Related papers: Where does output diversity collapse in post-train…
Reinforcement learning has been central to recent advances in large language model reasoning, but most algorithms rely on on-policy training that demands fresh rollouts at every update, limiting efficiency and scalability. Asynchronous RL…
Pre-trained language models (LMs) are capable of in-context learning (ICL): they can adapt to a task with only a few examples given in the prompt without any parameter update. However, it is unclear where this capability comes from as there…
Given the ease of creating synthetic data from machine learning models, new models can be potentially trained on synthetic data generated by previous models. This recursive training process raises concerns about the long-term impact on…
Diversity in training data, architecture, and providers is assumed to mitigate homogeneity in LLMs. However, we lack empirical evidence on whether different LLMs differ meaningfully. We conduct a large-scale empirical evaluation on over 350…
Group Relative Policy Optimization (GRPO) trains Chain-of-Thought reasoning with verifiable rewards, but estimating thought-level advantages without value functions often suffers from high variance. Although tree-style branching is used in…
Large Language Models (LLMs) that undergo recursive training on synthetically generated data are susceptible to model collapse, a phenomenon marked by the generation of meaningless output. Existing research has examined this issue from…
Continual learning for large language models is typically evaluated through accuracy retention under sequential fine-tuning. We argue that this perspective is incomplete, because uncertainty reliability can degrade earlier and more sharply…
Supervised fine-tuning (SFT) on chain-of-thought data is an essential post-training step for reasoning language models. Standard machine learning intuition suggests that training with more unique training samples yields better…
Reinforcement Learning (RL)-based post-training has significantly advanced the complex reasoning capabilities of language models, fostering sophisticated self-reflection processes. However, this ``slow thinking'' paradigm presents a…
Continual instruction tuning(CIT) during the post-training phase is crucial for adapting multimodal large language models (MLLMs) to evolving real-world demands. However, the progress is hampered by the lack of benchmarks with rigorous,…
Large language models (LLMs) tend to generate homogenous texts, which may impact the diversity of knowledge generated across different outputs. Given their potential to replace existing forms of knowledge acquisition, this poses a risk of…
Chain-of-thought (CoT) prompting has become central to mathematical reasoning in large language models, yet models remain brittle to early errors: a single arithmetic slip or unjustified inference typically propagates uncorrected to an…
Intermediate token generation (ITG), where a model produces output before the solution, has been proposed as a method to improve the performance of language models on reasoning tasks. While these reasoning traces or Chain of Thoughts (CoTs)…
Modern reasoning language models generate dense, sequential chain-of-thought traces implicitly assuming that every token contributes and that steps must be consumed in order. We challenge both assumptions through a systematic intervention…
Meta-learning aims to leverage information across related tasks to improve prediction on unlabeled data for new tasks when only a small number of labeled observations are available ("few-shot" learning). Increased task diversity is often…
Recent success of pre-trained language models (LMs) has spurred widespread interest in the language capabilities that they possess. However, efforts to understand whether LM representations are useful for symbolic reasoning tasks have been…
Reinforcement learning (RL) has shown extraordinary potential in aligning diffusion models to downstream tasks, yet most of them still suffer from significant reward hacking, which degrades generative diversity and quality by inducing…
We aim to develop a fundamental understanding of modality collapse, a recently observed empirical phenomenon wherein models trained for multimodal fusion tend to rely only on a subset of the modalities, ignoring the rest. We show that…
The remarkable performance of models like the OpenAI o1 can be attributed to their ability to emulate human-like long-time thinking during inference. These models employ extended chain-of-thought (CoT) processes, exploring multiple…
Multilingual Large Language Models considerably changed how technologies influence language. While previous technologies could mediate or assist humans, there is now a tendency to offload the task of writing itself to these technologies,…