Learning to Play Blackjack: A Curriculum Learning Perspective

Reinforcement Learning (RL) agents often struggle with efficiency and performance in complex environments. We propose a novel framework that uses a Large Language Model (LLM) to dynamically generate a curriculum over available actions, enabling the agent to incorporate each action individually. We apply this framework to the game of Blackjack, where the LLM creates a multi-stage training path that progressively introduces complex actions to a Tabular Q-Learning and a Deep Q-Network (DQN) agent. Our evaluation in a realistic 8-deck simulation over 10 independent runs demonstrates significant performance gains over standard training methods. The curriculum-based approach increases the DQN agent's average win rate from 43.97% to 47.41%, reduces the average bust rate from 32.9% to 28.0%, and accelerates the overall workflow by over 74%, with the agent's full training completing faster than the baseline's evaluation phase alone. These results validate that LLM-guided curricula can build more effective, robust, and efficient RL agents.

Learn by Surprise, Commit by Proof

We propose LSCP, a self-gated post-training framework for autonomous knowledge acquisition: learning only what a model does not already know, verified against what it does know, at a strength proportional to conviction, with no external oracle. When a passage produces anomalously high per-token loss, LSCP flags it, generates a Q&A chain that forces the model to articulate its own knowledge and identify gaps, then adjusts AdamW's $β_2$ proportionally to conviction depth k (the number of self-verification steps the passage survives) via $β_2 = 0.999 \cdot r^k$. The entire learning intensity is governed by a single parameter $r$. Beyond new knowledge, this process sharpens weakly encoded existing knowledge, which is a primary source of hallucination. The framework is self-extinguishing: as the model learns, per-token loss on learned passages decreases toward the surprisal threshold and the system progressively converges to standard AdamW. This models biological memory consolidation: temporary information in the context window is selectively consolidated into parametric weights, the model's long-term memory. Experiments on the reference model (Qwen3-14B) and across six models (8B--32B, four families) show that standard fine-tuning produces rote memorization (perturbation gap (the ratio of paraphrase to original perplexity) of 11.6 +- 0.2 x baseline) while all LSCP conditions learn semantically (2.7--3.0x). The r=1.0 condition (identical optimizer, nearly identical data, only Q&A format differs) confirms that the training data format, not $β_2$ gating, is the primary mechanism preventing memorization; gating instead protects neighboring knowledge from contamination by corrupt content (93 +- 7% accuracy on adjacent questions at r=0.98 vs. 90% baseline).

Residuals-based Offline Reinforcement Learning

Offline reinforcement learning (RL) has received increasing attention for learning policies from previously collected data without interaction with the real environment, which is particularly important in high-stakes applications. While a growing body of work has developed offline RL algorithms, these methods often rely on restrictive assumptions about data coverage and suffer from distribution shift. In this paper, we propose a residuals-based offline RL framework for general state and action spaces. Specifically, we define a residuals-based Bellman optimality operator that explicitly incorporates estimation error in learning transition dynamics into policy optimization by leveraging empirical residuals. We show that this Bellman operator is a contraction mapping and identify conditions under which its fixed point is asymptotically optimal and possesses finite-sample guarantees. We further develop a residuals-based offline deep Q-learning (DQN) algorithm. Using a stochastic CartPole environment, we demonstrate the effectiveness of our residuals-based offline DQN algorithm.

Full-Gradient Successor Feature Representations

Successor Features (SF) combined with Generalized Policy Improvement (GPI) provide a robust framework for transfer learning in Reinforcement Learning (RL) by decoupling environment dynamics from reward functions. However, standard SF learning methods typically rely on semi-gradient Temporal Difference (TD) updates. When combined with non-linear function approximation, semi-gradient methods lack robust convergence guarantees and can lead to instability, particularly in the multi-task setting where accurate feature estimation is critical for effective GPI. Inspired by Full Gradient DQN, we propose Full-Gradient Successor Feature Representations Q-Learning (FG-SFRQL), an algorithm that optimizes the successor features by minimizing the full Mean Squared Bellman Error. Unlike standard approaches, our method computes gradients with respect to parameters in both the online and target networks. We provide a theoretical proof of almost-sure convergence for FG-SFRQL and demonstrate empirically that minimizing the full residual leads to superior sample efficiency and transfer performance compared to semi-gradient baselines in both discrete and continuous domains.

