gains at p < 1 % on synthetic benchmarks. https://getnews.me/quantile-scaled-bayesian-optimization-for-rank-only-feedback/ #bayesianoptimization #rankfeedback
CCGBO, the method, assigns credit to past observations, achieving a sublinear regret bound and lower simple regret on benchmarks, with fewer evaluations. https://getnews.me/counterfactual-credit-guided-bayesian-optimization-advances-efficient-search/ #bayesianoptimization #counterfactualcredit
October 8, 2025 at 3:45 AM
CCGBO, the method, assigns credit to past observations, achieving a sublinear regret bound and lower simple regret on benchmarks, with fewer evaluations. https://getnews.me/counterfactual-credit-guided-bayesian-optimization-advances-efficient-search/ #bayesianoptimization #counterfactualcredit
ENS‑LoRA‑VBLL fine‑tunes LLMs with LoRA and a variational Bayesian last layer, beating standard GP surrogates on high‑dimensional benchmark and molecular design tasks. Read more: https://getnews.me/fine-tuning-llms-with-variational-bayesian-for-high-dim-optimization/ #llm #bayesianoptimization #lora
October 3, 2025 at 8:19 PM
ENS‑LoRA‑VBLL fine‑tunes LLMs with LoRA and a variational Bayesian last layer, beating standard GP surrogates on high‑dimensional benchmark and molecular design tasks. Read more: https://getnews.me/fine-tuning-llms-with-variational-bayesian-for-high-dim-optimization/ #llm #bayesianoptimization #lora
A PSD‑projected max kernel enables symmetry‑aware Bayesian Optimization to lower cumulative regret on synthetic and real‑world benchmarks without extra cost. Read more: https://getnews.me/symmetry-aware-bayesian-optimization-improves-performance-with-max-kernels/ #bayesianoptimization #symmetry
October 1, 2025 at 6:49 AM
A PSD‑projected max kernel enables symmetry‑aware Bayesian Optimization to lower cumulative regret on synthetic and real‑world benchmarks without extra cost. Read more: https://getnews.me/symmetry-aware-bayesian-optimization-improves-performance-with-max-kernels/ #bayesianoptimization #symmetry
Sequence‑only Bayesian optimization matches the affinity of structure‑enhanced models; a protein‑language constraint removes stability gaps. Read more: https://getnews.me/study-finds-sequence-data-sufficient-for-antibody-bayesian-optimization/ #bayesianoptimization #antibodydesign #proteinclanguage
October 1, 2025 at 1:18 AM
Sequence‑only Bayesian optimization matches the affinity of structure‑enhanced models; a protein‑language constraint removes stability gaps. Read more: https://getnews.me/study-finds-sequence-data-sufficient-for-antibody-bayesian-optimization/ #bayesianoptimization #antibodydesign #proteinclanguage
The new Contextual NeuroMHE system lets quadrotor drones adapt on‑the‑fly and cuts max position error by 20.3% versus prior methods. Read more: https://getnews.me/contextual-neural-moving-horizon-estimation-enhances-quadrotor-control/ #quadrotor #bayesianoptimization
September 30, 2025 at 7:34 PM
The new Contextual NeuroMHE system lets quadrotor drones adapt on‑the‑fly and cuts max position error by 20.3% versus prior methods. Read more: https://getnews.me/contextual-neural-moving-horizon-estimation-enhances-quadrotor-control/ #quadrotor #bayesianoptimization
Reasoning BO combines large language models with Bayesian optimization, boosting yield on a Direct Arylation reaction from 25.2% to 60.7% and was posted on 26 Sep 2025. Read more: https://getnews.me/reasoning-bo-boosting-bayesian-optimization-with-llm-reasoning/ #bayesianoptimization #llm
September 29, 2025 at 9:21 PM
Reasoning BO combines large language models with Bayesian optimization, boosting yield on a Direct Arylation reaction from 25.2% to 60.7% and was posted on 26 Sep 2025. Read more: https://getnews.me/reasoning-bo-boosting-bayesian-optimization-with-llm-reasoning/ #bayesianoptimization #llm
