arXiv subjects

Chenmien Tan

Publications and source records attributed to Chenmien Tan.

4 recordsLinked to original sources

arXiv2025

GEM: A Gym for Agentic LLMs

The training paradigm for large language models (LLMs) is moving from static datasets to experience-based learning, where agents acquire skills via interacting with complex environments. To facilitate this transition we introduce GEM (General Experience Maker), an open-source environment simulator designed for the age of LLMs. Analogous to OpenAI-Gym for traditional reinforcement learning (RL), GEM provides a standardized framework for the environment-agent interface, including asynchronous vectorized execution for high throughput, and flexible wrappers for easy extensibility. GEM also features a diverse suite of environments, robust integrated tools, and single-file example scripts demonstrating using GEM with five popular RL training frameworks. Along with this, we also provide a set of baselines across 24 environments using REINFORCE with Return Batch Normalization (ReBN), which -- unlike GRPO -- is compatible with the full RL setting of dense per-turn rewards and offers better credit assignment. We further conduct apple-to-apple benchmarking of PPO, GRPO and REINFORCE in both single- and multi-turn settings using GEM to shed light on the algorithmic designs. Lastly, GEM also functions as a convenient evaluation toolkit besides a training environment. We hope this framework can help accelerate future agentic LLM research.

cs.LG↗

arXiv2025

CHARM: Calibrating Reward Models With Chatbot Arena Scores

Reward models (RMs) play a crucial role in Reinforcement Learning from Human Feedback by serving as proxies for human preferences in aligning large language models. However, they suffer from various biases which could lead to reward hacking. In this paper, we identify a model preference bias in RMs, where they systematically assign disproportionately high scores to responses from certain policy models, leading to unfair judgments. To mitigate this bias, we propose a calibration method named CHatbot Arena calibrated Reward Modeling (CHARM) that leverages Elo scores from the Chatbot Arena to construct debiased preference datasets and adjust reward model scoring. We conduct extensive experiments on reward model benchmarks and human preference alignment. Results demonstrate that our calibrated RMs achieve improved evaluation accuracy on RM-Bench and the Chat-Hard domain of RewardBench, exhibit a stronger correlation with human preferences by producing scores more closely aligned with Elo rankings and improve downstream post-training performance. These results demonstrate that CHARM provides a simple, effective, and broadly applicable approach to building more reliable and fair reward models.

cs.AI↗

arXiv2024

Massive Editing for Large Language Models via Meta Learning

While large language models (LLMs) have enabled learning knowledge from the pre-training corpora, the acquired knowledge may be fundamentally incorrect or outdated over time, which necessitates rectifying the knowledge of the language model (LM) after the training. A promising approach involves employing a hyper-network to generate parameter shift, whereas existing hyper-networks suffer from inferior scalability in synchronous editing operation amount. To mitigate the problem, we propose the MAssive Language Model Editing Network (MALMEN), which formulates the parameter shift aggregation as the least square problem, subsequently updating the LM parameters using the normal equation. To accommodate editing multiple facts simultaneously with limited memory budgets, we separate the computation on the hyper-network and LM, enabling arbitrary batch size on both neural networks. Our method is evaluated by editing up to thousands of facts on LMs with different architectures, i.e., BERT-base, GPT-2, T5-XL (2.8B), and GPT-J (6B), across various knowledge-intensive NLP tasks, i.e., closed book fact-checking and question answering. Remarkably, MALMEN is capable of editing hundreds of times more facts than strong baselines with the identical hyper-network architecture and outperforms editor specifically designed for GPT. Our code is available at https://github.com/ChenmienTan/malmen.

cs.CL↗

arXiv2023

Learning Rewards to Optimize Global Performance Metrics in Deep Reinforcement Learning

When applying reinforcement learning (RL) to a new problem, reward engineering is a necessary, but often difficult and error-prone task a system designer has to face. To avoid this step, we propose LR4GPM, a novel (deep) RL method that can optimize a global performance metric, which is supposed to be available as part of the problem description. LR4GPM alternates between two phases: (1) learning a (possibly vector) reward function used to fit the performance metric, and (2) training a policy to optimize an approximation of this performance metric based on the learned rewards. Such RL training is not straightforward since both the reward function and the policy are trained using non-stationary data. To overcome this issue, we propose several training tricks. We demonstrate the efficiency of LR4GPM on several domains. Notably, LR4GPM outperforms the winner of a recent autonomous driving competition organized at DAI'2020.

cs.LG↗