We propose a new algorithm for fine-tuning large language models using reinforcement learning. Tapered Off-Policy REINFORCE (TOPR) uses an asymmetric, tapered variant of importance sampling to speed up learning while maintaining stable learning dynamics, even without the use of KL regularization. TOPR can be applied in a fully offline fashion, allows the handling of positive and negative examples in a unified framework, and benefits from the implementational simplicity that is typical of Monte Carlo algorithms. We demonstrate the effectiveness of our approach with a series of experiments on the GSM8K and MATH reasoning benchmarks, finding performance gains for training both a model for solution generation as a generative verifier, and on a learning to search task, using the model as a query expander. We show that properly leveraging positive and negative examples alike in the off-policy regime simultaneously increases test-time accuracy and training data efficiency, all the while avoiding the "wasted inference" that comes with discarding negative examples. We find that this advantage persists over multiple iterations of training and can be amplified by dataset curation techniques, enabling us to match 70B-parameter model performance with 8B language models. As a corollary to this work, we find that REINFORCE's baseline parameter plays an important and unexpected role in defining dataset composition in the presence of negative examples, and is consequently critical in driving off-policy performance.

Current AI advances largely rely on scaling neural models and expanding training datasets to achieve generalization and robustness.

The recent success of generative AI highlights the crucial role of high-quality human feedback in building trustworthy AI systems. However, the increasing use of large language models (LLMs) by crowdsourcing workers poses a significant challenge: datasets intended to reflect human input may be compromised by LLM-generated responses. Existing LLM detection approaches often rely on high-dimensional training data such as text, making them unsuitable for structured annotation tasks like multiple-choice labeling. In this work, we investigate the potential of peer prediction--a mechanism that evaluates the information within workers' responses--to mitigate LLM-assisted cheating in crowdsourcing with a focus on annotation tasks.

Initialization plays a critical role in Deep Neural Network training, directly influencing convergence, stability, and generalization. Common approaches such as Glorot and He initializations rely on randomness, which can produce uneven weight distributions across layer connections. In this paper, we introduce the Sinusoidal initialization, a novel deterministic method that employs sinusoidal functions to construct structured weight matrices expressly to improve the spread and balance of weights throughout the network while simultaneously fostering a more uniform, well-conditioned distribution of neuron activation states from the very first forward pass. Because Sinusoidal initialization begins with weights and activations that are already evenly and efficiently utilized, it delivers consistently faster convergence, greater training stability, and higher final accuracy across a wide range of models, including convolutional neural networks, vision transformers, and large language models. On average, our experiments show an increase of 4.9% in final validation accuracy and 20.9% in convergence speed. By replacing randomness with structure, this initialization provides a stronger and more reliable foundation for Deep Learning systems.

Recent advances in scene understanding have leveraged multimodal large language models (MLLMs) for 3D reasoning by capitalizing on their strong 2D pretraining.

Understanding the internal mechanisms of large language models (LLMs) remains a challenging and complex endeavor. Even fundamental questions, such as how fine-tuning affects model behavior, often require extensive empirical evaluation. In this paper, we introduce a novel perspective based on the geometric properties of contextual latent embeddings to study the effects of training and fine-tuning. To that end, we measure the local dimensions of a contextual language model's latent space and analyze their shifts during training and fine-tuning. We show that the local dimensions provide insights into the model's training dynamics and generalization ability. Specifically, the mean of the local dimensions predicts when the model's training capabilities are exhausted, as exemplified in a dialogue state tracking task, overfitting, as demonstrated in an emotion recognition task, and grokking, as illustrated with an arithmetic task. Furthermore, our experiments suggest a practical heuristic: reductions in the mean local dimension tend to accompany and predict subsequent performance gains. Through this exploration, we aim to provide practitioners with a deeper understanding of the implications of fine-tuning on embedding spaces, facilitating informed decisions when configuring models for specific applications. The results of this work contribute to the ongoing discourse on the interpretability, adaptability, and generalizability of LLMs by bridging the gap between intrinsic model mechanisms and geometric properties in embeddings.

