Rishabh Joshi

Polina Zablotskaia, Misha Khalman, Rishabh Joshi et al. · deepmind

Despite the recent advances in abstractive text summarization, current summarization models still suffer from generating factually inconsistent summaries, reducing their utility for real-world application. We argue that the main reason for such behavior is that the summarization models trained with maximum likelihood objective assign high probability to plausible sequences given the context, but they often do not accurately rank sequences by their consistency. In this work, we solve this problem by calibrating the likelihood of model generated sequences to better align with a consistency metric measured by natural language inference (NLI) models. The human evaluation study and automatic metrics show that the calibrated models generate more consistent and higher-quality summaries. We also show that the models trained using our method return probabilities that are better aligned with the NLI scores, which significantly increase reliability of summarization models.

Rishabh Joshi, Vidhisha Balachandran, Emily Saldanha et al. · cmu

Keyphrase extraction aims at automatically extracting a list of "important" phrases representing the key concepts in a document. Prior approaches for unsupervised keyphrase extraction resorted to heuristic notions of phrase importance via embedding clustering or graph centrality, requiring extensive domain expertise. Our work presents a simple alternative approach which defines keyphrases as document phrases that are salient for predicting the topic of the document. To this end, we propose INSPECT -- an approach that uses self-explaining models for identifying influential keyphrases in a document by measuring the predictive impact of input phrases on the downstream task of the document topic classification. We show that this novel method not only alleviates the need for ad-hoc heuristics but also achieves state-of-the-art results in unsupervised keyphrase extraction in four datasets across two domains: scientific publications and news articles.

Wei Xiong, Chengshuai Shi, Jiaming Shen et al.

Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and employing multi-turn Chain-of-Thought (CoT) reasoning. While current methods focus on synthetic data generation and Supervised Fine-Tuning (SFT), this paper studies the complementary direct preference learning approach to further improve model performance. However, existing direct preference learning algorithms are originally designed for the single-turn chat task, and do not fully address the complexities of multi-turn reasoning and external tool integration required for tool-integrated mathematical reasoning tasks. To fill in this gap, we introduce a multi-turn direct preference learning framework, tailored for this context, that leverages feedback from code interpreters and optimizes trajectory-level preferences. This framework includes multi-turn DPO and multi-turn KTO as specific implementations. The effectiveness of our framework is validated through training of various language models using an augmented prompt set from the GSM8K and MATH datasets. Our results demonstrate substantial improvements: a supervised fine-tuned Gemma-1.1-it-7B model's performance increased from 77.5% to 83.9% on GSM8K and from 46.1% to 51.2% on MATH. Similarly, a Gemma-2-it-9B model improved from 84.1% to 86.3% on GSM8K and from 51.0% to 54.5% on MATH.

Tianqi Liu, Wei Xiong, Jie Ren et al.

Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. However, traditional RM training, which relies on response pairs tied to specific prompts, struggles to disentangle prompt-driven preferences from prompt-independent artifacts, such as response length and format. In this work, we expose a fundamental limitation of current RM training methods, where RMs fail to effectively distinguish between contextual signals and irrelevant artifacts when determining preferences. To address this, we introduce a causal framework that learns preferences independent of these artifacts and propose a novel data augmentation technique designed to eliminate them. Extensive experiments show that our approach successfully filters out undesirable artifacts, yielding a more robust reward model (RRM). Our RRM improves the performance of a pairwise reward model trained on Gemma-2-9b-it, on RewardBench, increasing accuracy from 80.61% to 84.15%. Additionally, we train two DPO policies using both the RM and RRM, demonstrating that the RRM significantly enhances DPO-aligned policies, improving MT-Bench scores from 7.27 to 8.31 and length-controlled win-rates in AlpacaEval-2 from 33.46% to 52.49%.

Tianqi Liu, Yao Zhao, Rishabh Joshi et al.

Improving the alignment of language models with human preferences remains an active research challenge. Previous approaches have primarily utilized Reinforcement Learning from Human Feedback (RLHF) via online RL methods such as Proximal Policy Optimization (PPO). Recently, offline methods such as Sequence Likelihood Calibration (SLiC) and Direct Preference Optimization (DPO) have emerged as attractive alternatives, offering improvements in stability and scalability while maintaining competitive performance. SLiC refines its loss function using sequence pairs sampled from a supervised fine-tuned (SFT) policy, while DPO directly optimizes language models based on preference data, foregoing the need for a separate reward model. However, the maximum likelihood estimator (MLE) of the target optimal policy requires labeled preference pairs sampled from that policy. DPO's lack of a reward model constrains its ability to sample preference pairs from the optimal policy, and SLiC is restricted to sampling preference pairs only from the SFT policy. To address these limitations, we introduce a novel approach called Statistical Rejection Sampling Optimization (RSO) that aims to source preference data from the target optimal policy using rejection sampling, enabling a more accurate estimation of the optimal policy. We also propose a unified framework that enhances the loss functions used in both SLiC and DPO from a preference modeling standpoint. Through extensive experiments across three diverse tasks, we demonstrate that RSO consistently outperforms both SLiC and DPO on evaluations from both Large Language Model (LLM) and human raters.

