Daily Paper Cast

🤗 Upvotes: 25 | cs.LG, cs.AI, cs.CL Authors: Dawei Li, Renliang Sun, Yue Huang, Ming Zhong, Bohan Jiang, Jiawei Han, Xiangliang Zhang, Wei Wang, Huan Liu Title: Preference Leakage: A Contamination Problem in LLM-as-a-judge Arxiv: http://arxiv.org/abs/2502.01534v1 Abstract: Large Language Models (LLMs) as judges and LLM-based data synthesis have emerged as two fundamental LLM-driven data annotation methods in model development. While their combination significantly enhances the efficiency of model training and evaluation, little attention has been given to the potential contamination brought by this new model development paradigm. In this work, we expose preference leakage, a contamination problem in LLM-as-a-judge caused by the relatedness between the synthetic data generators and LLM-based evaluators. To study this issue, we first define three common relatednesses between data generator LLM and judge LLM: being the same model, having an inheritance relationship, and belonging to the same model family. Through extensive experiments, we empirically confirm the bias of judges towards their related student models caused by preference leakage across multiple LLM baselines and benchmarks. Further analysis suggests that preference leakage is a pervasive issue that is harder to detect compared to previously identified biases in LLM-as-a-judge scenarios. All of these findings imply that preference leakage is a widespread and challenging problem in the area of LLM-as-a-judge. We release all codes and data at: https://github.com/David-Li0406/Preference-Leakage.

🤗 Upvotes: 19 | cs.CV, cs.GR, cs.LG Authors: Rohit Gandikota, Zongze Wu, Richard Zhang, David Bau, Eli Shechtman, Nick Kolkin Title: SliderSpace: Decomposing the Visual Capabilities of Diffusion Models Arxiv: http://arxiv.org/abs/2502.01639v1 Abstract: We present SliderSpace, a framework for automatically decomposing the visual capabilities of diffusion models into controllable and human-understandable directions. Unlike existing control methods that require a user to specify attributes for each edit direction individually, SliderSpace discovers multiple interpretable and diverse directions simultaneously from a single text prompt. Each direction is trained as a low-rank adaptor, enabling compositional control and the discovery of surprising possibilities in the model's latent space. Through extensive experiments on state-of-the-art diffusion models, we demonstrate SliderSpace's effectiveness across three applications: concept decomposition, artistic style exploration, and diversity enhancement. Our quantitative evaluation shows that SliderSpace-discovered directions decompose the visual structure of model's knowledge effectively, offering insights into the latent capabilities encoded within diffusion models. User studies further validate that our method produces more diverse and useful variations compared to baselines. Our code, data and trained weights are available at https://sliderspace.baulab.info

🤗 Upvotes: 15 | cs.AI, cs.CV Authors: Huanqia Cai, Yijun Yang, Winston Hu Title: MM-IQ: Benchmarking Human-Like Abstraction and Reasoning in Multimodal Models Arxiv: http://arxiv.org/abs/2502.00698v1 Abstract: IQ testing has served as a foundational methodology for evaluating human cognitive capabilities, deliberately decoupling assessment from linguistic background, language proficiency, or domain-specific knowledge to isolate core competencies in abstraction and reasoning. Yet, artificial intelligence research currently lacks systematic benchmarks to quantify these critical cognitive dimensions in multimodal systems. To address this critical gap, we propose MM-IQ, a comprehensive evaluation framework comprising 2,710 meticulously curated test items spanning 8 distinct reasoning paradigms. Through systematic evaluation of leading open-source and proprietary multimodal models, our benchmark reveals striking limitations: even state-of-the-art architectures achieve only marginally superior performance to random chance (27.49% vs. 25% baseline accuracy). This substantial performance chasm highlights the inadequacy of current multimodal systems in approximating fundamental human reasoning capacities, underscoring the need for paradigm-shifting advancements to bridge this cognitive divide.

🤗 Upvotes: 12 | cs.CV, cs.AI, cs.CL, cs.HC, cs.LG Authors: Ahmed Heakl, Sara Ghaboura, Omkar Thawkar, Fahad Shahbaz Khan, Hisham Cholakkal, Rao Muhammad Anwer, Salman Khan Title: AIN: The Arabic INclusive Large Multimodal Model Arxiv: http://arxiv.org/abs/2502.00094v1 Abstract: Amid the swift progress of large language models (LLMs) and their evolution into large multimodal models (LMMs), significant strides have been made in high-resource languages such as English and Chinese. While Arabic LLMs have seen notable progress, Arabic LMMs remain largely unexplored, often narrowly focusing on a few specific aspects of the language and visual understanding. To bridge this gap, we introduce AIN-the Arabic Inclusive Multimodal Model-designed to excel across diverse domains. AIN is an English-Arabic bilingual LMM designed to excel in English and Arabic, leveraging carefully constructed 3.6 million high-quality Arabic-English multimodal data samples. AIN demonstrates state-of-the-art Arabic performance, while also possessing strong English-language visual capabilities. On the recent CAMEL-Bench benchmark comprising 38 sub-domains including, multi-image understanding, complex visual perception, handwritten document understanding, video understanding, medical imaging, plant diseases, and remote sensing-based land use understanding, our AIN demonstrates strong performance with the 7B model outperforming GPT-4o by an absolute gain of 3.4% averaged over eight domains and 38 sub-domains. AIN's superior capabilities position it as a significant step toward empowering Arabic speakers with advanced multimodal generative AI tools across diverse applications.

🤗 Upvotes: 54 | cs.CL, cs.AI, cs.LG Authors: Niklas Muennighoff, Zitong Yang, Weijia Shi, Xiang Lisa Li, Li Fei-Fei, Hannaneh Hajishirzi, Luke Zettlemoyer, Percy Liang, Emmanuel Candès, Tatsunori Hashimoto Title: s1: Simple test-time scaling Arxiv: http://arxiv.org/abs/2501.19393v1 Abstract: Test-time scaling is a promising new approach to language modeling that uses extra test-time compute to improve performance. Recently, OpenAI's o1 model showed this capability but did not publicly share its methodology, leading to many replication efforts. We seek the simplest approach to achieve test-time scaling and strong reasoning performance. First, we curate a small dataset s1K of 1,000 questions paired with reasoning traces relying on three criteria we validate through ablations: difficulty, diversity, and quality. Second, we develop budget forcing to control test-time compute by forcefully terminating the model's thinking process or lengthening it by appending "Wait" multiple times to the model's generation when it tries to end. This can lead the model to double-check its answer, often fixing incorrect reasoning steps. After supervised finetuning the Qwen2.5-32B-Instruct language model on s1K and equipping it with budget forcing, our model s1 exceeds o1-preview on competition math questions by up to 27% (MATH and AIME24). Further, scaling s1 with budget forcing allows extrapolating beyond its performance without test-time intervention: from 50% to 57% on AIME24. Our model, data, and code are open-source at https://github.com/simplescaling/s1.

🤗 Upvotes: 28 | cs.CL, cs.AI Authors: Baohao Liao, Yuhui Xu, Hanze Dong, Junnan Li, Christof Monz, Silvio Savarese, Doyen Sahoo, Caiming Xiong Title: Reward-Guided Speculative Decoding for Efficient LLM Reasoning Arxiv: http://arxiv.org/abs/2501.19324v1 Abstract: We introduce Reward-Guided Speculative Decoding (RSD), a novel framework aimed at improving the efficiency of inference in large language models (LLMs). RSD synergistically combines a lightweight draft model with a more powerful target model, incorporating a controlled bias to prioritize high-reward outputs, in contrast to existing speculative decoding methods that enforce strict unbiasedness. RSD employs a process reward model to evaluate intermediate decoding steps and dynamically decide whether to invoke the target model, optimizing the trade-off between computational cost and output quality. We theoretically demonstrate that a threshold-based mixture strategy achieves an optimal balance between resource utilization and performance. Extensive evaluations on challenging reasoning benchmarks, including Olympiad-level tasks, show that RSD delivers significant efficiency gains against decoding with the target model only (up to 4.4x fewer FLOPs), while achieving significant better accuracy than parallel decoding method on average (up to +3.5). These results highlight RSD as a robust and cost-effective approach for deploying LLMs in resource-intensive scenarios.

🤗 Upvotes: 12 | cs.CL, cs.AI Authors: Qika Lin, Tianzhe Zhao, Kai He, Zhen Peng, Fangzhi Xu, Ling Huang, Jingying Ma, Mengling Feng Title: Self-supervised Quantized Representation for Seamlessly Integrating Knowledge Graphs with Large Language Models Arxiv: http://arxiv.org/abs/2501.18119v1 Abstract: Due to the presence of the natural gap between Knowledge Graph (KG) structures and the natural language, the effective integration of holistic structural information of KGs with Large Language Models (LLMs) has emerged as a significant question. To this end, we propose a two-stage framework to learn and apply quantized codes for each entity, aiming for the seamless integration of KGs with LLMs. Firstly, a self-supervised quantized representation (SSQR) method is proposed to compress both KG structural and semantic knowledge into discrete codes (\ie, tokens) that align the format of language sentences. We further design KG instruction-following data by viewing these learned codes as features to directly input to LLMs, thereby achieving seamless integration. The experiment results demonstrate that SSQR outperforms existing unsupervised quantized methods, producing more distinguishable codes. Further, the fine-tuned LLaMA2 and LLaMA3.1 also have superior performance on KG link prediction and triple classification tasks, utilizing only 16 tokens per entity instead of thousands in conventional prompting methods.