Agent Q-Mix: Selecting the Right Action for LLM Multi-Agent Systems through Reinforcement Learning

Large Language Models (LLMs) have shown remarkable performance in completing various tasks. However, solving complex problems often requires the coordination of multiple agents, raising a fundamental question: how to effectively select and interconnect these agents. In this paper, we propose \textbf{Agent Q-Mix}, a reinforcement learning framework that reformulates topology selection as a cooperative Multi-Agent Reinforcement Learning (MARL) problem. Our method learns decentralized communication decisions using QMIX value factorization, where each agent selects from a set of communication actions that jointly induce a round-wise communication graph. At its core, Agent Q-Mix combines a topology-aware GNN encoder, GRU memory, and per-agent Q-heads under a Centralized Training with Decentralized Execution (CTDE) paradigm. The framework optimizes a reward function that balances task accuracy with token cost. Across seven core benchmarks in coding, reasoning, and mathematics, Agent Q-Mix achieves the highest average accuracy compared to existing methods while demonstrating superior token efficiency and robustness against agent failure. Notably, on the challenging Humanity's Last Exam (HLE) using Gemini-3.1-Flash-Lite as a backbone, Agent Q-Mix achieves 20.8\% accuracy, outperforming Microsoft Agent Framework (19.2\%) and LangGraph (19.2\%), followed by AutoGen and Lobster by OpenClaw. These results underscore the effectiveness of learned, decentralized topology optimization in pushing the boundaries of multi-agent reasoning.

Coupled Query-Key Dynamics for Attention

Standard scaled dot-product attention computes scores from static, independent projections of the input. We show that evolving queries and keys \emph{jointly} through shared learned dynamics before scoring - which we call \textbf{coupled QK dynamics} - improves language modeling perplexity and training stability. On WikiText-103 at 60M parameters, coupled dynamics achieves 22.55--22.62 perplexity vs.\ 24.22 for standard attention ($-$6.6--6.9\%), with only 0.11\% additional parameters (shared across both instantiations). A structural ablation isolates coupling as the active ingredient: a symplectic (Hamiltonian) and a non-symplectic (Euler) integrator perform identically when both couple Q and K, while an uncoupled MLP baseline of matched capacity reaches only 23.81 with 8$\times$ higher seed variance. The integration step count (1--7) is similarly irrelevant - a single coupled step suffices. A compute-matched comparison reveals that coupling is a \emph{sample-efficiency} mechanism: standard attention trained for 2.4$\times$ longer (matching wall-clock) reaches the same perplexity, but requires 2.4$\times$ more tokens. The advantage scales to 150M ($-$6.7\%) but narrows at 350M ($-$1.0\%), where Differential Attention (18.93) overtakes coupled dynamics (19.35). The benefit is corpus-dependent: coupling helps on domain-coherent text (WikiText-103 $-$6.6\%, PubMed $-$4.5\%) but degrades on heterogeneous web text ($+$10.3\%) and shows no benefit on GLUE. We characterize when coupling helps and when it does not, providing practical guidelines.

Soft MPCritic: Amortized Model Predictive Value Iteration

Reinforcement learning (RL) and model predictive control (MPC) offer complementary strengths, yet combining them at scale remains computationally challenging. We propose soft MPCritic, an RL-MPC framework that learns in (soft) value space while using sample-based planning for both online control and value target generation. soft MPCritic instantiates MPC through model predictive path integral control (MPPI) and trains a terminal Q-function with fitted value iteration, aligning the learned value function with the planner and implicitly extending the effective planning horizon. We introduce an amortized warm-start strategy that recycles planned open-loop action sequences from online observations when computing batched MPPI-based value targets. This makes soft MPCritic computationally practical, while preserving solution quality. soft MPCritic plans in a scenario-based fashion with an ensemble of dynamic models trained for next-step prediction accuracy. Together, these ingredients enable soft MPCritic to learn effectively through robust, short-horizon planning on classic and complex control tasks. These results establish soft MPCritic as a practical and scalable blueprint for synthesizing MPC policies in settings where policy extraction and direct, long-horizon planning may fail.