Yao Zhao, Misha Khalman, Rishabh Joshi et al.

Conditional language models are predominantly trained with maximum likelihood estimation (MLE), giving probability mass to sparsely observed target sequences. While MLE trained models assign high probability to plausible sequences given the context, the model probabilities often do not accurately rank-order generated sequences by quality. This has been empirically observed in beam search decoding as output quality degrading with large beam sizes, and decoding strategies benefiting from heuristics such as length normalization and repetition-blocking. In this work, we introduce sequence likelihood calibration (SLiC) where the likelihood of model generated sequences are calibrated to better align with reference sequences in the model's latent space. With SLiC, decoding heuristics become unnecessary and decoding candidates' quality significantly improves regardless of the decoding method. Furthermore, SLiC shows no sign of diminishing returns with model scale, and presents alternative ways to improve quality with limited training and inference budgets. With SLiC, we exceed or match SOTA results on a wide range of generation tasks spanning abstractive summarization, question generation, abstractive question answering and data-to-text generation, even with modest-sized models.

Gemini Team, Petko Georgiev, Ving Ian Lei et al. · deepmind, mila

In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.

Gheorghe Comanici, Eric Bieber, Mike Schaekermann et al. · amazon-science, baidu

In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving.

Ritam Dutt, Sayan Sinha, Rishabh Joshi et al.

Modelling persuasion strategies as predictors of task outcome has several real-world applications and has received considerable attention from the computational linguistics community. However, previous research has failed to account for the resisting strategies employed by an individual to foil such persuasion attempts. Grounded in prior literature in cognitive and social psychology, we propose a generalised framework for identifying resisting strategies in persuasive conversations. We instantiate our framework on two distinct datasets comprising persuasion and negotiation conversations. We also leverage a hierarchical sequence-labelling neural architecture to infer the aforementioned resisting strategies automatically. Our experiments reveal the asymmetry of power roles in non-collaborative goal-directed conversations and the benefits accrued from incorporating resisting strategies on the final conversation outcome. We also investigate the role of different resisting strategies on the conversation outcome and glean insights that corroborate with past findings. We also make the code and the dataset of this work publicly available at https://github.com/americast/resper.

Sopan Khosla, Rishabh Joshi, Ritam Dutt et al.

In this paper we describe our submission for the task of Propaganda Span Identification in news articles. We introduce a BERT-BiLSTM based span-level propaganda classification model that identifies which token spans within the sentence are indicative of propaganda. The "multi-granular" model incorporates linguistic knowledge at various levels of text granularity, including word, sentence and document level syntactic, semantic and pragmatic affect features, which significantly improve model performance, compared to its language-agnostic variant. To facilitate better representation learning, we also collect a corpus of 10k news articles, and use it for fine-tuning the model. The final model is a majority-voting ensemble which learns different propaganda class boundaries by leveraging different subsets of incorporated knowledge and attains $4^{th}$ position on the test leaderboard. Our final model and code is released at https://github.com/sopu/PropagandaSemEval2020.

Shikhar Vashishth, Rishabh Joshi, Sai Suman Prayaga et al.

Distantly-supervised Relation Extraction (RE) methods train an extractor by automatically aligning relation instances in a Knowledge Base (KB) with unstructured text. In addition to relation instances, KBs often contain other relevant side information, such as aliases of relations (e.g., founded and co-founded are aliases for the relation founderOfCompany). RE models usually ignore such readily available side information. In this paper, we propose RESIDE, a distantly-supervised neural relation extraction method which utilizes additional side information from KBs for improved relation extraction. It uses entity type and relation alias information for imposing soft constraints while predicting relations. RESIDE employs Graph Convolution Networks (GCN) to encode syntactic information from text and improves performance even when limited side information is available. Through extensive experiments on benchmark datasets, we demonstrate RESIDE's effectiveness. We have made RESIDE's source code available to encourage reproducible research.

Tianqi Liu, Zhen Qin, Junru Wu et al.

