Spontaneous Reward Hacking in Iterative Self-Refinement

Language models are capable of iteratively improving their outputs based on natural language feedback, thus enabling in-context optimization of user preference. In place of human users, a second language model can be used as an evaluator, providing feedback along with numerical ratings which the generator attempts to optimize. However, because the evaluator is an imperfect proxy of user preference, this optimization can lead to reward hacking, where the evaluator's ratings improve while the generation quality remains stagnant or even decreases as judged by actual user preference. The concern of reward hacking is heightened in iterative self-refinement where the generator and the evaluator use the same underlying language model, in which case the optimization pressure can drive them to exploit shared vulnerabilities. Using an essay editing task, we show that iterative self-refinement leads to deviation between the language model evaluator and human judgment, demonstrating that reward hacking can occur spontaneously in-context with the use of iterative self-refinement. In addition, we study conditions under which reward hacking occurs and observe two factors that affect reward hacking severity: model size and context sharing between the generator and the evaluator.

LLMs Are Prone to Fallacies in Causal Inference

Recent work shows that causal facts can be effectively extracted from LLMs through prompting, facilitating the creation of causal graphs for causal inference tasks. However, it is unclear if this success is limited to explicitly-mentioned causal facts in the pretraining data which the model can memorize. Thus, this work investigates: Can LLMs infer causal relations from other relational data in text? To disentangle the role of memorized causal facts vs inferred causal relations, we finetune LLMs on synthetic data containing temporal, spatial and counterfactual relations, and measure whether the LLM can then infer causal relations. We find that: (a) LLMs are susceptible to inferring causal relations from the order of two entity mentions in text (e.g. X mentioned before Y implies X causes Y); (b) if the order is randomized, LLMs still suffer from the post hoc fallacy, i.e. X occurs before Y (temporal relation) implies X causes Y. We also find that while LLMs can correctly deduce the absence of causal relations from temporal and spatial relations, they have difficulty inferring causal relations from counterfactuals, questioning their understanding of causality.

Iterative Reasoning Preference Optimization

Iterative preference optimization methods have recently been shown to perform well for general instruction tuning tasks, but typically make little improvement on reasoning tasks (Yuan et al., 2024, Chen et al., 2024). In this work we develop an iterative approach that optimizes the preference between competing generated Chain-of-Thought (CoT) candidates by optimizing for winning vs. losing reasoning steps that lead to the correct answer. We train using a modified DPO loss (Rafailov et al., 2023) with an additional negative log-likelihood term, which we find to be crucial. We show reasoning improves across repeated iterations of this scheme. While only relying on examples in the training set, our approach results in increasing accuracy for Llama-2-70B-Chat from 55.6% to 81.6% on GSM8K (and 88.7% with majority voting out of 32 samples), from 12.5% to 20.8% on MATH, and from 77.8% to 86.7% on ARC-Challenge, which outperforms other Llama-2-based models not relying on additionally sourced datasets.

The PRISM Alignment Project: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models

Human feedback plays a central role in the alignment of Large Language Models (LLMs). However, open questions remain about the methods (how), domains (where), people (who) and objectives (to what end) of human feedback collection. To navigate these questions, we introduce PRISM, a new dataset which maps the sociodemographics and stated preferences of 1,500 diverse participants from 75 countries, to their contextual preferences and fine-grained feedback in 8,011 live conversations with 21 LLMs. PRISM contributes (i) wide geographic and demographic participation in human feedback data; (ii) two census-representative samples for understanding collective welfare (UK and US); and (iii) individualised feedback where every rating is linked to a detailed participant profile, thus permitting exploration of personalisation and attribution of sample artefacts. We focus on collecting conversations that centre subjective and multicultural perspectives on value-laden and controversial topics, where we expect the most interpersonal and cross-cultural disagreement. We demonstrate the usefulness of PRISM via three case studies of dialogue diversity, preference diversity, and welfare outcomes, showing that it matters which humans set alignment norms. As well as offering a rich community resource, we advocate for broader participation in AI development and a more inclusive approach to technology design.

Foundational Challenges in Assuring Alignment and Safety of Large Language Models

This work identifies 18 foundational challenges in assuring the alignment and safety of large language models (LLMs). These challenges are organized into three different categories: scientific understanding of LLMs, development and deployment methods, and sociotechnical challenges. Based on the identified challenges, we pose $200+$ concrete research questions.