🤗 Upvotes: 10 | cs.CV, cs.CL Authors: Zhiheng Lyu, Xueguang Ma, Wenhu Chen Title: PixelWorld: Towards Perceiving Everything as Pixels Arxiv: http://arxiv.org/abs/2501.19339v1 Abstract: Existing foundation models typically process visual input as pixels and textual input as tokens, a paradigm that contrasts with human perception, where both modalities are processed in a unified manner. With the rise of embodied and agentic AI, where inputs primarily come from camera pixels, the need for a unified perception framework becomes increasingly evident. In this paper, we propose to unify all modalities (text, tables, code, diagrams, images, etc) as pixel inputs, i.e. "Perceive Everything as Pixels" (PEAP). We introduce PixelWorld, a novel evaluation suite that unifies all the mentioned modalities into pixel space to gauge the existing models' performance. Our findings show that (1) PEAP outperforms baseline with token-based input in multimodal datasets, benefiting from unified input for better disambiguation, (2) significant declines in reasoning and coding capabilities across all models when processing pixel-based input, underscoring the need to enhance foundation models' perceptual abilities, (3) larger models can maintain strong performance on non-reasoning tasks under PEAP, while smaller models like Phi-3.5-V suffer significant performance degradation, (4) the attention pattern of PEAP is highly aligned with text token input, (5) PEAP can be accelerated significantly by exploiting the spatial sparsity. We conclude that the existing frontier models are competent in pixel perception, however, there is still headroom for improvement. Our code, dataset will be released upon acceptance.

🤗 Upvotes: 8 | cs.RO, cs.AI Authors: Gaoyue Zhou, Hengkai Pan, Yann LeCun, Lerrel Pinto Title: DINO-WM: World Models on Pre-trained Visual Features enable Zero-shot Planning Arxiv: http://arxiv.org/abs/2411.04983v2 Abstract: The ability to predict future outcomes given control actions is fundamental for physical reasoning. However, such predictive models, often called world models, remains challenging to learn and are typically developed for task-specific solutions with online policy learning. To unlock world models' true potential, we argue that they should 1) be trainable on offline, pre-collected trajectories, 2) support test-time behavior optimization, and 3) facilitate task-agnostic reasoning. To this end, we present DINO World Model (DINO-WM), a new method to model visual dynamics without reconstructing the visual world. DINO-WM leverages spatial patch features pre-trained with DINOv2, enabling it to learn from offline behavioral trajectories by predicting future patch features. This allows DINO-WM to achieve observational goals through action sequence optimization, facilitating task-agnostic planning by treating goal features as prediction targets. We demonstrate that DINO-WM achieves zero-shot behavioral solutions at test time on six environments without expert demonstrations, reward modeling, or pre-learned inverse models, outperforming prior state-of-the-art work across diverse task families such as arbitrarily configured mazes, push manipulation with varied object shapes, and multi-particle scenarios.

🤗 Upvotes: 6 | cs.CL, cs.AI, cs.CR, cs.LG Authors: Mrinank Sharma, Meg Tong, Jesse Mu, Jerry Wei, Jorrit Kruthoff, Scott Goodfriend, Euan Ong, Alwin Peng, Raj Agarwal, Cem Anil, Amanda Askell, Nathan Bailey, Joe Benton, Emma Bluemke, Samuel R. Bowman, Eric Christiansen, Hoagy Cunningham, Andy Dau, Anjali Gopal, Rob Gilson, Logan Graham, Logan Howard, Nimit Kalra, Taesung Lee, Kevin Lin, Peter Lofgren, Francesco Mosconi, Clare O'Hara, Catherine Olsson, Linda Petrini, Samir Rajani, Nikhil Saxena, Alex Silverstein, Tanya Singh, Theodore Sumers, Leonard Tang, Kevin K. Troy, Constantin Weisser, Ruiqi Zhong, Giulio Zhou, Jan Leike, Jared Kaplan, Ethan Perez Title: Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming Arxiv: http://arxiv.org/abs/2501.18837v1 Abstract: Large language models (LLMs) are vulnerable to universal jailbreaks-prompting strategies that systematically bypass model safeguards and enable users to carry out harmful processes that require many model interactions, like manufacturing illegal substances at scale. To defend against these attacks, we introduce Constitutional Classifiers: safeguards trained on synthetic data, generated by prompting LLMs with natural language rules (i.e., a constitution) specifying permitted and restricted content. In over 3,000 estimated hours of red teaming, no red teamer found a universal jailbreak that could extract information from an early classifier-guarded LLM at a similar level of detail to an unguarded model across most target queries. On automated evaluations, enhanced classifiers demonstrated robust defense against held-out domain-specific jailbreaks. These classifiers also maintain deployment viability, with an absolute 0.38% increase in production-traffic refusals and a 23.7% inference overhead. Our work demonstrates that defending against universal jailbreaks while maintaining practical deployment viability is tractable.

🤗 Upvotes: 6 | cs.CL, cs.AI, cs.LG Authors: Ken M. Nakanishi Title: Scalable-Softmax Is Superior for Attention Arxiv: http://arxiv.org/abs/2501.19399v1 Abstract: The maximum element of the vector output by the Softmax function approaches zero as the input vector size increases. Transformer-based language models rely on Softmax to compute attention scores, causing the attention distribution to flatten as the context size grows. This reduces the model's ability to prioritize key information effectively and potentially limits its length generalization. To address this problem, we propose Scalable-Softmax (SSMax), which replaces Softmax in scenarios where the input vector size varies. SSMax can be seamlessly integrated into existing Transformer-based architectures. Experimental results in language modeling show that models using SSMax not only achieve faster loss reduction during pretraining but also significantly improve performance in long contexts and key information retrieval. Furthermore, an analysis of attention scores reveals that SSMax enables the model to focus attention on key information even in long contexts. Additionally, although models that use SSMax from the beginning of pretraining achieve better length generalization, those that have already started pretraining can still gain some of this ability by replacing Softmax in the attention layers with SSMax, either during or after pretraining.

🤗 Upvotes: 3 | cs.LG, math.OC, stat.ML Authors: Fabian Schaipp, Alexander Hägele, Adrien Taylor, Umut Simsekli, Francis Bach Title: The Surprising Agreement Between Convex Optimization Theory and Learning-Rate Scheduling for Large Model Training Arxiv: http://arxiv.org/abs/2501.18965v1 Abstract: We show that learning-rate schedules for large model training behave surprisingly similar to a performance bound from non-smooth convex optimization theory. We provide a bound for the constant schedule with linear cooldown; in particular, the practical benefit of cooldown is reflected in the bound due to the absence of logarithmic terms. Further, we show that this surprisingly close match between optimization theory and practice can be exploited for learning-rate tuning: we achieve noticeable improvements for training 124M and 210M Llama-type models by (i) extending the schedule for continued training with optimal learning-rate, and (ii) transferring the optimal learning-rate across schedules.

🤗 Upvotes: 3 | cs.LG, cs.AI Authors: Bartosz Cywiński, Kamil Deja Title: SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders Arxiv: http://arxiv.org/abs/2501.18052v2 Abstract: Diffusion models, while powerful, can inadvertently generate harmful or undesirable content, raising significant ethical and safety concerns. Recent machine unlearning approaches offer potential solutions but often lack transparency, making it difficult to understand the changes they introduce to the base model. In this work, we introduce SAeUron, a novel method leveraging features learned by sparse autoencoders (SAEs) to remove unwanted concepts in text-to-image diffusion models. First, we demonstrate that SAEs, trained in an unsupervised manner on activations from multiple denoising timesteps of the diffusion model, capture sparse and interpretable features corresponding to specific concepts. Building on this, we propose a feature selection method that enables precise interventions on model activations to block targeted content while preserving overall performance. Evaluation with the competitive UnlearnCanvas benchmark on object and style unlearning highlights SAeUron's state-of-the-art performance. Moreover, we show that with a single SAE, we can remove multiple concepts simultaneously and that in contrast to other methods, SAeUron mitigates the possibility of generating unwanted content, even under adversarial attack. Code and checkpoints are available at: https://github.com/cywinski/SAeUron.

🤗 Upvotes: 46 | cs.CR, cs.AI, cs.LG Authors: Yue Liu, Hongcheng Gao, Shengfang Zhai, Jun Xia, Tianyi Wu, Zhiwei Xue, Yulin Chen, Kenji Kawaguchi, Jiaheng Zhang, Bryan Hooi Title: GuardReasoner: Towards Reasoning-based LLM Safeguards Arxiv: http://arxiv.org/abs/2501.18492v1 Abstract: As LLMs increasingly impact safety-critical applications, ensuring their safety using guardrails remains a key challenge. This paper proposes GuardReasoner, a new safeguard for LLMs, by guiding the guard model to learn to reason. Concretely, we first create the GuardReasonerTrain dataset, which consists of 127K samples with 460K detailed reasoning steps. Then, we introduce reasoning SFT to unlock the reasoning capability of guard models. In addition, we present hard sample DPO to further strengthen their reasoning ability. In this manner, GuardReasoner achieves better performance, explainability, and generalizability. Extensive experiments and analyses on 13 benchmarks of 3 guardrail tasks demonstrate its superiority. Remarkably, GuardReasoner 8B surpasses GPT-4o+CoT by 5.74% and LLaMA Guard 3 8B by 20.84% F1 score on average. We release the training data, code, and models with different scales (1B, 3B, 8B) of GuardReasoner : https://github.com/yueliu1999/GuardReasoner/.

