# AI Training Manager: Bounded Closed-Loop Control of Adaptive Training Recipes

arXiv: https://arxiv.org/abs/2606.29871 · HTML: https://mlsanity.com/paper/2606.29871/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 40 references that are in the map.
- First posted on arXiv 29 Jun 2026.

## Builds on (most cited first, 27 of 27)

- [SGDR: Stochastic Gradient Descent with Warm Restarts](https://mlsanity.com/paper/1608.03983/) (2016)
- [Optuna: A Next-generation Hyperparameter Optimization Framework](https://mlsanity.com/paper/1907.10902/) (2019)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics](https://mlsanity.com/paper/1705.07115/) (2017)
- [The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery](https://mlsanity.com/paper/2408.06292/) (2024)
- [Constrained Policy Optimization](https://mlsanity.com/paper/1705.10528/) (2017)
- [Large Language Models as Optimizers](https://mlsanity.com/paper/2309.03409/) (2023)
- [GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks](https://mlsanity.com/paper/1711.02257/) (2017)
- [Hyperband: A Novel Bandit-Based Approach to Hyperparameter Optimization](https://mlsanity.com/paper/1603.06560/) (2016)
- [Cyclical Learning Rates for Training Neural Networks](https://mlsanity.com/paper/1506.01186/) (2015)
- [Learning to learn by gradient descent by gradient descent](https://mlsanity.com/paper/1606.04474/) (2016)
- [TinyStories: How Small Can Language Models Be and Still Speak Coherent English?](https://mlsanity.com/paper/2305.07759/) (2023)
- [Population Based Training of Neural Networks](https://mlsanity.com/paper/1711.09846/) (2017)
- [MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation](https://mlsanity.com/paper/2310.03302/) (2023)
- [Learning to Optimize](https://mlsanity.com/paper/1606.01885/) (2016)
- [Neural Optimizer Search with Reinforcement Learning](https://mlsanity.com/paper/1709.07417/) (2017)
- [Using Large Language Models for Hyperparameter Optimization](https://mlsanity.com/paper/2312.04528/) (2023)
- [Automated Reinforcement Learning (AutoRL): A Survey and Open Problems](https://mlsanity.com/paper/2201.03916/) (2022)
- [Large Language Model Agent for Hyper-Parameter Optimization](https://mlsanity.com/paper/2402.01881/) (2024)
- [VeLO: Training Versatile Learned Optimizers by Scaling Up](https://mlsanity.com/paper/2211.09760/) (2022)
- [Automated Dynamic Algorithm Configuration](https://mlsanity.com/paper/2205.13881/) (2022)
- [Sequential Large Language Model-Based Hyper-parameter Optimization](https://mlsanity.com/paper/2410.20302/) (2024)
- [Learning to Optimize for Reinforcement Learning](https://mlsanity.com/paper/2302.01470/) (2023)
- [Theory-inspired Parameter Control Benchmarks for Dynamic Algorithm Configuration](https://mlsanity.com/paper/2202.03259/) (2022)
- [AutoRL Hyperparameter Landscapes](https://mlsanity.com/paper/2304.02396/) (2023)
- [A Proposal to Extend the Common Model of Cognition with Metacognition](https://mlsanity.com/paper/2506.07807/) (2025)
- [Small LLMs with Expert Blocks Are Good Enough for Hyperparamter Tuning](https://mlsanity.com/paper/2509.15561/) (2025)

## Built on by (most cited first, 1 of 1)

- [Interactive Training 2: Auditable Control Plane for Live Model Training](https://mlsanity.com/paper/2607.18314/) (2026)

## Abstract (arXiv)

We present the AI Training Manager, a bounded LLM-based metacognitive monitoring-and-control layer for machine-learning training. The Manager asynchronously observes structured telemetry from an active training run, assesses the current training regime, and selects adaptive interventions through a constrained, deterministically verified action interface. We evaluate the feasibility of this architecture using deliberately induced, interpretable failure modes in supervised and reinforcement learning tasks. For supervised learning, we induce a multi-objective overfitting failure in a GPT-2-style model trained on TinyStories. The Manager prevents the resulting late-run validation collapse, reducing final validation loss by 54.3% relative to the same stressed recipe. For reinforcement learning, we study robotic reaching under opposing multifactor stress regimes. In a conservative regime, the Manager raises final deterministic safe success from 0.413 to 0.705. In an aggressive regime, it raises safe success from 0.121 to 0.764. The same Manager instructions and action surface are used in both regimes. These results suggest that bounded LLM reasoning can provide a practical metacognitive control layer over ongoing learning processes.