Consistency Training Along the Transformer Stack

Consistency training encourages models to behave similarly across different contexts, and has shown promise for reducing misalignment. We broaden the scope of consistency training in two ways. First, we introduce two new internal consistency targets: MLP Consistency Training (MLPCT), which matches post-activation MLP states, and Attention Consistency Training (AttCT), which matches per-head attention distributions. Second, we apply consistency training to four additional safety threats: persona in-context learning attacks, adversarial frustration, prefill attacks, and conditional misalignment. Across several models and threat settings, we find that consistency training reduces misalignment well beyond the sycophancy and jailbreak settings studied in prior work. We also find cases of cross-threat generalization, where training against one failure mode improves robustness to another, and identify a shared residual-stream mechanism underlying ACT, MLPCT, and AttCT, while distinguishing BCT as mechanistically distinct. Our results suggest that consistency training is a flexible and extensible framework for alignment, capable of unifying defenses against a broader class of model pathologies.

Consistency Training Can Entrench Misalignment

Consistency training encourages a model to produce similar outputs across related inputs or sampling procedures. Such methods are simple, scalable, and largely label-free, but their effects on model alignment remain poorly understood. Could the self-bootstrapping nature of these methods amplify undesired behavior in models? We test seven consistency training methods on 108 model organisms: open-source models (7B--70B) fine-tuned to exhibit various forms of controlled misaligned behavior. We find that outcomes vary significantly: consistency training generally suppresses reward hacking and emergent misalignment but amplifies sycophancy. We present evidence that distribution shifts induced by the consistency labeling process, rather than variation in the selection operators, may be the primary driver of systematic alignment effects. Finally, we present a unifying theoretical framework to derive conditions under which consistency training will amplify or suppress misalignment. In total, our study establishes that consistency training is not alignment-neutral, and that its use in critical systems should be carefully audited.

Consistency Training while Mitigating Obfuscation via Rate Matching

Large language models are often influenced by extraneous input features, such as cues revealing a user's preferred answer. Consistency training reduces this influence by training models to behave similarly across inputs with and without the extraneous feature. However, existing methods train for consistency over entire responses or internal activations, which also constrains whether the model verbalises said extraneous features. We show this leads to obfuscation, where the model learns not to mention a cue while remaining influenced by it, which may undermine monitorability. To address this, we introduce Rate Matching Consistency Training (RMCT), which trains for consistency over selected behavioural properties without constraining how this behaviour is expressed. RMCT matches the rate at which the model exhibits a target behaviour (e.g., following a bias cue) across input perturbations, rather than requiring paired inputs with and without the extraneous feature, extending consistency training to settings where the extraneous features cannot be removed. We evaluate RMCT on sycophancy reduction in two open-weight language models, achieving reductions in bias-following comparable to a standard consistency-training baseline on held-out bias types, while largely preserving the model's tendency to verbalise the bias cue. Further, we find that RMCT is more data-efficient at the expense of being less compute-efficient in our experiments. Overall, RMCT shows that consistency training can improve behavioural robustness without directly trading off against monitorability.

Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time

Language model finetuning often results in learning undesirable traits in combination with desired ones. To address this, we propose inoculation prompting: modifying finetuning data by prepending a short system-prompt instruction that deliberately elicits the undesirable trait. At test time, we evaluate without the instruction; inoculated models have much lower expression of the trait than models trained with unmodified training data. Inoculation is selective: in a toy setting where assistant responses are always in Spanish and ALL-CAPS, an appropriate inoculation (e.g., ``You always speak in Spanish.'') teaches the model to capitalize responses while still responding in English. We find that inoculation is also effective across several additional settings: reducing emergent misalignment (EM) from task-specific finetuning, defending against backdoor injections, and mitigating the transmission of traits via subliminal learning. Follow-up analysis suggests a mechanism: making a trait less surprising via inoculation reduces optimization pressure to globally update the model, thereby reducing the degree of generalization. Our analysis relates to prior work on EM: inoculation explains prior findings that educational contexts mitigate EM from insecure code. Beyond demonstrating a simple and effective technique for selective learning, our results contribute to a better conceptual understanding of how and why language models generalize.

Investigating ReLoRA: Effects on the Learning Dynamics of Small Language Models

Parameter-efficient methods such as LoRA have revolutionised the fine-tuning of LLMs. Still, their extension to pretraining via ReLoRA is less well understood, especially for small language models (SLMs), which offer lower computational and environmental costs. This work is the first systematic study of ReLoRA in SLMs (11M-66M parameters), evaluating both performance and learning dynamics. Through ablation experiments, we find that ReLoRA generally performs worse than standard training on loss, Paloma perplexity and BLiMP, with the gap widening for the larger models. Further analysis of the learning dynamics of the models indicates that ReLoRA reinforces the rank deficiencies found in smaller models. These results indicate that low-rank update strategies may not transfer easily to SLM pretraining, highlighting the need for more research in the low-compute regime.

Batayan: A Filipino NLP benchmark for evaluating Large Language Models

Recent advances in large language models (LLMs) have demonstrated remarkable capabilities on widely benchmarked high-resource languages; however, linguistic nuances of under-resourced languages remain unexplored. We introduce Batayan, a holistic Filipino benchmark designed to systematically evaluate LLMs across three key natural language processing (NLP) competencies: understanding, reasoning, and generation. Batayan consolidates eight tasks, covering both Tagalog and code-switched Taglish utterances. Our rigorous, native-speaker-driven annotation process ensures fluency and authenticity to the complex morphological and syntactic structures of Filipino, alleviating a pervasive translationese bias in existing Filipino corpora. We report empirical results on a variety of multilingual LLMs, highlighting significant performance gaps that signal the under-representation of Filipino in pretraining corpora, the unique hurdles in modeling Filipino's rich morphology and construction, and the importance of explicit Filipino language support and instruction tuning. Moreover, we discuss the practical challenges encountered in dataset construction and propose principled solutions for building culturally and linguistically-faithful resources in under-represented languages. We also provide a public benchmark and leaderboard as a clear foundation for iterative, community-driven progress in Filipino NLP.