Bayesian black-box function paradigmknowledg expected [33,25].

Self-play fine-tuning has demonstrated promising abilities in adapting large language models (LLMs) to downstream tasks with limited real-world data. The basic principle is to iteratively refine the model with real samples and synthetic ones generated from itself. However, the existing methods primarily focus on the relative gaps between the rewards for two types of data, neglecting their absolute values. Through theoretical analysis, we identify that the gap-based methods suffer from unstable evolution, due to the potentially degenerated objectives. To address this limitation, we introduce a novel self-play fine-tuning method, namely Self-PlAy via Noise Contrastive Estimation (SPACE), which leverages noise contrastive estimation to capture the real-world data distribution. Specifically, SPACE treats synthetic samples as auxiliary components, and discriminates them from the real ones in a binary classification manner. As a result, SPACE independently optimizes the absolute reward values for each type of data, ensuring a consistently meaningful objective and thereby avoiding the instability issue. Theoretically, we show that the optimal solution of the objective in SPACE aligns with the underlying distribution of real-world data, and SPACE guarantees a provably stable convergence to the optimal distribution. Empirically, we show that SPACE significantly improves the performance of LLMs over various tasks, and outperforms supervised fine-tuning that employs much more real-world samples. Compared to gap-based self-play fine-tuning methods, SPACE exhibits remarkable superiority and stable evolution.

Agents are predominantly evaluated and optimized via task success metrics, which are coarse, rely on manual design from experts, and fail to reward intermediate emergent behaviors. We propose **AutoLibra**, a framework for agent evaluation, that transforms open-ended human feedback *e.g.* "If you find that the button is disabled, don't click it again", or "This agent has too much autonomy to decide what to do on its own" into metrics for evaluating fine-grained behaviors in agent trajectories. AutoLibra accomplishes this by grounding feedback to an agent's behavior, clustering similar positive and negative behaviors, and creating concrete metrics with clear definitions and concrete examples, which can be used for prompting LLM-as-a-Judge as evaluators. We further propose two meta metrics to evaluate the alignment of a set of (induced) metrics with open feedback: "coverage" and "redundancy". Through optimizing these meta-metrics, we experimentally demonstrate AutoLibra's ability to induce more concrete agent evaluation metrics than the ones proposed in previous agent evaluation benchmarks and discover new metrics to analyze agents. We also present two applications of AutoLibra in agent improvement: First, we show that AutoLibra serve human prompt engineers for diagonalize agent failures and improve prompts iterative. Moreover, we find that AutoLibra can induce metrics for automatic optimization for agents, which makes agents improve through self-regulation. Our results suggest that AutoLibra is a powerful task-agnostic tool for evaluating and improving language agents.

Estimating individualized treatment effects from observational data presents a persistent challenge due to unmeasured confounding and structural bias. Causal Machine Learning (causal ML) methods, such as causal trees and doubly robust estimators, provide tools for estimating conditional average treatment effects. These methods have limited effectiveness in complex real-world environments due to the presence of latent confounders or those described in unstructured formats. Moreover, reliance on domain experts for confounder identification and rule interpretation introduces high annotation cost and scalability concerns. In this work, we proposed Large Language Model-based agents for automated confounder discovery and subgroup analysis that integrate agents into the causal ML pipeline to simulate domain expertise. Our framework systematically performs subgroup identification and confounding structure discovery by leveraging the reasoning capabilities of LLM-based agents, which reduces human dependency while preserving interpretability. Experiments on real-world medical datasets show that our proposed approach enhances treatment effect estimation robustness by narrowing confidence intervals and uncovering unrecognized confounding biases. Our findings suggest that LLM-based agents offer a promising path toward scalable, trustworthy, and semantically aware causal inference.

The Dawid-Skene estimator has been widely used for inferring the true labels from the noisy labels provided by non-expert crowdsourcing workers. However, since the estimator maximizes a non-convex log-likelihood function, it is hard to theoretically justify its performance. In this paper, we propose a two-stage efficient algorithm for multi-class crowd labeling problems. The first stage uses the spectral method to obtain an initial estimate of parameters.

