Disha Shrivastava

Disha Shrivastava, Hugo Larochelle, Daniel Tarlow · mila

With the success of large language models (LLMs) of code and their use as code assistants (e.g. Codex used in GitHub Copilot), techniques for introducing domain-specific knowledge in the prompt design process become important. In this work, we propose a framework called Repo-Level Prompt Generator that learns to generate example-specific prompts using prompt proposals. The prompt proposals take context from the entire repository, thereby incorporating both the structure of the repository and the context from other relevant files (e.g. imports, parent class files). Our technique doesn't require any access to the weights of the LLM, making it applicable in cases where we only have black-box access to the LLM. We conduct experiments on the task of single-line code-autocompletion using code repositories taken from Google Code archives. We demonstrate that an oracle constructed from our prompt proposals gives a remarkably high relative improvement of 36% over Codex, showing the quality of these proposals. Further, we show that when we train a model to predict a prompt proposal, we can achieve significant performance gains over Codex and other baselines. We release our code, data, and trained checkpoints at: \url{https://github.com/shrivastavadisha/repo_level_prompt_generation}.

Disha Shrivastava, Denis Kocetkov, Harm de Vries et al. · mila

Despite the huge success of Large Language Models (LLMs) in coding assistants like GitHub Copilot, these models struggle to understand the context present in the repository (e.g., imports, parent classes, files with similar names, etc.), thereby producing inaccurate code completions. This effect is more pronounced when using these assistants for repositories that the model has not seen during training, such as proprietary software or work-in-progress code projects. Recent work has shown the promise of using context from the repository during inference. In this work, we extend this idea and propose RepoFusion, a framework to train models to incorporate relevant repository context. Experiments on single-line code completion show that our models trained with repository context significantly outperform much larger code models as CodeGen-16B-multi ($\sim73\times$ larger) and closely match the performance of the $\sim 70\times$ larger StarCoderBase model that was trained with the Fill-in-the-Middle objective. We find these results to be a novel and compelling demonstration of the gains that training with repository context can bring. We carry out extensive ablation studies to investigate the impact of design choices such as context type, number of contexts, context length, and initialization within our framework. Lastly, we release Stack-Repo, a dataset of 200 Java repositories with permissive licenses and near-deduplicated files that are augmented with three types of repository contexts. Additionally, we are making available the code and trained checkpoints for our work. Our released resources can be found at \url{https://huggingface.co/RepoFusion}.

Aviral Kumar, Vincent Zhuang, Rishabh Agarwal et al. · deepmind

Self-correction is a highly desirable capability of large language models (LLMs), yet it has consistently been found to be largely ineffective in modern LLMs. Current methods for training self-correction typically depend on either multiple models, a more advanced model, or additional forms of supervision. To address these shortcomings, we develop a multi-turn online reinforcement learning (RL) approach, SCoRe, that significantly improves an LLM's self-correction ability using entirely self-generated data. To build SCoRe, we first show that variants of supervised fine-tuning (SFT) on offline model-generated correction traces are often insufficient for instilling self-correction behavior. In particular, we observe that training via SFT falls prey to either a distribution mismatch between mistakes made by the data-collection policy and the model's own responses, or to behavior collapse, where learning implicitly prefers only a certain mode of correction behavior that is often not effective at self-correction on test problems. SCoRe addresses these challenges by training under the model's own distribution of self-generated correction traces and using appropriate regularization to steer the learning process into learning a self-correction behavior that is effective at test time as opposed to fitting high-reward responses for a given prompt. This regularization process includes an initial phase of multi-turn RL on a base model to generate a policy initialization that is less susceptible to collapse, followed by using a reward bonus to amplify self-correction. With Gemini 1.0 Pro and 1.5 Flash models, we find that SCoRe achieves state-of-the-art self-correction performance, improving the base models' self-correction by 15.6% and 9.1% respectively on MATH and HumanEval.