Your Co-Workers Matter: Evaluating Collaborative Capabilities of Language Models in Blocks World

Language agents that interact with the world on their own have great potential for automating digital tasks. While large language model (LLM) agents have made progress in understanding and executing tasks such as textual games and webpage control, many real-world tasks also require collaboration with humans or other LLMs in equal roles, which involves intent understanding, task coordination, and communication. To test LLM's ability to collaborate, we design a blocks-world environment, where two agents, each having unique goals and skills, build a target structure together. To complete the goals, they can act in the world and communicate in natural language. Under this environment, we design increasingly challenging settings to evaluate different collaboration perspectives, from independent to more complex, dependent tasks. We further adopt chain-of-thought prompts that include intermediate reasoning steps to model the partner's state and identify and correct execution errors. Both human-machine and machine-machine experiments show that LLM agents have strong grounding capacities, and our approach significantly improves the evaluation metric.

Parallel Structures in Pre-training Data Yield In-Context Learning

Pre-trained language models (LMs) are capable of in-context learning (ICL): they can adapt to a task with only a few examples given in the prompt without any parameter update. However, it is unclear where this capability comes from as there is a stark distribution shift between pre-training text and ICL prompts. In this work, we study what patterns of the pre-training data contribute to ICL. We find that LMs' ICL ability depends on $\textit{parallel structures}$ in the pre-training data -- pairs of phrases following similar templates in the same context window. Specifically, we detect parallel structures by checking whether training on one phrase improves prediction of the other, and conduct ablation experiments to study their effect on ICL. We show that removing parallel structures in the pre-training data reduces LMs' ICL accuracy by 51% (vs 2% from random ablation). This drop persists even when excluding common patterns such as n-gram repetitions and long-range dependency, showing the diversity and generality of parallel structures. A closer look at the detected parallel structures indicates that they cover diverse linguistic tasks and span long distances in the data.

Towards Consistent Natural-Language Explanations via Explanation-Consistency Finetuning

Large language models (LLMs) often generate convincing, fluent explanations. However, different from humans, they often generate inconsistent explanations on different inputs. For example, an LLM may generate the explanation "all birds can fly" when answering the question "Can sparrows fly?" but meanwhile answer "no" to the related question "Can penguins fly?". Explanations should be consistent across related examples so that they allow a human to simulate the LLM's decision process on multiple examples. We propose explanation-consistency finetuning (EC-finetuning), a method that adapts LLMs to generate more consistent natural-language explanations on related examples. EC-finetuning involves finetuning LLMs on synthetic data that is carefully constructed to contain consistent explanations. Across a variety of question-answering datasets in various domains, EC-finetuning yields a 10.0% relative explanation consistency improvement on four finetuning datasets, and generalizes to seven out-of-distribution datasets not seen during finetuning (+4.5% relative). Code is available at https://github.com/yandachen/explanation-consistency-finetuning .

Pragmatic Radiology Report Generation

When pneumonia is not found on a chest X-ray, should the report describe this negative observation or omit it? We argue that this question cannot be answered from the X-ray alone and requires a pragmatic perspective, which captures the communicative goal that radiology reports serve between radiologists and patients. However, the standard image-to-text formulation for radiology report generation fails to incorporate such pragmatic intents. Following this pragmatic perspective, we demonstrate that the indication, which describes why a patient comes for an X-ray, drives the mentions of negative observations and introduce indications as additional input to report generation. With respect to the output, we develop a framework to identify uninferable information from the image as a source of model hallucinations, and limit them by cleaning groundtruth reports. Finally, we use indications and cleaned groundtruth reports to develop pragmatic models, and show that they outperform existing methods not only in new pragmatics-inspired metrics (+4.3 Negative F1) but also in standard metrics (+6.3 Positive F1 and +11.0 BLEU-2).

Show Your Work with Confidence: Confidence Bands for Tuning Curves

The choice of hyperparameters greatly impacts performance in natural language processing. Often, it is hard to tell if a method is better than another or just better tuned. Tuning curves fix this ambiguity by accounting for tuning effort. Specifically, they plot validation performance as a function of the number of hyperparameter choices tried so far. While several estimators exist for these curves, it is common to use point estimates, which we show fail silently and give contradictory results when given too little data. Beyond point estimates, confidence bands are necessary to rigorously establish the relationship between different approaches. We present the first method to construct valid confidence bands for tuning curves. The bands are exact, simultaneous, and distribution-free, thus they provide a robust basis for comparing methods. Empirical analysis shows that while bootstrap confidence bands, which serve as a baseline, fail to approximate their target confidence, ours achieve it exactly. We validate our design with ablations, analyze the effect of sample size, and provide guidance on comparing models with our method. To promote confident comparisons in future work, we release a library implementing the method at https://github.com/nalourie/opda .