🤗 Upvotes: 22 | cs.CL Authors: Yue Wang, Qiuzhi Liu, Jiahao Xu, Tian Liang, Xingyu Chen, Zhiwei He, Linfeng Song, Dian Yu, Juntao Li, Zhuosheng Zhang, Rui Wang, Zhaopeng Tu, Haitao Mi, Dong Yu Title: Thoughts Are All Over the Place: On the Underthinking of o1-Like LLMs Arxiv: http://arxiv.org/abs/2501.18585v1 Abstract: Large language models (LLMs) such as OpenAI's o1 have demonstrated remarkable abilities in complex reasoning tasks by scaling test-time compute and exhibiting human-like deep thinking. However, we identify a phenomenon we term underthinking, where o1-like LLMs frequently switch between different reasoning thoughts without sufficiently exploring promising paths to reach a correct solution. This behavior leads to inadequate depth of reasoning and decreased performance, particularly on challenging mathematical problems. To systematically analyze this issue, we conduct experiments on three challenging test sets and two representative open-source o1-like models, revealing that frequent thought switching correlates with incorrect responses. We introduce a novel metric to quantify underthinking by measuring token efficiency in incorrect answers. To address underthinking, we propose a decoding strategy with thought switching penalty TIP that discourages premature transitions between thoughts, encouraging deeper exploration of each reasoning path. Experimental results demonstrate that our approach improves accuracy across challenging datasets without requiring model fine-tuning. Our findings contribute to understanding reasoning inefficiencies in o1-like LLMs and offer a practical solution to enhance their problem-solving capabilities.

🤗 Upvotes: 15 | cs.CL Authors: Arthur Douillard, Yanislav Donchev, Keith Rush, Satyen Kale, Zachary Charles, Zachary Garrett, Gabriel Teston, Dave Lacey, Ross McIlroy, Jiajun Shen, Alexandre Ramé, Arthur Szlam, Marc'Aurelio Ranzato, Paul Barham Title: Streaming DiLoCo with overlapping communication: Towards a Distributed Free Lunch Arxiv: http://arxiv.org/abs/2501.18512v1 Abstract: Training of large language models (LLMs) is typically distributed across a large number of accelerators to reduce training time. Since internal states and parameter gradients need to be exchanged at each and every single gradient step, all devices need to be co-located using low-latency high-bandwidth communication links to support the required high volume of exchanged bits. Recently, distributed algorithms like DiLoCo have relaxed such co-location constraint: accelerators can be grouped into ``workers'', where synchronizations between workers only occur infrequently. This in turn means that workers can afford being connected by lower bandwidth communication links without affecting learning quality. However, in these methods, communication across workers still requires the same peak bandwidth as before, as the synchronizations require all parameters to be exchanged across all workers. In this paper, we improve DiLoCo in three ways. First, we synchronize only subsets of parameters in sequence, rather than all at once, which greatly reduces peak bandwidth. Second, we allow workers to continue training while synchronizing, which decreases wall clock time. Third, we quantize the data exchanged by workers, which further reduces bandwidth across workers. By properly combining these modifications, we show experimentally that we can distribute training of billion-scale parameters and reach similar quality as before, but reducing required bandwidth by two orders of magnitude.

🤗 Upvotes: 15 | cs.AI, cs.CL, cs.CV, cs.LG Authors: Yuxin Zuo, Shang Qu, Yifei Li, Zhangren Chen, Xuekai Zhu, Ermo Hua, Kaiyan Zhang, Ning Ding, Bowen Zhou Title: MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding Arxiv: http://arxiv.org/abs/2501.18362v1 Abstract: We introduce MedXpertQA, a highly challenging and comprehensive benchmark to evaluate expert-level medical knowledge and advanced reasoning. MedXpertQA includes 4,460 questions spanning 17 specialties and 11 body systems. It includes two subsets, Text for text evaluation and MM for multimodal evaluation. Notably, MM introduces expert-level exam questions with diverse images and rich clinical information, including patient records and examination results, setting it apart from traditional medical multimodal benchmarks with simple QA pairs generated from image captions. MedXpertQA applies rigorous filtering and augmentation to address the insufficient difficulty of existing benchmarks like MedQA, and incorporates specialty board questions to improve clinical relevance and comprehensiveness. We perform data synthesis to mitigate data leakage risk and conduct multiple rounds of expert reviews to ensure accuracy and reliability. We evaluate 16 leading models on MedXpertQA. Moreover, medicine is deeply connected to real-world decision-making, providing a rich and representative setting for assessing reasoning abilities beyond mathematics and code. To this end, we develop a reasoning-oriented subset to facilitate the assessment of o1-like models.

🤗 Upvotes: 10 | cs.AI, q-bio.NC Authors: Lan Pan, Hanbo Xie, Robert C. Wilson Title: Large Language Models Think Too Fast To Explore Effectively Arxiv: http://arxiv.org/abs/2501.18009v1 Abstract: Large Language Models have emerged many intellectual capacities. While numerous benchmarks assess their intelligence, limited attention has been given to their ability to explore, an essential capacity for discovering new information and adapting to novel environments in both natural and artificial systems. The extent to which LLMs can effectively explore, particularly in open-ended tasks, remains unclear. This study investigates whether LLMs can surpass humans in exploration during an open-ended task, using Little Alchemy 2 as a paradigm, where agents combine elements to discover new ones. Results show most LLMs underperform compared to humans, except for the o1 model, with those traditional LLMs relying primarily on uncertainty driven strategies, unlike humans who balance uncertainty and empowerment. Representational analysis of the models with Sparse Autoencoders revealed that uncertainty and choices are represented at earlier transformer blocks, while empowerment values are processed later, causing LLMs to think too fast and make premature decisions, hindering effective exploration. These findings shed light on the limitations of LLM exploration and suggest directions for improving their adaptability.

🤗 Upvotes: 10 | cs.LG, cs.CL Authors: Benjamin Feuer, Chinmay Hegde Title: WILDCHAT-50M: A Deep Dive Into the Role of Synthetic Data in Post-Training Arxiv: http://arxiv.org/abs/2501.18511v1 Abstract: Language model (LLM) post-training, from DPO to distillation, can refine behaviors and unlock new skills, but the open science supporting these post-training techniques is still in its infancy. One limiting factor has been the difficulty of conducting large-scale comparative analyses of synthetic data generating models and LLM judges. To close this gap, we introduce WILDCHAT-50M, the largest public chat dataset to date. We extend the existing WildChat dataset to include responses not only from GPT, but from over 50 different open-weight models, ranging in size from 0.5B to 104B parameters. We conduct an extensive comparative analysis and demonstrate the potential of this dataset by creating RE-WILD, our own public SFT mix, which outperforms the recent Tulu-3 SFT mixture from Allen AI with only 40% as many samples. Our dataset, samples and code are available at https://github.com/penfever/wildchat-50m.

🤗 Upvotes: 10 | cs.CV, cs.AI, cs.CL, cs.LG, cs.RO Authors: Wei Chow, Jiageng Mao, Boyi Li, Daniel Seita, Vitor Guizilini, Yue Wang Title: PhysBench: Benchmarking and Enhancing Vision-Language Models for Physical World Understanding Arxiv: http://arxiv.org/abs/2501.16411v2 Abstract: Understanding the physical world is a fundamental challenge in embodied AI, critical for enabling agents to perform complex tasks and operate safely in real-world environments. While Vision-Language Models (VLMs) have shown great promise in reasoning and task planning for embodied agents, their ability to comprehend physical phenomena remains extremely limited. To close this gap, we introduce PhysBench, a comprehensive benchmark designed to evaluate VLMs' physical world understanding capability across a diverse set of tasks. PhysBench contains 10,002 entries of interleaved video-image-text data, categorized into four major domains: physical object properties, physical object relationships, physical scene understanding, and physics-based dynamics, further divided into 19 subclasses and 8 distinct capability dimensions. Our extensive experiments, conducted on 75 representative VLMs, reveal that while these models excel in common-sense reasoning, they struggle with understanding the physical world -- likely due to the absence of physical knowledge in their training data and the lack of embedded physical priors. To tackle the shortfall, we introduce PhysAgent, a novel framework that combines the generalization strengths of VLMs with the specialized expertise of vision models, significantly enhancing VLMs' physical understanding across a variety of tasks, including an 18.4\% improvement on GPT-4o. Furthermore, our results demonstrate that enhancing VLMs' physical world understanding capabilities can help embodied agents such as MOKA. We believe that PhysBench and PhysAgent offer valuable insights and contribute to bridging the gap between VLMs and physical world understanding.

🤗 Upvotes: 6 | cs.SE, cs.AI Authors: Aitor Arrieta, Miriam Ugarte, Pablo Valle, José Antonio Parejo, Sergio Segura Title: o3-mini vs DeepSeek-R1: Which One is Safer? Arxiv: http://arxiv.org/abs/2501.18438v1 Abstract: The irruption of DeepSeek-R1 constitutes a turning point for the AI industry in general and the LLMs in particular. Its capabilities have demonstrated outstanding performance in several tasks, including creative thinking, code generation, maths and automated program repair, at apparently lower execution cost. However, LLMs must adhere to an important qualitative property, i.e., their alignment with safety and human values. A clear competitor of DeepSeek-R1 is its American counterpart, OpenAI's o3-mini model, which is expected to set high standards in terms of performance, safety and cost. In this paper we conduct a systematic assessment of the safety level of both, DeepSeek-R1 (70b version) and OpenAI's o3-mini (beta version). To this end, we make use of our recently released automated safety testing tool, named ASTRAL. By leveraging this tool, we automatically and systematically generate and execute a total of 1260 unsafe test inputs on both models. After conducting a semi-automated assessment of the outcomes provided by both LLMs, the results indicate that DeepSeek-R1 is highly unsafe as compared to OpenAI's o3-mini. Based on our evaluation, DeepSeek-R1 answered unsafely to 11.98% of the executed prompts whereas o3-mini only to 1.19%.