Avinash Bhat, Disha Shrivastava, Jin L. C. Guo · mila

Despite the potential of Large Language Models (LLMs) as writing assistants, they are plagued by issues like coherence and fluency of the model output, trustworthiness, ownership of the generated content, and predictability of model performance, thereby limiting their usability. In this position paper, we propose to adopt Norman's seven stages of action as a framework to approach the interaction design of intelligent writing assistants. We illustrate the framework's applicability to writing tasks by providing an example of software tutorial authoring. The paper also discusses the framework as a tool to synthesize research on the interaction design of LLM-based tools and presents examples of tools that support the stages of action. Finally, we briefly outline the potential of a framework for human-LLM interaction research.

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.

Mina Lee, Katy Ilonka Gero, John Joon Young Chung et al. · allen-ai, deepmind

In our era of rapid technological advancement, the research landscape for writing assistants has become increasingly fragmented across various research communities. We seek to address this challenge by proposing a design space as a structured way to examine and explore the multidimensional space of intelligent and interactive writing assistants. Through a large community collaboration, we explore five aspects of writing assistants: task, user, technology, interaction, and ecosystem. Within each aspect, we define dimensions (i.e., fundamental components of an aspect) and codes (i.e., potential options for each dimension) by systematically reviewing 115 papers. Our design space aims to offer researchers and designers a practical tool to navigate, comprehend, and compare the various possibilities of writing assistants, and aid in the envisioning and design of new writing assistants.

Disha Shrivastava, Hugo Larochelle, Daniel Tarlow

The goal of program synthesis from examples is to find a computer program that is consistent with a given set of input-output examples. Most learning-based approaches try to find a program that satisfies all examples at once. Our work, by contrast, considers an approach that breaks the problem into two stages: (a) find programs that satisfy only one example, and (b) leverage these per-example solutions to yield a program that satisfies all examples. We introduce the Cross Aggregator neural network module based on a multi-head attention mechanism that learns to combine the cues present in these per-example solutions to synthesize a global solution. Evaluation across programs of different lengths and under two different experimental settings reveal that when given the same time budget, our technique significantly improves the success rate over PCCoder [Zohar et. al 2018] and other ablation baselines. The code, data and trained models for our work can be found at https://github.com/shrivastavadisha/N-PEPS.

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.

Edoardo Maria Ponti, Rahul Aralikatte, Disha Shrivastava et al.

Model-agnostic meta-learning (MAML) has been recently put forth as a strategy to learn resource-poor languages in a sample-efficient fashion. Nevertheless, the properties of these languages are often not well represented by those available during training. Hence, we argue that the i.i.d. assumption ingrained in MAML makes it ill-suited for cross-lingual NLP. In fact, under a decision-theoretic framework, MAML can be interpreted as minimising the expected risk across training languages (with a uniform prior), which is known as Bayes criterion. To increase its robustness to outlier languages, we create two variants of MAML based on alternative criteria: Minimax MAML reduces the maximum risk across languages, while Neyman-Pearson MAML constrains the risk in each language to a maximum threshold. Both criteria constitute fully differentiable two-player games. In light of this, we propose a new adaptive optimiser solving for a local approximation to their Nash equilibrium. We evaluate both model variants on two popular NLP tasks, part-of-speech tagging and question answering. We report gains for their average and minimum performance across low-resource languages in zero- and few-shot settings, compared to joint multi-source transfer and vanilla MAML.

Disha Shrivastava, Hugo Larochelle, Daniel Tarlow

The ability to adapt to unseen, local contexts is an important challenge that successful models of source code must overcome. One of the most popular approaches for the adaptation of such models is dynamic evaluation. With dynamic evaluation, when running a model on an unseen file, the model is updated immediately after having observed each token in that file. In this work, we propose instead to frame the problem of context adaptation as a meta-learning problem. We aim to train a base source code model that is best able to learn from information in a file to deliver improved predictions of missing tokens. Unlike dynamic evaluation, this formulation allows us to select more targeted information (support tokens) for adaptation, that is both before and after a target hole in a file. We consider an evaluation setting that we call line-level maintenance, designed to reflect the downstream task of code auto-completion in an IDE. Leveraging recent developments in meta-learning such as first-order MAML and Reptile, we demonstrate improved performance in experiments on a large scale Java GitHub corpus, compared to other adaptation baselines including dynamic evaluation. Moreover, our analysis shows that, compared to a non-adaptive baseline, our approach improves performance on identifiers and literals by 44\% and 15\%, respectively.