Learning Compact Terrain-Context Representations for Feasibility-Aware Offline Reinforcement Learning in UAV Relaying Networks

Offline reinforcement learning (RL) is an attractive tool for unmanned aerial vehicle (UAV) systems, where online exploration is costly and raises safety concerns. In terrain-aware UAV relaying, agents may observe high-dimensional inputs such as terrain and land-cover maps, which describe the propagation environment, but complicate offline learning from fixed datasets. This paper investigates the impact of compact state representations on offline RL for UAV relaying. End-to-end service is jointly constrained by UAV--user access links and a base-station--to--UAV backhaul link, yielding feasibility limits driven by user mobility and independent of UAV control. To distinguish feasibility limits from control-induced sub-optimality, a candidate-set feasibility upper bound (CS-FUB) is introduced, which estimates the maximum achievable user coverage over a restricted set of UAV placements. To address high-dimensional terrain context, map-like observations are compressed into low-dimensional latent representations using a variational autoencoder (VAE) and policies are trained via Conservative Q-Learning (CQL). Simulation results show that training CQL directly on raw high-dimensional terrain-context states leads to slow convergence and large feasibility gaps. In contrast, VAE-encoded representations improve learning stability, enable earlier convergence to feasible relay configurations, and reduce sub-optimality relative to physical limits. Comparisons with autoencoder and linear compression baselines further demonstrate the benefit of structured representation learning for effective offline RL in terrain-aware UAV systems.

Activation Saturation and Floquet Spectrum Collapse in Neural ODEs

We prove that activation saturation imposes a structural dynamical limitation on autonomous Neural ODEs $\dot{h}=f_θ(h)$ with saturating activations ($\tanh$, sigmoid, etc.): if $q$ hidden layers of the MLP $f_θ$ satisfy $|σ'|\leδ$ on a region~$U$, the input Jacobian is attenuated as $\norm{Df_θ(x)}\le C(U)$ (for activations with $\sup_{x}|σ'(x)|\le 1$, e.g.\ $\tanh$ and sigmoid, this reduces to $C_Wδ^q$), forcing every Floquet (Lyapunov) exponen along any $T$-periodic orbit $γ\subset U$ into the interval $[-C(U),\;C(U)]$. This is a collapse of the Floquet spectrum: as saturation deepens ($δ\to 0$), all exponents are driven to zero, limiting both strong contraction and chaotic sensitivity. The obstruction is structural -- it constrains the learned vector field at inference time, independent of training quality. As a secondary contribution, for activations with $σ'>0$, a saturation-weighted spectral factorisation yields a refined bound $\widetilde{C}(U)\le C(U)$ whose improvement is amplified exponentially in~$T$ at the flow level. All results are numerically illustrated on the Stuart--Landau oscillator; the bounds provide a theoretical explanation for the empirically observed failure of $\tanh$-NODEs on the Morris--Lecar neuron model.

Q-DIVER: Integrated Quantum Transfer Learning and Differentiable Quantum Architecture Search with EEG Data

Integrating quantum circuits into deep learning pipelines remains challenging due to heuristic design limitations. We propose Q-DIVER, a hybrid framework combining a large-scale pretrained EEG encoder (DIVER-1) with a differentiable quantum classifier. Unlike fixed-ansatz approaches, we employ Differentiable Quantum Architecture Search to autonomously discover task-optimal circuit topologies during end-to-end fine-tuning. On the PhysioNet Motor Imagery dataset, our quantum classifier achieves predictive performance comparable to classical multi-layer perceptrons (Test F1: 63.49\%) while using approximately \textbf{50$\times$ fewer task-specific head parameters} (2.10M vs. 105.02M). These results validate quantum transfer learning as a parameter-efficient strategy for high-dimensional biological signal processing.