🤗 Upvotes: 1 | cs.AI, cs.CL, cs.HC Authors: Faria Huq, Zora Zhiruo Wang, Frank F. Xu, Tianyue Ou, Shuyan Zhou, Jeffrey P. Bigham, Graham Neubig Title: CowPilot: A Framework for Autonomous and Human-Agent Collaborative Web Navigation Arxiv: http://arxiv.org/abs/2501.16609v1 Abstract: While much work on web agents emphasizes the promise of autonomously performing tasks on behalf of users, in reality, agents often fall short on complex tasks in real-world contexts and modeling user preference. This presents an opportunity for humans to collaborate with the agent and leverage the agent's capabilities effectively. We propose CowPilot, a framework supporting autonomous as well as human-agent collaborative web navigation, and evaluation across task success and task efficiency. CowPilot reduces the number of steps humans need to perform by allowing agents to propose next steps, while users are able to pause, reject, or take alternative actions. During execution, users can interleave their actions with the agent by overriding suggestions or resuming agent control when needed. We conducted case studies on five common websites and found that the human-agent collaborative mode achieves the highest success rate of 95% while requiring humans to perform only 15.2% of the total steps. Even with human interventions during task execution, the agent successfully drives up to half of task success on its own. CowPilot can serve as a useful tool for data collection and agent evaluation across websites, which we believe will enable research in how users and agents can work together. Video demonstrations are available at https://oaishi.github.io/cowpilot.html

🤗 Upvotes: 28 | cs.CL Authors: Yubo Wang, Xiang Yue, Wenhu Chen Title: Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate Arxiv: http://arxiv.org/abs/2501.17703v2 Abstract: Supervised Fine-Tuning (SFT) is commonly used to train language models to imitate annotated responses for given instructions. In this paper, we challenge this paradigm and propose Critique Fine-Tuning (CFT), a strategy where models learn to critique noisy responses rather than simply imitate correct ones. Inspired by human learning processes that emphasize critical thinking, CFT encourages deeper analysis and nuanced understanding-traits often overlooked by standard SFT. To validate the effectiveness of CFT, we construct a 50K-sample dataset from WebInstruct, using GPT-4o as the teacher to generate critiques in the form of ([query; noisy response], critique). CFT on this dataset yields a consistent 4-10% improvement over SFT on six math benchmarks with different base models like Qwen2.5, Qwen2.5-Math and DeepSeek-Math. We further expand to MetaMath and NuminaMath datasets and observe similar gains over SFT. Notably, our model Qwen2.5-Math-CFT only requires 1 hour training on 8xH100 over the 50K examples. It can match or outperform strong competitors like Qwen2.5-Math-Instruct on most benchmarks, which use over 2M samples. Moreover, it can match the performance of SimpleRL, which is a deepseek-r1 replication trained with 140x more compute. Ablation studies show that CFT is robust to the source of noisy response and teacher critique model. Through these findings, we argue that CFT offers a more effective alternative to advance the reasoning of language models.

🤗 Upvotes: 24 | cs.CL, cs.AI Authors: Andrei Alexandru, Antonia Calvi, Henry Broomfield, Jackson Golden, Kyle Dai, Mathias Leys, Maurice Burger, Max Bartolo, Roman Engeler, Sashank Pisupati, Toby Drane, Young Sun Park Title: Atla Selene Mini: A General Purpose Evaluation Model Arxiv: http://arxiv.org/abs/2501.17195v1 Abstract: We introduce Atla Selene Mini, a state-of-the-art small language model-as-a-judge (SLMJ). Selene Mini is a general-purpose evaluator that outperforms the best SLMJs and GPT-4o-mini on overall performance across 11 out-of-distribution benchmarks, spanning absolute scoring, classification, and pairwise preference tasks. It is the highest-scoring 8B generative model on RewardBench, surpassing strong baselines like GPT-4o and specialized judges. To achieve this, we develop a principled data curation strategy that augments public datasets with synthetically generated critiques and ensures high quality through filtering and dataset ablations. We train our model on a combined direct preference optimization (DPO) and supervised fine-tuning (SFT) loss, and produce a highly promptable evaluator that excels in real-world scenarios. Selene Mini shows dramatically improved zero-shot agreement with human expert evaluations on financial and medical industry datasets. It is also robust to variations in prompt format. Preliminary results indicate that Selene Mini is the top-ranking evaluator in a live, community-driven Judge Arena. We release the model weights on HuggingFace (https://hf.co/AtlaAI/Selene-1-Mini-Llama-3.1-8B) and Ollama to encourage widespread community adoption.

🤗 Upvotes: 14 | cs.AI, cs.CY, cs.LG Authors: Clément Desroches, Martin Chauvin, Louis Ladan, Caroline Vateau, Simon Gosset, Philippe Cordier Title: Exploring the sustainable scaling of AI dilemma: A projective study of corporations' AI environmental impacts Arxiv: http://arxiv.org/abs/2501.14334v2 Abstract: The rapid growth of artificial intelligence (AI), particularly Large Language Models (LLMs), has raised concerns regarding its global environmental impact that extends beyond greenhouse gas emissions to include consideration of hardware fabrication and end-of-life processes. The opacity from major providers hinders companies' abilities to evaluate their AI-related environmental impacts and achieve net-zero targets. In this paper, we propose a methodology to estimate the environmental impact of a company's AI portfolio, providing actionable insights without necessitating extensive AI and Life-Cycle Assessment (LCA) expertise. Results confirm that large generative AI models consume up to 4600x more energy than traditional models. Our modelling approach, which accounts for increased AI usage, hardware computing efficiency, and changes in electricity mix in line with IPCC scenarios, forecasts AI electricity use up to 2030. Under a high adoption scenario, driven by widespread Generative AI and agents adoption associated to increasingly complex models and frameworks, AI electricity use is projected to rise by a factor of 24.4. Mitigating the environmental impact of Generative AI by 2030 requires coordinated efforts across the AI value chain. Isolated measures in hardware efficiency, model efficiency, or grid improvements alone are insufficient. We advocate for standardized environmental assessment frameworks, greater transparency from the all actors of the value chain and the introduction of a "Return on Environment" metric to align AI development with net-zero goals.

🤗 Upvotes: 8 | cs.SE, cs.AI Authors: Aitor Arrieta, Miriam Ugarte, Pablo Valle, José Antonio Parejo, Sergio Segura Title: Early External Safety Testing of OpenAI's o3-mini: Insights from the Pre-Deployment Evaluation Arxiv: http://arxiv.org/abs/2501.17749v1 Abstract: Large Language Models (LLMs) have become an integral part of our daily lives. However, they impose certain risks, including those that can harm individuals' privacy, perpetuate biases and spread misinformation. These risks highlight the need for robust safety mechanisms, ethical guidelines, and thorough testing to ensure their responsible deployment. Safety of LLMs is a key property that needs to be thoroughly tested prior the model to be deployed and accessible to the general users. This paper reports the external safety testing experience conducted by researchers from Mondragon University and University of Seville on OpenAI's new o3-mini LLM as part of OpenAI's early access for safety testing program. In particular, we apply our tool, ASTRAL, to automatically and systematically generate up to date unsafe test inputs (i.e., prompts) that helps us test and assess different safety categories of LLMs. We automatically generate and execute a total of 10,080 unsafe test input on a early o3-mini beta version. After manually verifying the test cases classified as unsafe by ASTRAL, we identify a total of 87 actual instances of unsafe LLM behavior. We highlight key insights and findings uncovered during the pre-deployment external testing phase of OpenAI's latest LLM.

