Archie Chaudhury

Archie Chaudhury

Catastrophic forgetting is often framed as a representational problem: after sequential training, a model appears to lose the features that supported performance on earlier tasks. We challenge the stronger form of this view. Across controlled continual-learning settings, we find that a significant portion of apparent forgetting can be attributed to interface drift between internal stages rather than permanent erasure of task-relevant computation. We study this phenomenon through a stitched evaluation protocol that combines early computation from a post-update network with late computation from its predecessor, optionally mediated by a compact, task-specific transport key. We describe transport keys at a systems level as compact interface-alignment operators estimated from a small set of paired anchor activations and evaluated through model stitching. On split CIFAR-100 with a ResNet-style network, transport keys recover most of the original Task A performance after sequential training on Task B. On a compact vision transformer, we observe a similar recovery pattern. These results suggest that continual learning may require better mechanisms for indexing and re-accessing latent computations, not only methods that prevent weight change.

Archie Chaudhury

Reinforcement Learning frameworks, particularly those utilizing human annotations, have become an increasingly popular method for preference fine-tuning, where the outputs of a language model are tuned to match a certain set of behavioral policies or guidelines. Reinforcement Learning through Human Feedback (RLHF) is perhaps the most popular implementation of such a framework, particularly for aligning LMs toward safety and human intent. However, the internal workings of how such alignment is achieved remain largely opaque. In this work, we systematically analyze preference optimization for language model alignment by applying layer-wide causal patching between a base model and its tuned counterpart across human preference pairs. We implement our methodology on \textit{Llama-3.2-1B}, and find that alignment is spatially localized: mid-layer activations encode a distinct subspace that causally determines reward-consistent behavior, while early and late layers remain largely unaffected. Utilizing LASSO regression, we also find that only a small number of layers possess non-zero coefficients linking activation distances to reward gains. Overall, we show that, at least for some language models, alignment from human-based, preferential tuning is a directional, low rank process, rather than diffuse and parameteric.

Leon Eshuijs, Archie Chaudhury, Alan McBeth et al.

Detecting subtle forms of dishonesty like sycophancy and manipulation in Large Language Models (LLMs) remains challenging for both humans and automated evaluators, as these behaviors often appear through small biases rather than clear false statements. We introduce Judge Using Safety-Steered Alternatives (JUSSA), a novel framework that employs steering vectors not to improve model behavior directly, but to enhance LLM judges' evaluation capabilities. JUSSA applies steering vectors during inference to generate more honest alternatives, providing judges with contrastive examples that make subtle dishonest patterns easier to detect. While existing evaluation methods rely on black-box evaluation, JUSSA leverages model internals to create targeted comparisons from single examples. We evaluate our method on sycophancy detection and introduce a new manipulation dataset covering multiple types of manipulation. Our results demonstrate that JUSSA effectively improves detection accuracy over single-response evaluation in various cases. Analysis across judge models reveals that JUSSA helps weaker judges on easier dishonesty detection tasks, and stronger judges on harder tasks. Layer-wise experiments show how dishonest prompts cause representations to diverge from honest ones in middle layers, revealing where steering interventions are most effective for generating contrastive examples. By demonstrating that steering vectors can enhance safety evaluation rather than just modify behavior, our work opens new directions for scalable model auditing as systems become increasingly sophisticated.

Vishakha Agrawal, Archie Chaudhury, Shreya Agrawal

While Large Language Models (LLMs) are fundamentally next-token prediction systems, their practical applications extend far beyond this basic function. From natural language processing and text generation to conversational assistants and software use, LLMs have numerous use-cases, and have already acquired a significant degree of enterprise adoption. To evaluate such models, static evaluation datasets, consisting of a set of prompts and their corresponding ground truths, are often used to benchmark the efficacy of the model for a particular task. In this paper, we provide the basis for a more comprehensive evaluation framework, based upon a traditional game and tool-based architecture that enables a more overarching measurement of a model's capabilities. For simplicity, we provide a generalized foundation that can be extended, without significant alteration, to numerous scenarios, from specific use cases such as supply chain management or financial reasoning, to abstract measurements such as ethics or safety.