Aligning language models (LMs) with curated human feedback is critical to control their behaviors in real-world applications. Several recent policy optimization methods, such as DPO and SLiC, serve as promising alternatives to the traditional Reinforcement Learning from Human Feedback (RLHF) approach. In practice, human feedback often comes in a format of a ranked list over multiple responses to amortize the cost of reading prompt. Multiple responses can also be ranked by reward models or AI feedback. There lacks such a thorough study on directly fitting upon a list of responses. In this work, we formulate the LM alignment as a \textit{listwise} ranking problem and describe the LiPO framework, where the policy can potentially learn more effectively from a ranked list of plausible responses given the prompt. This view draws an explicit connection to Learning-to-Rank (LTR), where most existing preference optimization work can be mapped to existing ranking objectives. Following this connection, we provide an examination of ranking objectives that are not well studied for LM alignment with DPO and SLiC as special cases when list size is two. In particular, we highlight a specific method, LiPO-$λ$, which leverages a state-of-the-art \textit{listwise} ranking objective and weights each preference pair in a more advanced manner. We show that LiPO-$λ$ can outperform DPO variants and SLiC by a clear margin on several preference alignment tasks with both curated and real rankwise preference data.

Daniele Calandriello, Daniel Guo, Remi Munos et al.

Ensuring alignment of language models' outputs with human preferences is critical to guarantee a useful, safe, and pleasant user experience. Thus, human alignment has been extensively studied recently and several methods such as Reinforcement Learning from Human Feedback (RLHF), Direct Policy Optimisation (DPO) and Sequence Likelihood Calibration (SLiC) have emerged. In this paper, our contribution is two-fold. First, we show the equivalence between two recent alignment methods, namely Identity Policy Optimisation (IPO) and Nash Mirror Descent (Nash-MD). Second, we introduce a generalisation of IPO, named IPO-MD, that leverages the regularised sampling approach proposed by Nash-MD. This equivalence may seem surprising at first sight, since IPO is an offline method whereas Nash-MD is an online method using a preference model. However, this equivalence can be proven when we consider the online version of IPO, that is when both generations are sampled by the online policy and annotated by a trained preference model. Optimising the IPO loss with such a stream of data becomes then equivalent to finding the Nash equilibrium of the preference model through self-play. Building on this equivalence, we introduce the IPO-MD algorithm that generates data with a mixture policy (between the online and reference policy) similarly as the general Nash-MD algorithm. We compare online-IPO and IPO-MD to different online versions of existing losses on preference data such as DPO and SLiC on a summarisation task.

Ziyu Ye, Rishabh Agarwal, Tianqi Liu et al.

Current reinforcement learning (RL) frameworks for large language models (LLM) post-training typically assume a fixed prompt distribution, which is sub-optimal and bottlenecks scalability. Prior works have explored prompt evolving, but are often limited to the supervised fine-tuning stage, and prompts are sampled and evolved uniformly without signals. This empirical work presents a paradigm shift: Evolving Alignment via Asymmetric Self-Play (eva), that casts post-training as an infinite game with regret-based signals for 2 players: (i) a creator, who strategically samples and creates new informative prompts and (ii) a solver, who learns to produce preferred responses. eva is the first method that allows language models to adaptively create training prompts in both offline and online RL post-training. The design is simple, easy-to-use yet remarkably effective: eva sets a new SOTA on challenging benchmarks, without any extra human prompts, e.g. it boosts the win-rate of gemma-2-9b-it on Arena-Hard by 51.6% -> 60.1% for DPO and 52.6% -> 62.4% for RLOO, surpassing claude-3-opus and catching up to gemini-1.5-pro, both of which are orders of magnitude larger. Extensive experiments show eva can create effective RL curricula and is robust across ablations. We believe adaptively evolving prompts are key to designing the next-generation RL post-training scheme.

Gemini Team, Rohan Anil, Sebastian Borgeaud et al.

This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.

Yao Zhao, Rishabh Joshi, Tianqi Liu et al.

Learning from human feedback has been shown to be effective at aligning language models with human preferences. Past work has often relied on Reinforcement Learning from Human Feedback (RLHF), which optimizes the language model using reward scores assigned from a reward model trained on human preference data. In this work we show how the recently introduced Sequence Likelihood Calibration (SLiC), can also be used to effectively learn from human preferences (SLiC-HF). Furthermore, we demonstrate this can be done with human feedback data collected for a different model, similar to off-policy, offline RL data. Automatic and human evaluation experiments on the TL;DR summarization task show that SLiC-HF significantly improves supervised fine-tuning baselines. Furthermore, SLiC-HF presents a competitive alternative to the PPO RLHF implementation used in past work while being much simpler to implement, easier to tune and more computationally efficient in practice.

Rishabh Joshi, Vidhisha Balachandran, Shikhar Vashishth et al.

To successfully negotiate a deal, it is not enough to communicate fluently: pragmatic planning of persuasive negotiation strategies is essential. While modern dialogue agents excel at generating fluent sentences, they still lack pragmatic grounding and cannot reason strategically. We present DialoGraph, a negotiation system that incorporates pragmatic strategies in a negotiation dialogue using graph neural networks. DialoGraph explicitly incorporates dependencies between sequences of strategies to enable improved and interpretable prediction of next optimal strategies, given the dialogue context. Our graph-based method outperforms prior state-of-the-art negotiation models both in the accuracy of strategy/dialogue act prediction and in the quality of downstream dialogue response generation. We qualitatively show further benefits of learned strategy-graphs in providing explicit associations between effective negotiation strategies over the course of the dialogue, leading to interpretable and strategic dialogues.