One of the main challenges in optimal scaling of large language models (LLMs) is the prohibitive cost of hyperparameter tuning, particularly learning rate $\eta$ and batch size $B$. While techniques like $\mu$P (Yang et al., 2022) provide scaling rules for optimal $\eta$ transfer in the infinite model size limit, the optimal scaling behavior in the infinite data size limit remains unknown. We fill in this gap by observing for the first time an intricate dependence of optimal $\eta$ scaling on the pretraining token budget $T$, $B$ and its relation to the critical batch size $B_\mathrm{crit}$, which we measure to evolve as $B_\mathrm{crit} \propto T$. Furthermore, we show that the optimal batch size is positively correlated with $B_\mathrm{crit}$: keeping it fixed becomes suboptimal over time even if learning rate is scaled optimally. Surprisingly, our results demonstrate that the observed optimal $\eta$ and $B$ dynamics are preserved with $\mu$P model scaling, challenging the conventional view of $B_\mathrm{crit}$ dependence solely on loss value. Complementing optimality, we examine the sensitivity of loss to changes in learning rate, where we find the sensitivity to decrease with increase of $T$ and to remain constant with $\mu$P model scaling. We hope our results make the first step towards a unified picture of the joint optimal data and model scaling.

In previous research, we developed methods to train decision trees (DT) as agents for reinforcement learning tasks, based on deep reinforcement learning (DRL) networks. The samples from which the DTs are built, use the environment's state as features and the corresponding action as label. To solve the nontrivial task of selecting samples, which on one hand reflect the DRL agent's capabilities of choosing the right action but on the other hand also cover enough state space to generalize well, we developed an algorithm to iteratively train DTs. In this short paper, we apply this algorithm to a real-world implementation of a robotic task for the first time. Real-world tasks pose additional challenges compared to simulations, such as noise and delays. The task consists of a physical pendulum attached to a cart, which moves on a linear track. By movements to the left and to the right, the pendulum is to be swung in the upright position and balanced in the unstable equilibrium. Our results demonstrate the applicability of the algorithm to real-world tasks by generating a DT whose performance matches the performance of the DRL agent, while consisting of fewer parameters. This research could be a starting point for distilling DTs from DRL agents to obtain transparent, lightweight models for real-world reinforcement learning tasks.

Topology optimization is a structural design methodology widely utilized to address engineering challenges. However, sensitivity-based topology optimization methods struggle to solve optimization problems characterized by strong non-linearity. Leveraging the sensitivity-free nature and high capacity of deep generative models, data-driven topology design (DDTD) methodology is considered an effective solution to this problem. Despite this, the training effectiveness of deep generative models diminishes when input size exceeds a threshold while maintaining high degrees of freedom is crucial for accurately characterizing complex structures. To resolve the conflict between the both, we propose DDTD based on principal component analysis (PCA). Its core idea is to replace the direct training of deep generative models with material distributions by using a principal component score matrix obtained from PCA computation and to obtain the generated material distributions with new features through the restoration process. We apply the proposed PCA-based DDTD to the problem of minimizing the maximum stress in 3D structural mechanics and demonstrate it can effectively address the current challenges faced by DDTD that fail to handle 3D structural design problems. Various experiments are conducted to demonstrate the effectiveness and practicability of the proposed PCA-based DDTD.

The Dawid-Skene estimator has been widely used for inferring the true labels from the noisy labels provided by non-expert crowdsourcing workers. However, since the estimator maximizes a non-convex log-likelihood function, it is hard to theoretically justify its performance. In this paper, we propose a two-stage efficient algorithm for multi-class crowd labeling problems. The first stage uses the spectral method to obtain an initial estimate of parameters.

Harnessing the power of human-annotated data through Supervised Fine-Tuning (SFT) is pivotal for advancing Large Language Models (LLMs). In this paper, we delve into the prospect of growing a strong LLM out of a weak one without the need for acquiring additional human-annotated data. We propose a new fine-tuning method called Self-Play fIne-tuNing (SPIN), which starts from a supervised fine-tuned model. At the heart of SPIN lies a self-play mechanism, where the LLM refines its capability by playing against instances of itself. More specifically, the LLM generates its own training data from its previous iterations, refining its policy by discerning these self-generated responses from those obtained from human-annotated data. Our method progressively elevates the LLM from a nascent model to a formidable one, unlocking the full potential of human-annotated demonstration data for SFT. Theoretically, we prove that the global optimum to the training objective function of our method is achieved only when the LLM policy aligns with the target data distribution. Empirically, we evaluate our method on several benchmark datasets including the HuggingFace Open LLM Leaderboard, MT-Bench, and datasets from Big-Bench. Our results show that SPIN can significantly improve the LLM's performance across a variety of benchmarks and even outperform models trained through direct preference optimization (DPO) supplemented with extra GPT-4 preference data. This sheds light on the promise of self-play, enabling the achievement of human-level performance in LLMs without the need for expert opponents.