🤗 Upvotes: 8 | cs.CV Authors: Hailong Guo, Bohan Zeng, Yiren Song, Wentao Zhang, Chuang Zhang, Jiaming Liu Title: Any2AnyTryon: Leveraging Adaptive Position Embeddings for Versatile Virtual Clothing Tasks Arxiv: http://arxiv.org/abs/2501.15891v1 Abstract: Image-based virtual try-on (VTON) aims to generate a virtual try-on result by transferring an input garment onto a target person's image. However, the scarcity of paired garment-model data makes it challenging for existing methods to achieve high generalization and quality in VTON. Also, it limits the ability to generate mask-free try-ons. To tackle the data scarcity problem, approaches such as Stable Garment and MMTryon use a synthetic data strategy, effectively increasing the amount of paired data on the model side. However, existing methods are typically limited to performing specific try-on tasks and lack user-friendliness. To enhance the generalization and controllability of VTON generation, we propose Any2AnyTryon, which can generate try-on results based on different textual instructions and model garment images to meet various needs, eliminating the reliance on masks, poses, or other conditions. Specifically, we first construct the virtual try-on dataset LAION-Garment, the largest known open-source garment try-on dataset. Then, we introduce adaptive position embedding, which enables the model to generate satisfactory outfitted model images or garment images based on input images of different sizes and categories, significantly enhancing the generalization and controllability of VTON generation. In our experiments, we demonstrate the effectiveness of our Any2AnyTryon and compare it with existing methods. The results show that Any2AnyTryon enables flexible, controllable, and high-quality image-based virtual try-on generation.https://logn-2024.github.io/Any2anyTryonProjectPage/

🤗 Upvotes: 6 | cs.CR, cs.AI, cs.CL, cs.LG Authors: Tiansheng Huang, Sihao Hu, Fatih Ilhan, Selim Furkan Tekin, Ling Liu Title: Virus: Harmful Fine-tuning Attack for Large Language Models Bypassing Guardrail Moderation Arxiv: http://arxiv.org/abs/2501.17433v1 Abstract: Recent research shows that Large Language Models (LLMs) are vulnerable to harmful fine-tuning attacks -- models lose their safety alignment ability after fine-tuning on a few harmful samples. For risk mitigation, a guardrail is typically used to filter out harmful samples before fine-tuning. By designing a new red-teaming method, we in this paper show that purely relying on the moderation guardrail for data filtration is not reliable. Our proposed attack method, dubbed Virus, easily bypasses the guardrail moderation by slightly modifying the harmful data. Experimental results show that the harmful data optimized by Virus is not detectable by the guardrail with up to 100\% leakage ratio, and can simultaneously achieve superior attack performance. Finally, the key message we want to convey through this paper is that: \textbf{it is reckless to consider guardrail moderation as a clutch at straws towards harmful fine-tuning attack}, as it cannot solve the inherent safety issue of the pre-trained LLMs. Our code is available at https://github.com/git-disl/Virus

🤗 Upvotes: 6 | cs.CL, cs.AI Authors: Jenna Russell, Marzena Karpinska, Mohit Iyyer Title: People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text Arxiv: http://arxiv.org/abs/2501.15654v1 Abstract: In this paper, we study how well humans can detect text generated by commercial LLMs (GPT-4o, Claude, o1). We hire annotators to read 300 non-fiction English articles, label them as either human-written or AI-generated, and provide paragraph-length explanations for their decisions. Our experiments show that annotators who frequently use LLMs for writing tasks excel at detecting AI-generated text, even without any specialized training or feedback. In fact, the majority vote among five such "expert" annotators misclassifies only 1 of 300 articles, significantly outperforming most commercial and open-source detectors we evaluated even in the presence of evasion tactics like paraphrasing and humanization. Qualitative analysis of the experts' free-form explanations shows that while they rely heavily on specific lexical clues ('AI vocabulary'), they also pick up on more complex phenomena within the text (e.g., formality, originality, clarity) that are challenging to assess for automatic detectors. We release our annotated dataset and code to spur future research into both human and automated detection of AI-generated text.

🤗 Upvotes: 29 | cs.AI, cs.CV, cs.LG Authors: Tianzhe Chu, Yuexiang Zhai, Jihan Yang, Shengbang Tong, Saining Xie, Dale Schuurmans, Quoc V. Le, Sergey Levine, Yi Ma Title: SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training Arxiv: http://arxiv.org/abs/2501.17161v1 Abstract: Supervised fine-tuning (SFT) and reinforcement learning (RL) are widely used post-training techniques for foundation models. However, their roles in enhancing model generalization capabilities remain unclear. This paper studies the difference between SFT and RL on generalization and memorization, focusing on text-based rule variants and visual variants. We introduce GeneralPoints, an arithmetic reasoning card game, and adopt V-IRL, a real-world navigation environment, to assess how models trained with SFT and RL generalize to unseen variants in both textual and visual domains. We show that RL, especially when trained with an outcome-based reward, generalizes across both rule-based textual and visual variants. SFT, in contrast, tends to memorize training data and struggles to generalize out-of-distribution scenarios. Further analysis reveals that RL improves the model's underlying visual recognition capabilities, contributing to its enhanced generalization in the visual domain. Despite RL's superior generalization, we show that SFT remains essential for effective RL training; SFT stabilizes the model's output format, enabling subsequent RL to achieve its performance gains. These findings demonstrates the capability of RL for acquiring generalizable knowledge in complex, multi-modal tasks.

🤗 Upvotes: 15 | cs.LG, cs.CL Authors: Ruizhe Wang, Yeyun Gong, Xiao Liu, Guoshuai Zhao, Ziyue Yang, Baining Guo, Zhengjun Zha, Peng Cheng Title: Optimizing Large Language Model Training Using FP4 Quantization Arxiv: http://arxiv.org/abs/2501.17116v1 Abstract: The growing computational demands of training large language models (LLMs) necessitate more efficient methods. Quantized training presents a promising solution by enabling low-bit arithmetic operations to reduce these costs. While FP8 precision has demonstrated feasibility, leveraging FP4 remains a challenge due to significant quantization errors and limited representational capacity. This work introduces the first FP4 training framework for LLMs, addressing these challenges with two key innovations: a differentiable quantization estimator for precise weight updates and an outlier clamping and compensation strategy to prevent activation collapse. To ensure stability, the framework integrates a mixed-precision training scheme and vector-wise quantization. Experimental results demonstrate that our FP4 framework achieves accuracy comparable to BF16 and FP8, with minimal degradation, scaling effectively to 13B-parameter LLMs trained on up to 100B tokens. With the emergence of next-generation hardware supporting FP4, our framework sets a foundation for efficient ultra-low precision training.

🤗 Upvotes: 11 | cs.CV Authors: Chenguo Lin, Panwang Pan, Bangbang Yang, Zeming Li, Yadong Mu Title: DiffSplat: Repurposing Image Diffusion Models for Scalable Gaussian Splat Generation Arxiv: http://arxiv.org/abs/2501.16764v1 Abstract: Recent advancements in 3D content generation from text or a single image struggle with limited high-quality 3D datasets and inconsistency from 2D multi-view generation. We introduce DiffSplat, a novel 3D generative framework that natively generates 3D Gaussian splats by taming large-scale text-to-image diffusion models. It differs from previous 3D generative models by effectively utilizing web-scale 2D priors while maintaining 3D consistency in a unified model. To bootstrap the training, a lightweight reconstruction model is proposed to instantly produce multi-view Gaussian splat grids for scalable dataset curation. In conjunction with the regular diffusion loss on these grids, a 3D rendering loss is introduced to facilitate 3D coherence across arbitrary views. The compatibility with image diffusion models enables seamless adaptions of numerous techniques for image generation to the 3D realm. Extensive experiments reveal the superiority of DiffSplat in text- and image-conditioned generation tasks and downstream applications. Thorough ablation studies validate the efficacy of each critical design choice and provide insights into the underlying mechanism.

🤗 Upvotes: 10 | cs.CL, cs.LG Authors: Hongzhi Huang, Defa Zhu, Banggu Wu, Yutao Zeng, Ya Wang, Qiyang Min, Xun Zhou Title: Over-Tokenized Transformer: Vocabulary is Generally Worth Scaling Arxiv: http://arxiv.org/abs/2501.16975v1 Abstract: Tokenization is a fundamental component of large language models (LLMs), yet its influence on model scaling and performance is not fully explored. In this paper, we introduce Over-Tokenized Transformers, a novel framework that decouples input and output vocabularies to improve language modeling performance. Specifically, our approach scales up input vocabularies to leverage multi-gram tokens. Through extensive experiments, we uncover a log-linear relationship between input vocabulary size and training loss, demonstrating that larger input vocabularies consistently enhance model performance, regardless of model size. Using a large input vocabulary, we achieve performance comparable to double-sized baselines with no additional cost. Our findings highlight the importance of tokenization in scaling laws and provide practical insight for tokenizer design, paving the way for more efficient and powerful LLMs.

🤗 Upvotes: 10 | cs.LG Authors: Lee Sharkey, Bilal Chughtai, Joshua Batson, Jack Lindsey, Jeff Wu, Lucius Bushnaq, Nicholas Goldowsky-Dill, Stefan Heimersheim, Alejandro Ortega, Joseph Bloom, Stella Biderman, Adria Garriga-Alonso, Arthur Conmy, Neel Nanda, Jessica Rumbelow, Martin Wattenberg, Nandi Schoots, Joseph Miller, Eric J. Michaud, Stephen Casper, Max Tegmark, William Saunders, David Bau, Eric Todd, Atticus Geiger, Mor Geva, Jesse Hoogland, Daniel Murfet, Tom McGrath Title: Open Problems in Mechanistic Interpretability Arxiv: http://arxiv.org/abs/2501.16496v1 Abstract: Mechanistic interpretability aims to understand the computational mechanisms underlying neural networks' capabilities in order to accomplish concrete scientific and engineering goals. Progress in this field thus promises to provide greater assurance over AI system behavior and shed light on exciting scientific questions about the nature of intelligence. Despite recent progress toward these goals, there are many open problems in the field that require solutions before many scientific and practical benefits can be realized: Our methods require both conceptual and practical improvements to reveal deeper insights; we must figure out how best to apply our methods in pursuit of specific goals; and the field must grapple with socio-technical challenges that influence and are influenced by our work. This forward-facing review discusses the current frontier of mechanistic interpretability and the open problems that the field may benefit from prioritizing.