Ritam Dutt, Rishabh Joshi, Carolyn Penstein Rose

The notion of face refers to the public self-image of an individual that emerges both from the individual's own actions as well as from the interaction with others. Modeling face and understanding its state changes throughout a conversation is critical to the study of maintenance of basic human needs in and through interaction. Grounded in the politeness theory of Brown and Levinson (1978), we propose a generalized framework for modeling face acts in persuasion conversations, resulting in a reliable coding manual, an annotated corpus, and computational models. The framework reveals insights about differences in face act utilization between asymmetric roles in persuasion conversations. Using computational models, we are able to successfully identify face acts as well as predict a key conversational outcome (e.g. donation success). Finally, we model a latent representation of the conversational state to analyze the impact of predicted face acts on the probability of a positive conversational outcome and observe several correlations that corroborate previous findings.

Shikhar Vashishth, Denis Newman-Griffis, Rishabh Joshi et al.

Medical entity linking is the task of identifying and standardizing medical concepts referred to in an unstructured text. Most of the existing methods adopt a three-step approach of (1) detecting mentions, (2) generating a list of candidate concepts, and finally (3) picking the best concept among them. In this paper, we probe into alleviating the problem of overgeneration of candidate concepts in the candidate generation module, the most under-studied component of medical entity linking. For this, we present MedType, a fully modular system that prunes out irrelevant candidate concepts based on the predicted semantic type of an entity mention. We incorporate MedType into five off-the-shelf toolkits for medical entity linking and demonstrate that it consistently improves entity linking performance across several benchmark datasets. To address the dearth of annotated training data for medical entity linking, we present WikiMed and PubMedDS, two large-scale medical entity linking datasets, and demonstrate that pre-training MedType on these datasets further improves entity linking performance. We make our source code and datasets publicly available for medical entity linking research.

Rakesh Bal, Sayan Sinha, Swastika Dutta et al.

Twitter has become one of the most sought after places to discuss a wide variety of topics, including medically relevant issues such as cancer. This helps spread awareness regarding the various causes, cures and prevention methods of cancer. However, no proper analysis has been performed, which discusses the validity of such claims. In this work, we aim to tackle the misinformation spread in such platforms. We collect and present a dataset regarding tweets which talk specifically about cancer and propose an attention-based deep learning model for automated detection of misinformation along with its spread. We then do a comparative analysis of the linguistic variation in the text corresponding to misinformation and truth. This analysis helps us gather relevant insights on various social aspects related to misinformed tweets.

Gaurav Kumar, Rishabh Joshi, Jaspreet Singh et al.

The problem of building a coherent and non-monotonous conversational agent with proper discourse and coverage is still an area of open research. Current architectures only take care of semantic and contextual information for a given query and fail to completely account for syntactic and external knowledge which are crucial for generating responses in a chit-chat system. To overcome this problem, we propose an end to end multi-stream deep learning architecture which learns unified embeddings for query-response pairs by leveraging contextual information from memory networks and syntactic information by incorporating Graph Convolution Networks (GCN) over their dependency parse. A stream of this network also utilizes transfer learning by pre-training a bidirectional transformer to extract semantic representation for each input sentence and incorporates external knowledge through the the neighborhood of the entities from a Knowledge Base (KB). We benchmark these embeddings on next sentence prediction task and significantly improve upon the existing techniques. Furthermore, we use AMUSED to represent query and responses along with its context to develop a retrieval based conversational agent which has been validated by expert linguists to have comprehensive engagement with humans.

Dhruv Gosain, Rishabh Joshi

Frangi Filters are one of the widely used filters for enhancing vessels in medical images. Since they were first proposed, the threshold of the vesselness function of Frangi Filters is to be arranged for each individual application. These thresholds are changed manually for individual fluoroscope, for enhancing coronary angiogram images. Hence it is felt, there is a need of mitigating the tuning procedure of threshold values for every fluoroscope. The current papers approach has been devised in order to treat the coronary angiogram images uniformly, irrespective of the fluoroscopes through which they were obtained and the patient demographics for further stenosis detection. This problem to the best of our knowledge has not been addressed yet. In the approach, before feeding the image to Frangi Filters, non uniform illumination of the input image is removed using homomorphic filters and the image is enhanced using Non Subsampled Contourlet Transform (NSCT). The experiment was conducted on the data that has been accumulated from various hospitals in India and the results obtained verifies dependency removal of parameters without compromising the results obtained by Frangi filters.