Disha Shrivastava, Eeshan Gunesh Dhekane, Riashat Islam

Exploration and adaptation to new tasks in a transfer learning setup is a central challenge in reinforcement learning. In this work, we build on the idea of modeling a distribution over policies in a Bayesian deep reinforcement learning setup to propose a transfer strategy. Recent works have shown to induce diversity in the learned policies by maximizing the entropy of a distribution of policies (Bachman et al., 2018; Garnelo et al., 2018) and thus, we postulate that our proposed approach leads to faster exploration resulting in improved transfer learning. We support our hypothesis by demonstrating favorable experimental results on a variety of settings on fully-observable GridWorld and partially observable MiniGrid (Chevalier-Boisvert et al., 2018) environments.

Disha Shrivastava, Sreyash Kenkre, Santosh Penubothula

We introduce a new measure for unsupervised hypernym detection and directionality. The motivation is to keep the measure computationally light and portatable across languages. We show that the relative physical location of words in explanatory articles captures the directionality property. Further, the phrases in section titles of articles about the word, capture the semantic similarity needed for hypernym detection task. We experimentally show that the combination of features coming from these two simple measures suffices to produce results comparable with the best unsupervised measures in terms of the average precision.

Disha Shrivastava, Abhijit Mishra, Karthik Sankaranarayanan

Coherence of text is an important attribute to be measured for both manually and automatically generated discourse; but well-defined quantitative metrics for it are still elusive. In this paper, we present a metric for scoring topical coherence of an input paragraph on a real-valued scale by analyzing its underlying topical structure. We first extract all possible topics that the sentences of a paragraph of text are related to. Coherence of this text is then measured by computing: (a) the degree of uncertainty of the topics with respect to the paragraph, and (b) the relatedness between these topics. All components of our modular framework rely only on unlabeled data and WordNet, thus making it completely unsupervised, which is an important feature for general-purpose usage of any metric. Experiments are conducted on two datasets - a publicly available dataset for essay grading (representing human discourse), and a synthetic dataset constructed by mixing content from multiple paragraphs covering diverse topics. Our evaluation shows that the measured coherence scores are positively correlated with the ground truth for both the datasets. Further validation to our coherence scores is provided by conducting human evaluation on the synthetic data, showing a significant agreement of 79.3%

Disha Shrivastava, Santanu Chaudhury, Dr. Jayadeva

Training deep networks is expensive and time-consuming with the training period increasing with data size and growth in model parameters. In this paper, we provide a framework for distributed training of deep networks over a cluster of CPUs in Apache Spark. The framework implements both Data Parallelism and Model Parallelism making it suitable to use for deep networks which require huge training data and model parameters which are too big to fit into the memory of a single machine. It can be scaled easily over a cluster of cheap commodity hardware to attain significant speedup and obtain better results making it quite economical as compared to farm of GPUs and supercomputers. We have proposed a new algorithm for training of deep networks for the case when the network is partitioned across the machines (Model Parallelism) along with detailed cost analysis and proof of convergence of the same. We have developed implementations for Fully-Connected Feedforward Networks, Convolutional Neural Networks, Recurrent Neural Networks and Long Short-Term Memory architectures. We present the results of extensive simulations demonstrating the speedup and accuracy obtained by our framework for different sizes of the data and model parameters with variation in the number of worker cores/partitions; thereby showing that our proposed framework can achieve significant speedup (upto 11X for CNN) and is also quite scalable.

Disha Shrivastava, Saneem Ahmed CG, Anirban Laha et al.

Much work has been done in understanding human creativity and defining measures to evaluate creativity. This is necessary mainly for the reason of having an objective and automatic way of quantifying creative artifacts. In this work, we propose a regression-based learning framework which takes into account quantitatively the essential criteria for creativity like novelty, influence, value and unexpectedness. As it is often the case with most creative domains, there is no clear ground truth available for creativity. Our proposed learning framework is applicable to all creative domains; yet we evaluate it on a dataset of movies created from IMDb and Rotten Tomatoes due to availability of audience and critic scores, which can be used as proxy ground truth labels for creativity. We report promising results and observations from our experiments in the following ways : 1) Correlation of creative criteria with critic scores, 2) Improvement in movie rating prediction with inclusion of various creative criteria, and 3) Identification of creative movies.