🤗 Upvotes: 5 | cs.LG, cs.AI, cs.CL Authors: J. Pablo Muñoz, Jinjie Yuan, Nilesh Jain Title: Low-Rank Adapters Meet Neural Architecture Search for LLM Compression Arxiv: http://arxiv.org/abs/2501.16372v1 Abstract: The rapid expansion of Large Language Models (LLMs) has posed significant challenges regarding the computational resources required for fine-tuning and deployment. Recent advancements in low-rank adapters have demonstrated their efficacy in parameter-efficient fine-tuning (PEFT) of these models. This retrospective paper comprehensively discusses innovative approaches that synergize low-rank representations with Neural Architecture Search (NAS) techniques, particularly weight-sharing super-networks. Robust solutions for compressing and fine-tuning large pre-trained models are developed by integrating these methodologies. Our analysis highlights the potential of these combined strategies to democratize the use of LLMs, making them more accessible for deployment in resource-constrained environments. The resulting models exhibit reduced memory footprints and faster inference times, paving the way for more practical and scalable applications of LLMs. Models and code are available at https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning.

🤗 Upvotes: 4 | cs.CL, cs.AI Authors: Sankalp KJ, Ashutosh Kumar, Laxmaan Balaji, Nikunj Kotecha, Vinija Jain, Aman Chadha, Sreyoshi Bhaduri Title: IndicMMLU-Pro: Benchmarking Indic Large Language Models on Multi-Task Language Understanding Arxiv: http://arxiv.org/abs/2501.15747v2 Abstract: Known by more than 1.5 billion people in the Indian subcontinent, Indic languages present unique challenges and opportunities for natural language processing (NLP) research due to their rich cultural heritage, linguistic diversity, and complex structures. IndicMMLU-Pro is a comprehensive benchmark designed to evaluate Large Language Models (LLMs) across Indic languages, building upon the MMLU Pro (Massive Multitask Language Understanding) framework. Covering major languages such as Hindi, Bengali, Gujarati, Marathi, Kannada, Punjabi, Tamil, Telugu, and Urdu, our benchmark addresses the unique challenges and opportunities presented by the linguistic diversity of the Indian subcontinent. This benchmark encompasses a wide range of tasks in language comprehension, reasoning, and generation, meticulously crafted to capture the intricacies of Indian languages. IndicMMLU-Pro provides a standardized evaluation framework to push the research boundaries in Indic language AI, facilitating the development of more accurate, efficient, and culturally sensitive models. This paper outlines the benchmarks' design principles, task taxonomy, and data collection methodology, and presents baseline results from state-of-the-art multilingual models.

🤗 Upvotes: 3 | cs.CL, cs.AI Authors: Thibaud Leteno, Irina Proskurina, Antoine Gourru, Julien Velcin, Charlotte Laclau, Guillaume Metzler, Christophe Gravier Title: Histoires Morales: A French Dataset for Assessing Moral Alignment Arxiv: http://arxiv.org/abs/2501.17117v1 Abstract: Aligning language models with human values is crucial, especially as they become more integrated into everyday life. While models are often adapted to user preferences, it is equally important to ensure they align with moral norms and behaviours in real-world social situations. Despite significant progress in languages like English and Chinese, French has seen little attention in this area, leaving a gap in understanding how LLMs handle moral reasoning in this language. To address this gap, we introduce Histoires Morales, a French dataset derived from Moral Stories, created through translation and subsequently refined with the assistance of native speakers to guarantee grammatical accuracy and adaptation to the French cultural context. We also rely on annotations of the moral values within the dataset to ensure their alignment with French norms. Histoires Morales covers a wide range of social situations, including differences in tipping practices, expressions of honesty in relationships, and responsibilities toward animals. To foster future research, we also conduct preliminary experiments on the alignment of multilingual models on French and English data and the robustness of the alignment. We find that while LLMs are generally aligned with human moral norms by default, they can be easily influenced with user-preference optimization for both moral and immoral data.

🤗 Upvotes: 26 | cs.CL Authors: An Yang, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoyan Huang, Jiandong Jiang, Jianhong Tu, Jianwei Zhang, Jingren Zhou, Junyang Lin, Kai Dang, Kexin Yang, Le Yu, Mei Li, Minmin Sun, Qin Zhu, Rui Men, Tao He, Weijia Xu, Wenbiao Yin, Wenyuan Yu, Xiafei Qiu, Xingzhang Ren, Xinlong Yang, Yong Li, Zhiying Xu, Zipeng Zhang Title: Qwen2.5-1M Technical Report Arxiv: http://arxiv.org/abs/2501.15383v1 Abstract: We introduce Qwen2.5-1M, a series of models that extend the context length to 1 million tokens. Compared to the previous 128K version, the Qwen2.5-1M series have significantly enhanced long-context capabilities through long-context pre-training and post-training. Key techniques such as long data synthesis, progressive pre-training, and multi-stage supervised fine-tuning are employed to effectively enhance long-context performance while reducing training costs. To promote the use of long-context models among a broader user base, we present and open-source our inference framework. This framework includes a length extrapolation method that can expand the model context lengths by at least four times, or even more, without additional training. To reduce inference costs, we implement a sparse attention method along with chunked prefill optimization for deployment scenarios and a sparsity refinement method to improve precision. Additionally, we detail our optimizations in the inference engine, including kernel optimization, pipeline parallelism, and scheduling optimization, which significantly enhance overall inference performance. By leveraging our inference framework, the Qwen2.5-1M models achieve a remarkable 3x to 7x prefill speedup in scenarios with 1 million tokens of context. This framework provides an efficient and powerful solution for developing applications that require long-context processing using open-source models. The Qwen2.5-1M series currently includes the open-source models Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M, as well as the API-accessed model Qwen2.5-Turbo. Evaluations show that Qwen2.5-1M models have been greatly improved in long-context tasks without compromising performance in short-context scenarios. Specifically, the Qwen2.5-14B-Instruct-1M model significantly outperforms GPT-4o-mini in long-context tasks and supports contexts eight times longer.

🤗 Upvotes: 13 | cs.CL Authors: Lin Yueyu, Li Zhiyuan, Peter Yue, Liu Xiao Title: ARWKV: Pretrain is not what we need, an RNN-Attention-Based Language Model Born from Transformer Arxiv: http://arxiv.org/abs/2501.15570v1 Abstract: As is known, hybrid quadratic and subquadratic attention models in multi-head architectures have surpassed both Transformer and Linear RNN models , with these works primarily focusing on reducing KV complexity and improving efficiency. For further research on expressiveness, we introduce our series of models distilled from Qwen 2.5, based on pure native RWKV-7 attention, which aims to make RNN more expressive and demonstrates state tracking ability beyond transformers. We work with QRWK 32B based on RWKV-6 architecture, another approach that reduces the entire knowledge processing time to just 8 hours using 16 AMD MI300X GPUs while maintaining Qwen 2.5's performance. In fact, the distillation process can utilize any LLM, not just Qwen, and enables knowledge transfer from larger LLMs to smaller ones with more fewer tokens. We will explain the detailed process and share our insights on building more powerful foundation models. Please note that this is an ongoing work that will be updated continuously. The model checkpoints and source code are available at \href{https://github.com/yynil/RWKVInside}{https://github.com/yynil/RWKVInside}, \href{https://huggingface.co/RWKV-Red-Team/ARWKV-7B-Preview-0.1}{https://huggingface.co/RWKV-Red-Team/ARWKV-7B-Preview-0.1}.

🤗 Upvotes: 13 | cs.LG, cs.AI Authors: Scott Fujimoto, Pierluca D'Oro, Amy Zhang, Yuandong Tian, Michael Rabbat Title: Towards General-Purpose Model-Free Reinforcement Learning Arxiv: http://arxiv.org/abs/2501.16142v1 Abstract: Reinforcement learning (RL) promises a framework for near-universal problem-solving. In practice however, RL algorithms are often tailored to specific benchmarks, relying on carefully tuned hyperparameters and algorithmic choices. Recently, powerful model-based RL methods have shown impressive general results across benchmarks but come at the cost of increased complexity and slow run times, limiting their broader applicability. In this paper, we attempt to find a unifying model-free deep RL algorithm that can address a diverse class of domains and problem settings. To achieve this, we leverage model-based representations that approximately linearize the value function, taking advantage of the denser task objectives used by model-based RL while avoiding the costs associated with planning or simulated trajectories. We evaluate our algorithm, MR.Q, on a variety of common RL benchmarks with a single set of hyperparameters and show a competitive performance against domain-specific and general baselines, providing a concrete step towards building general-purpose model-free deep RL algorithms.

🤗 Upvotes: 11 | cs.SD, cs.CL, eess.AS Authors: Haorui He, Zengqiang Shang, Chaoren Wang, Xuyuan Li, Yicheng Gu, Hua Hua, Liwei Liu, Chen Yang, Jiaqi Li, Peiyang Shi, Yuancheng Wang, Kai Chen, Pengyuan Zhang, Zhizheng Wu Title: Emilia: A Large-Scale, Extensive, Multilingual, and Diverse Dataset for Speech Generation Arxiv: http://arxiv.org/abs/2501.15907v1 Abstract: Recent advancements in speech generation have been driven by the large-scale training datasets. However, current models fall short of capturing the spontaneity and variability inherent in real-world human speech, due to their reliance on audiobook datasets limited to formal read-aloud speech styles. To bridge this gap, we introduce Emilia-Pipe, an open-source preprocessing pipeline to extract high-quality training data from valuable yet underexplored in-the-wild data that capture spontaneous human speech in real-world contexts. By leveraging Emilia-Pipe, we construct Emilia, the first multilingual speech generation dataset derived from in-the-wild speech data. This dataset comprises over 101k hours of speech across six languages: English, Chinese, German, French, Japanese, and Korean. Besides, we expand Emilia to Emilia-Large, a dataset exceeding 216k hours, making it the largest open-source speech generation dataset available. Extensive experiments demonstrate that Emilia significantly outperforms traditional audiobook datasets in generating spontaneous and human-like speech, showcasing superior performance in capturing diverse speaker timbre and speaking styles of real-world human speech. Furthermore, this work underscores the importance of scaling dataset size to advance speech generation research and validates the effectiveness of Emilia for both multilingual and crosslingual speech generation.