Recent studies in Machine Learning (ML) for biological research focus on investigating molecular properties to accelerate drug discovery. However, limited labeled molecular data often hampers the performance of ML models. A common strategy to mitigate data scarcity is leveraging auxiliary learning tasks to provide additional supervision, but selecting effective auxiliary tasks requires substantial domain expertise and manual effort, and their inclusion does not always guarantee performance gains. To overcome these challenges, we introduce Automatic Auxiliary Task Selection (AUTAUT), a fully automated framework that seamlessly retrieves auxiliary tasks using large language models and adaptively integrates them through a novel gradient alignment weighting mechanism. By automatically emphasizing auxiliary tasks aligned with the primary objective, AUTAUT significantly enhances predictive accuracy while reducing negative impacts from irrelevant tasks. Extensive evaluations demonstrate that AUTAUT outperforms 10 auxiliary task-based approaches and 18 advanced molecular property prediction models.

Knowledge Graph Question Answering (KGQA) systems rely on high-quality benchmarks to evaluate complex multi-hop reasoning. However, despite their widespread use, popular datasets such as WebQSP and CWQ suffer from critical quality issues, including inaccurate or incomplete ground-truth annotations, poorly constructed questions that are ambiguous, trivial, or unanswerable, and outdated or inconsistent knowledge. Through a manual audit of 16 popular KGQA datasets--including WebQSPand CWQ--we find that the average factual correctness rate is only 57%. To address these issues, we introduce KGQAGen, an LLM-inthe-loop framework that systematically resolves these pitfalls. KGQAGencombines structured knowledge grounding, LLM-guided generation, and symbolic verification to produce challenging and verifiable QA instances. Using KGQAGen, we construct KGQAGen-10k, a 10K-scale benchmark grounded in Wikidata, and evaluate a diverse set of KG-RAG models. Experimental results demonstrate that even state-of-the-art systems struggle on this benchmark, highlighting its ability to expose limitations of existing models. Our findings advocate for more rigorous benchmark construction and position KGQAGen as a scalable framework for advancing KGQA evaluation 1.

Recent studies have demonstrated the feasibility of modeling single-cell data as natural languages and the potential of leveraging powerful large language models (LLMs) for understanding cell biology. However, a comprehensive evaluation of LLMs' performance on language-driven single-cell analysis tasks remains unexplored. Motivated by this challenge, we introduce CELLVERSE, a unified language-centric question-answering benchmark that integrates four types of single-cell multi-omics data and encompasses three hierarchical levels of single-cell analysis tasks: cell type annotation (cell-level), drug response prediction (drug-level), and perturbation analysis (gene-level). Going beyond this, we systematically evaluate the performance across 14 open-source and closed-source LLMs ranging 160M 671B on CELLVERSE. Remarkably, the experimental results reveal: Existing specialist models (e.g., C2S-Pythia) fail to make reasonable decisions across all sub-tasks within CELLVERSE, while generalist models such as Qwen, Llama, GPT, and DeepSeekfamily models exhibit preliminary understanding capabilities within the realm of cell biology. The performance of current LLMs falls short of expectations and has substantial room for improvement. Notably, in the widely studied drug response prediction task, none of the evaluated LLMs demonstrate significant performance improvement over random guessing. CELLVERSE offers the first large-scale empirical demonstration that significant challenges still remain in applying LLMs to cell biology. By introducing CELLVERSE, we lay the foundation for advancing cell biology through natural languages and hope this paradigm could facilitate next-generation single-cell analysis.

The problem of over-smoothing has emerged as a fundamental issue for Graph Convolutional Networks (GCNs). While existing efforts primarily focus on enhancing the discriminability of node representations for node classification, they tend to overlook the over-smoothing at the graph level, significantly influencing the performance of graph classification. In this paper, we provide an explanation of the graph-level over-smoothing phenomenon and propose a novel Adaptive MultiViewed Subgraph Convolutional Network (MultiNet) to address this challenge. Specifically, the MultiNet introduces a local subgraph convolution module that adaptively divides each input graph into multiple subgraph views. Then a number of subgraph-based view-specific convolution operations are applied to constrain the extent of node information propagation over the original global graph structure, not only mitigating the over-smoothing issue but also generating more discriminative local node representations. Moreover, we develop an alignment-based readout that establishes correspondences between nodes over different graphs, thereby effectively preserving the local node-level structure information and improving the discriminative ability of the resulting graph-level representations. Theoretical analysis and empirical studies show that the MultiNet mitigates the graph-level over-smoothing and achieves excellent performance for graph classification.