이진욱님의 블로그

https://arxiv.org/pdf/2411.00855

https://arxiv.org/pdf/2411.10442Here's the text with each sentence on a new line: Existing open-source multimodal large language models (MLLMs) generally follow a training process involving pretraining and supervised fine-tuning. However, these models suffer from distribution shifts, which limit their multimodal reasoning, particularly in the Chain-of-Thought (CoT) performance. To address this, ..

https://arxiv.org/abs/2411.11694 Technical Report: Enhancing LLM Reasoning with Reward-guided Tree SearchRecently, test-time scaling has garnered significant attention from the research community, largely due to the substantial advancements of the o1 model released by OpenAI. By allocating more computational resources during the inference phase, large languagarxiv.orgRecently, test-time scaling ..

https://arxiv.org/pdf/2411.06559Language agents have demonstrated promising capabilities in automating webbased tasks, though their current reactive approaches still underperform largely compared to humans. While incorporating advanced planning algorithms, particularly tree search methods, could enhance these agents' performance, implementing tree search directly on live websites poses significa..

https://arxiv.org/pdf/2411.11504The evolution of machine learning has increasingly prioritized the development of powerful models and more scalable supervision signals. However, the emergence of foundation models presents significant challenges in providing effective supervision signals necessary for further enhancing their capabilities. Consequently, there is an urgent need to explore novel sup..

CLIP is one of the most important multimodal foundational models today, aligning visual and textual signals into a shared feature space using a simple contrastive learning loss on large-scale image-text pairs.What powers CLIP's capabilities?The rich supervision signals provided by natural language — the carrier of human knowledge — shape a powerful cross-modal represen-tation space.As a result, ..

https://arxiv.org/abs/2411.10440    Large language models have demonstrated substantial advancements in reasoning capabilities, particularly through inference-time scaling, as illustrated by models such as OpenAI's o1.However, current Vision-Language Models (VLMs) often struggle to perform systematic and structured reasoning, especially when handling complex visual question-answering tasks.In th..

https://arxiv.org/pdf/2410.13298

https://arxiv.org/pdf/2411.00418 Reinforcement Learning from Human Feedback (RLHF) is a crucial technique for aligning language models with human preferences, playing a pivotal role in the success of conversational models like GPT-4, ChatGPT, and Llama 2. A core challenge in employing RLHF lies in training a reliable reward model (RM), which relies on high-quality labels typically provided by hu..

https://arxiv.org/abs/2410.21728Let me help you add line breaks between sentences from the abstract: Teaching large language models (LLMs) to generate text with citations to evidence sources can mitigate hallucinations and enhance verifiability in information-seeking systems. However, improving this capability requires highquality attribution data, which is costly and labor-intensive. Inspired b..