🤗 Upvotes: 9 | cs.CV, cs.AI Authors: Chuanyang Zheng Title: iFormer: Integrating ConvNet and Transformer for Mobile Application Arxiv: http://arxiv.org/abs/2501.15369v1 Abstract: We present a new family of mobile hybrid vision networks, called iFormer, with a focus on optimizing latency and accuracy on mobile applications. iFormer effectively integrates the fast local representation capacity of convolution with the efficient global modeling ability of self-attention. The local interactions are derived from transforming a standard convolutional network, \textit{i.e.}, ConvNeXt, to design a more lightweight mobile network. Our newly introduced mobile modulation attention removes memory-intensive operations in MHA and employs an efficient modulation mechanism to boost dynamic global representational capacity. We conduct comprehensive experiments demonstrating that iFormer outperforms existing lightweight networks across various tasks. Notably, iFormer achieves an impressive Top-1 accuracy of 80.4\% on ImageNet-1k with a latency of only 1.10 ms on an iPhone 13, surpassing the recently proposed MobileNetV4 under similar latency constraints. Additionally, our method shows significant improvements in downstream tasks, including COCO object detection, instance segmentation, and ADE20k semantic segmentation, while still maintaining low latency on mobile devices for high-resolution inputs in these scenarios.

🤗 Upvotes: 7 | cs.CV, cs.AI, cs.LG, q-bio.NC Authors: Paul Gavrikov, Jovita Lukasik, Steffen Jung, Robert Geirhos, Bianca Lamm, Muhammad Jehanzeb Mirza, Margret Keuper, Janis Keuper Title: Are Vision Language Models Texture or Shape Biased and Can We Steer Them? Arxiv: http://arxiv.org/abs/2403.09193v1 Abstract: Vision language models (VLMs) have drastically changed the computer vision model landscape in only a few years, opening an exciting array of new applications from zero-shot image classification, over to image captioning, and visual question answering. Unlike pure vision models, they offer an intuitive way to access visual content through language prompting. The wide applicability of such models encourages us to ask whether they also align with human vision - specifically, how far they adopt human-induced visual biases through multimodal fusion, or whether they simply inherit biases from pure vision models. One important visual bias is the texture vs. shape bias, or the dominance of local over global information. In this paper, we study this bias in a wide range of popular VLMs. Interestingly, we find that VLMs are often more shape-biased than their vision encoders, indicating that visual biases are modulated to some extent through text in multimodal models. If text does indeed influence visual biases, this suggests that we may be able to steer visual biases not just through visual input but also through language: a hypothesis that we confirm through extensive experiments. For instance, we are able to steer shape bias from as low as 49% to as high as 72% through prompting alone. For now, the strong human bias towards shape (96%) remains out of reach for all tested VLMs.

🤗 Upvotes: 5 | cs.LG Authors: Ryan Ehrlich, Bradley Brown, Jordan Juravsky, Ronald Clark, Christopher Ré, Azalia Mirhoseini Title: CodeMonkeys: Scaling Test-Time Compute for Software Engineering Arxiv: http://arxiv.org/abs/2501.14723v1 Abstract: Scaling test-time compute is a promising axis for improving LLM capabilities. However, test-time compute can be scaled in a variety of ways, and effectively combining different approaches remains an active area of research. Here, we explore this problem in the context of solving real-world GitHub issues from the SWE-bench dataset. Our system, named CodeMonkeys, allows models to iteratively edit a codebase by jointly generating and running a testing script alongside their draft edit. We sample many of these multi-turn trajectories for every issue to generate a collection of candidate edits. This approach lets us scale "serial" test-time compute by increasing the number of iterations per trajectory and "parallel" test-time compute by increasing the number of trajectories per problem. With parallel scaling, we can amortize up-front costs across multiple downstream samples, allowing us to identify relevant codebase context using the simple method of letting an LLM read every file. In order to select between candidate edits, we combine voting using model-generated tests with a final multi-turn trajectory dedicated to selection. Overall, CodeMonkeys resolves 57.4% of issues from SWE-bench Verified using a budget of approximately 2300 USD. Our selection method can also be used to combine candidates from different sources. Selecting over an ensemble of edits from existing top SWE-bench Verified submissions obtains a score of 66.2% and outperforms the best member of the ensemble on its own. We fully release our code and data at https://scalingintelligence.stanford.edu/pubs/codemonkeys.

🤗 Upvotes: 4 | cs.LG, cs.AI Authors: Samira Abnar, Harshay Shah, Dan Busbridge, Alaaeldin Mohamed Elnouby Ali, Josh Susskind, Vimal Thilak Title: Parameters vs FLOPs: Scaling Laws for Optimal Sparsity for Mixture-of-Experts Language Models Arxiv: http://arxiv.org/abs/2501.12370v2 Abstract: Scaling the capacity of language models has consistently proven to be a reliable approach for improving performance and unlocking new capabilities. Capacity can be primarily defined by two dimensions: the number of model parameters and the compute per example. While scaling typically involves increasing both, the precise interplay between these factors and their combined contribution to overall capacity remains not fully understood. We explore this relationship in the context of sparse Mixture-of-Experts (MoEs), which allow scaling the number of parameters without proportionally increasing the FLOPs per example. We investigate how varying the sparsity level, i.e., the fraction of inactive parameters, impacts model's performance during pretraining and downstream few-shot evaluation. We find that under different constraints (e.g., parameter size and total training compute), there is an optimal level of sparsity that improves both training efficiency and model performance. These results provide a better understanding of the impact of sparsity in scaling laws for MoEs and complement existing works in this area, offering insights for designing more efficient architectures.

