Author name cluster

Suhong Moon

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

4 papers

2 author rows

TMLR Journal 2025 Journal Article

Hard-Negative Prototype-Based Regularization for Few-Shot Class-Incremental Learning

Seongbeom Park
Hyunju Yun
Daewon Chae
Sungyoon Kim
Suhong Moon
Minwoo Kang
Seunghyun Park
Jinkyu Kim

Few-shot class-incremental learning (FSCIL)---involving abundant base training data followed by novel classes with limited labeled samples---poses challenges such as catastrophic forgetting and overfitting, leading to significant performance degradation across incremental sessions. As a remedy, recent work focuses on minimizing the interference of embeddings between base and incremental classes. However, previous studies have not explicitly considered variation in discriminative difficulty across samples and classes, leaving room for improvement: we observe that hard-negative (i.e., difficult to discriminate from the label) samples and classes significantly affect FSCIL performance, whereas easy ones have little impact. To this end, we propose a hard-negative prototype-based regularization approach that enhances discrimination between similar classes by imposing a penalty margin between each sample and its most similar class prototypes based on cosine similarity. To select hard-negative prototypes, we explore two distinct mining strategies: dynamic selection that leverages the model's decision boundary, and static selection that utilizes a pre-defined class-wise similarity matrix derived from external sources such as pre-trained models. We evaluate our approach on three widely used benchmarks, miniImageNet, CIFAR100, and CUB200, achieving state-of-the-art performance on each. Comprehensive analyses demonstrate that our proposed method enhances intra-class cohesion and inter-class separability of embeddings, both of which are crucial for FSCIL to better accommodate novel classes.

PDF Details

ICML Conference 2025 Conference Paper

Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks

Lutfi Eren Erdogan
Nicholas Lee
Sehoon Kim 0001
Suhong Moon
Hiroki Furuta
Gopala Anumanchipalli
Kurt Keutzer
Amir Gholami

Large language models (LLMs) have shown remarkable advancements in enabling language agents to tackle simple tasks. However, applying them for complex, multi-step, long-horizon tasks remains a challenge. Recent work have found success by separating high-level planning from low-level execution, which enables the model to effectively balance high-level planning objectives and low-level execution details. However, generating accurate plans remains difficult since LLMs are not inherently trained for this task. To address this, we propose Plan-and-Act, a novel framework that incorporates explicit planning into LLM-based agents and introduces a scalable method to enhance plan generation through a novel synthetic data generation method. Plan-and-Act consists of a Planner model which generates structured, high-level plans to achieve user goals, and an Executor model that translates these plans into environment-specific actions. To train the Planner effectively, we introduce a synthetic data generation method that annotates ground-truth trajectories with feasible plans, augmented with diverse and extensive examples to enhance generalization. We evaluate Plan-and-Act using web navigation as a representative long-horizon planning environment, demonstrating a state-of-the-art 57. 58% success rate on the WebArena-Lite benchmark as well as a text-only state-of-the-art 81. 36% success rate on WebVoyager.

Details

ICML Conference 2024 Conference Paper

An LLM Compiler for Parallel Function Calling

Sehoon Kim 0001
Suhong Moon
Ryan Tabrizi
Nicholas Lee
Michael W. Mahoney
Kurt Keutzer
Amir Gholami

The reasoning capabilities of the recent LLMs enable them to execute external function calls to overcome their inherent limitations, such as knowledge cutoffs, poor arithmetic skills, or lack of access to private data. This development has allowed LLMs to select and coordinate multiple functions based on the context to tackle more complex problems. However, current methods for function calling often require sequential reasoning and acting for each function which can result in high latency, cost, and sometimes inaccurate behavior. To address this, we introduce LLMCompiler, which executes functions in parallel to efficiently orchestrate multiple function calls. Drawing inspiration from the principles of classical compilers, LLMCompiler enables parallel function calling with three components: (i) a Function Calling Planner, formulating execution plans for function calling; (ii) a Task Fetching Unit, dispatching function calling tasks; and (iii) an Executor, executing these tasks in parallel. LLMCompiler automatically generates an optimized orchestration for the function calls and can be used with both open-source and closed-source models. We have benchmarked LLMCompiler on a range of tasks with different patterns of function calling. We observe consistent latency speedup of up to $3. 7 \times$, cost savings of up to $6. 7 \times$, and accuracy improvement of up to $\sim 9 %$ compared to ReAct. Our code is available at https: //github. com/SqueezeAILab/LLMCompiler.

Details

NeurIPS Conference 2023 Conference Paper

Speculative Decoding with Big Little Decoder

Sehoon Kim
Karttikeya Mangalam
Suhong Moon
Jitendra Malik
Michael W. Mahoney
Amir Gholami
Kurt Keutzer

The recent emergence of Large Language Models based on the Transformer architecture has enabled dramatic advancements in the field of Natural Language Processing. However, these models have long inference latency, which limits their deployment and makes them prohibitively expensive for various real-time applications. The inference latency is further exacerbated by autoregressive generative tasks, as models need to run iteratively to generate tokens sequentially without leveraging token-level parallelization. To address this, we propose Big Little Decoder (BiLD), a framework that can improve inference efficiency and latency for a wide range of text generation applications. The BiLD framework contains two models with different sizes that collaboratively generate text. The small model runs autoregressively to generate text with a low inference cost, and the large model is only invoked occasionally to refine the small model’s inaccurate predictions in a non-autoregressive manner. To coordinate the small and large models, BiLD introduces two simple yet effective policies: (1) the fallback policy that determines when to hand control over to the large model; and (2) the rollback policy that determines when the large model needs to correct the small model's inaccurate predictions. To evaluate our framework across different tasks and models, we apply BiLD to various text generation scenarios encompassing machine translation on IWSLT 2017 De-En and WMT 2014 De-En, and summarization on XSUM and CNN/DailyMail. On an NVIDIA T4 GPU, our framework achieves a speedup of up to 2. 12x speedup with minimal generation quality degradation. Furthermore, our framework is fully plug-and-play and can be applied without any modifications in the training process or model architecture. Our code is open-sourced.

PDF Details