Closed-loop Bayesian optimization reduced sensor characterization from a year to a few days, tuning a low‑noise CCD to performance that previously took months of manual effort. https://getnews.me/bayesian-optimization-automates-sensor-characterization/ #bayesianoptimization #sensorcharacterization
September 29, 2025 at 5:42 PM
Closed-loop Bayesian optimization reduced sensor characterization from a year to a few days, tuning a low‑noise CCD to performance that previously took months of manual effort. https://getnews.me/bayesian-optimization-automates-sensor-characterization/ #bayesianoptimization #sensorcharacterization
BioBO, a biology‑informed Bayesian optimization method, improves labeling efficiency by roughly 25%‑40% over standard BO using gene embeddings and enrichment analysis. Read more: https://getnews.me/biobo-boosts-bayesian-optimization-for-gene-perturbation-design/ #biobo #bayesianoptimization
September 26, 2025 at 11:20 PM
BioBO, a biology‑informed Bayesian optimization method, improves labeling efficiency by roughly 25%‑40% over standard BO using gene embeddings and enrichment analysis. Read more: https://getnews.me/biobo-boosts-bayesian-optimization-for-gene-perturbation-design/ #biobo #bayesianoptimization
CAKE uses LLMs to evolve GP kernels and BAKER ranking; tests on hyper‑parameter tuning, robotics control and photonic chip design show faster convergence. https://getnews.me/ai-powered-cake-method-boosts-bayesian-optimization-kernels/ #bayesianoptimization #ai
September 25, 2025 at 5:42 AM
CAKE uses LLMs to evolve GP kernels and BAKER ranking; tests on hyper‑parameter tuning, robotics control and photonic chip design show faster convergence. https://getnews.me/ai-powered-cake-method-boosts-bayesian-optimization-kernels/ #bayesianoptimization #ai
Guided multi‑fidelity Bayesian optimization merges corrected digital‑twin sims with real‑world data, cutting robotic drive trials needed for comparable controller performance. https://getnews.me/guided-multi-fidelity-bayesian-optimization-improves-controller-tuning/ #bayesianoptimization #robotics
September 25, 2025 at 4:46 AM
Guided multi‑fidelity Bayesian optimization merges corrected digital‑twin sims with real‑world data, cutting robotic drive trials needed for comparable controller performance. https://getnews.me/guided-multi-fidelity-bayesian-optimization-improves-controller-tuning/ #bayesianoptimization #robotics
Bayesian optimization selects projection directions for sliced Wasserstein distance, delivering accurate results. The authors present four new selectors. Read more: https://getnews.me/adaptive-bayesian-optimization-improves-sliced-wasserstein-computation/ #bayesianoptimization #slicedwasserstein
September 24, 2025 at 10:03 PM
Bayesian optimization selects projection directions for sliced Wasserstein distance, delivering accurate results. The authors present four new selectors. Read more: https://getnews.me/adaptive-bayesian-optimization-improves-sliced-wasserstein-computation/ #bayesianoptimization #slicedwasserstein
Martin Hammerschmidt, CEO @JCMwave.bsky.social, presents MODAL-#NanoLab results at the 15th International Conference on Metamaterials, Photonic Crystals and Plasmonics, META 2025, Malaga, Spain, 23 July 2025. #MachineLearning #BayesianOptimization @cnoatzib.bsky.social,
Inv Talk in Session 2A10.
Inv Talk in Session 2A10.
July 22, 2025 at 11:35 AM
Martin Hammerschmidt, CEO @JCMwave.bsky.social, presents MODAL-#NanoLab results at the 15th International Conference on Metamaterials, Photonic Crystals and Plasmonics, META 2025, Malaga, Spain, 23 July 2025. #MachineLearning #BayesianOptimization @cnoatzib.bsky.social,
Inv Talk in Session 2A10.
Inv Talk in Session 2A10.