🤗 Upvotes: 33 | cs.LG, cs.AI, cs.CL Authors: Long Phan, Alice Gatti, Ziwen Han, Nathaniel Li, Josephina Hu, Hugh Zhang, Sean Shi, Michael Choi, Anish Agrawal, Arnav Chopra, Adam Khoja, Ryan Kim, Jason Hausenloy, Oliver Zhang, Mantas Mazeika, Daron Anderson, Tung Nguyen, Mobeen Mahmood, Fiona Feng, Steven Y. Feng, Haoran Zhao, Michael Yu, Varun Gangal, Chelsea Zou, Zihan Wang, Jessica P. Wang, Pawan Kumar, Oleksandr Pokutnyi, Robert Gerbicz, Serguei Popov, John-Clark Levin, Mstyslav Kazakov, Johannes Schmitt, Geoff Galgon, Alvaro Sanchez, Yongki Lee, Will Yeadon, Scott Sauers, Marc Roth, Chidozie Agu, Søren Riis, Fabian Giska, Saiteja Utpala, Zachary Giboney, Gashaw M. Goshu, Joan of Arc Xavier, Sarah-Jane Crowson, Mohinder Maheshbhai Naiya, Noah Burns, Lennart Finke, Zerui Cheng, Hyunwoo Park, Francesco Fournier-Facio, John Wydallis, Mark Nandor, Ankit Singh, Tim Gehrunger, Jiaqi Cai, Ben McCarty, Darling Duclosel, Jungbae Nam, Jennifer Zampese, Ryan G. Hoerr, Aras Bacho, Gautier Abou Loume, Abdallah Galal, Hangrui Cao, Alexis C Garretson, Damien Sileo, Qiuyu Ren, Doru Cojoc, Pavel Arkhipov, Usman Qazi, Lianghui Li, Sumeet Motwani, Christian Schroeder de Witt, Edwin Taylor, Johannes Veith, Eric Singer, Taylor D. Hartman, Paolo Rissone, Jaehyeok Jin, Jack Wei Lun Shi, Chris G. Willcocks, Joshua Robinson, Aleksandar Mikov, Ameya Prabhu, Longke Tang, Xavier Alapont, Justine Leon Uro, Kevin Zhou, Emily de Oliveira Santos, Andrey Pupasov Maksimov, Edward Vendrow, Kengo Zenitani, Julien Guillod, Yuqi Li, Joshua Vendrow, Vladyslav Kuchkin, Ng Ze-An, Pierre Marion, Denis Efremov, Jayson Lynch, Kaiqu Liang, Andrew Gritsevskiy, Dakotah Martinez, Ben Pageler, Nick Crispino, Dimitri Zvonkine, Natanael Wildner Fraga, Saeed Soori, Ori Press, Henry Tang, Julian Salazar, Sean R. Green, Lina Brüssel, Moon Twayana, Aymeric Dieuleveut, T. Ryan Rogers, Wenjin Zhang, Bikun Li, Jinzhou Yang, Arun Rao, Gabriel Loiseau, Mikhail Kalinin, Marco Lukas, Ciprian Manolescu, Subrata Mishra, Ariel Ghislain Kemogne Kamdoum, Tobias Kreiman, Tad Hogg, Alvin Jin, Carlo Bosio, Gongbo Sun, Brian P Coppola, Tim Tarver, Haline Heidinger, Rafael Sayous, Stefan Ivanov, Joseph M Cavanagh, Jiawei Shen, Joseph Marvin Imperial, Philippe Schwaller, Shaipranesh Senthilkuma, Andres M Bran, Ali Dehghan, Andres Algaba, Brecht Verbeken, David Noever, Ragavendran P V, Lisa Schut, Ilia Sucholutsky, Evgenii Zheltonozhskii, Derek Lim, Richard Stanley, Shankar Sivarajan, Tong Yang, John Maar, Julian Wykowski, Martí Oller, Jennifer Sandlin, Anmol Sahu, Yuzheng Hu, Sara Fish, Nasser Heydari, Archimedes Apronti, Kaivalya Rawal, Tobias Garcia Vilchis, Yuexuan Zu, Martin Lackner, James Koppel, Jeremy Nguyen, Daniil S. Antonenko, Steffi Chern, Bingchen Zhao, Pierrot Arsene, Alan Goldfarb, Sergey Ivanov, Rafał Poświata, Chenguang Wang, Daofeng Li, Donato Crisostomi, Andrea Achilleos, Benjamin Myklebust, Archan Sen, David Perrella, Nurdin Kaparov, Mark H Inlow, Allen Zang, Elliott Thornley, Daniil Orel, Vladislav Poritski, Shalev Ben-David, Zachary Berger, Parker Whitfill, Michael Foster, Daniel Munro, Linh Ho, Dan Bar Hava, Aleksey Kuchkin, Robert Lauff, David Holmes, Frank Sommerhage, Keith Schneider, Zakayo Kazibwe, Nate Stambaugh, Mukhwinder Singh, Ilias Magoulas, Don Clarke, Dae Hyun Kim, Felipe Meneguitti Dias, Veit Elser, Kanu Priya Agarwal, Victor Efren Guadarrama Vilchis, Immo Klose, Christoph Demian, Ujjwala Anantheswaran, Adam Zweiger, Guglielmo Albani, Jeffery Li, Nicolas Daans, Maksim Radionov, Václav Rozhoň, Ziqiao Ma, Christian Stump, Mohammed Berkani, Jacob Platnick, Volodymyr Nevirkovets, Luke Basler, Marco Piccardo, Ferenc Jeanplong, Niv Cohen, Josef Tkadlec, Paul Rosu, Piotr Padlewski, Stanislaw Barzowski, Kyle Montgomery, Aline Menezes, Arkil Patel, Zixuan Wang, Jamie Tucker-Foltz, Jack Stade, Tom Goertzen, Fereshteh Kazemi, Jeremiah Milbauer, John Arnold Ambay, Abhishek Shukla, Yan Carlos Leyva Labrador, Alan Givré, Hew Wolff, Vivien Rossbach, Muhammad Fayez Aziz, Younesse Kaddar, Yanxu Chen, Robin Zhang, Jiayi Pan, Antonio Terpin, Niklas Muennighoff, Hailey Schoelkopf, Eric Zheng, Avishy Carmi, Adam Jones, Jainam Shah, Ethan D. L. Brown, Kelin Zhu, Max Bartolo, Richard Wheeler, Andrew Ho, Shaul Barkan, Jiaqi Wang, Martin Stehberger, Egor Kretov, Kaustubh Sridhar, Zienab EL-Wasif, Anji Zhang, Daniel Pyda, Joanna Tam, David M. Cunningham, Vladimir Goryachev, Demosthenes Patramanis, Michael Krause, Andrew Redenti, Daniel Bugas, David Aldous, Jesyin Lai, Shannon Coleman, Mohsen Bahaloo, Jiangnan Xu, Sangwon Lee, Sandy Zhao, Ning Tang, Michael K. Cohen, Micah Carroll, Orr Paradise, Jan Hendrik Kirchner, Stefan Steinerberger, Maksym Ovchynnikov, Jason O. Matos, Adithya Shenoy, Benedito Alves de Oliveira Junior, Michael Wang, Yuzhou Nie, Paolo Giordano, Philipp Petersen, Anna Sztyber-Betley, Priti Shukla, Jonathan Crozier, Antonella Pinto, Shreyas Verma, Prashant Joshi, Zheng-Xin Yong, Allison Tee, Jérémy Andréoletti, Orion Weller, Raghav Singhal, Gang Zh

🤗 Upvotes: 26 | cs.IR, cs.CL Authors: Liang Wang, Haonan Chen, Nan Yang, Xiaolong Huang, Zhicheng Dou, Furu Wei Title: Chain-of-Retrieval Augmented Generation Arxiv: http://arxiv.org/abs/2501.14342v1 Abstract: This paper introduces an approach for training o1-like RAG models that retrieve and reason over relevant information step by step before generating the final answer. Conventional RAG methods usually perform a single retrieval step before the generation process, which limits their effectiveness in addressing complex queries due to imperfect retrieval results. In contrast, our proposed method, CoRAG (Chain-of-Retrieval Augmented Generation), allows the model to dynamically reformulate the query based on the evolving state. To train CoRAG effectively, we utilize rejection sampling to automatically generate intermediate retrieval chains, thereby augmenting existing RAG datasets that only provide the correct final answer. At test time, we propose various decoding strategies to scale the model's test-time compute by controlling the length and number of sampled retrieval chains. Experimental results across multiple benchmarks validate the efficacy of CoRAG, particularly in multi-hop question answering tasks, where we observe more than 10 points improvement in EM score compared to strong baselines. On the KILT benchmark, CoRAG establishes a new state-of-the-art performance across a diverse range of knowledge-intensive tasks. Furthermore, we offer comprehensive analyses to understand the scaling behavior of CoRAG, laying the groundwork for future research aimed at developing factual and grounded foundation models.

🤗 Upvotes: 22 | cs.CL, cs.AI Authors: Zicheng Zhang, Xiangyu Zhao, Xinyu Fang, Chunyi Li, Xiaohong Liu, Xiongkuo Min, Haodong Duan, Kai Chen, Guangtao Zhai Title: Redundancy Principles for MLLMs Benchmarks Arxiv: http://arxiv.org/abs/2501.13953v1 Abstract: With the rapid iteration of Multi-modality Large Language Models (MLLMs) and the evolving demands of the field, the number of benchmarks produced annually has surged into the hundreds. The rapid growth has inevitably led to significant redundancy among benchmarks. Therefore, it is crucial to take a step back and critically assess the current state of redundancy and propose targeted principles for constructing effective MLLM benchmarks. In this paper, we focus on redundancy from three key perspectives: 1) Redundancy of benchmark capability dimensions, 2) Redundancy in the number of test questions, and 3) Cross-benchmark redundancy within specific domains. Through the comprehensive analysis over hundreds of MLLMs' performance across more than 20 benchmarks, we aim to quantitatively measure the level of redundancy lies in existing MLLM evaluations, provide valuable insights to guide the future development of MLLM benchmarks, and offer strategies to refine and address redundancy issues effectively.

🤗 Upvotes: 13 | cs.CL, cs.AI, cs.LG Authors: Zhengyang Tang, Ziniu Li, Zhenyang Xiao, Tian Ding, Ruoyu Sun, Benyou Wang, Dayiheng Liu, Fei Huang, Tianyu Liu, Bowen Yu, Junyang Lin Title: RealCritic: Towards Effectiveness-Driven Evaluation of Language Model Critiques Arxiv: http://arxiv.org/abs/2501.14492v1 Abstract: Critiques are important for enhancing the performance of Large Language Models (LLMs), enabling both self-improvement and constructive feedback for others by identifying flaws and suggesting improvements. However, evaluating the critique capabilities of LLMs presents a significant challenge due to the open-ended nature of the task. In this work, we introduce a new benchmark designed to assess the critique capabilities of LLMs. Unlike existing benchmarks, which typically function in an open-loop fashion, our approach employs a closed-loop methodology that evaluates the quality of corrections generated from critiques. Moreover, the benchmark incorporates features such as self-critique, cross-critique, and iterative critique, which are crucial for distinguishing the abilities of advanced reasoning models from more classical ones. We implement this benchmark using eight challenging reasoning tasks. We have several interesting findings. First, despite demonstrating comparable performance in direct chain-of-thought generation, classical LLMs significantly lag behind the advanced reasoning-based model o1-mini across all critique scenarios. Second, in self-critique and iterative critique settings, classical LLMs may even underperform relative to their baseline capabilities. We hope that this benchmark will serve as a valuable resource to guide future advancements. The code and data are available at \url{https://github.com/tangzhy/RealCritic}.

🤗 Upvotes: 7 | cs.LG, cs.AI Authors: Micah Rentschler, Jesse Roberts Title: RL + Transformer = A General-Purpose Problem Solver Arxiv: http://arxiv.org/abs/2501.14176v1 Abstract: What if artificial intelligence could not only solve problems for which it was trained but also learn to teach itself to solve new problems (i.e., meta-learn)? In this study, we demonstrate that a pre-trained transformer fine-tuned with reinforcement learning over multiple episodes develops the ability to solve problems that it has never encountered before - an emergent ability called In-Context Reinforcement Learning (ICRL). This powerful meta-learner not only excels in solving unseen in-distribution environments with remarkable sample efficiency, but also shows strong performance in out-of-distribution environments. In addition, we show that it exhibits robustness to the quality of its training data, seamlessly stitches together behaviors from its context, and adapts to non-stationary environments. These behaviors demonstrate that an RL-trained transformer can iteratively improve upon its own solutions, making it an excellent general-purpose problem solver.