Indraneil Paul

Clifton Poth, Hannah Sterz, Indraneil Paul et al.

We introduce Adapters, an open-source library that unifies parameter-efficient and modular transfer learning in large language models. By integrating 10 diverse adapter methods into a unified interface, Adapters offers ease of use and flexible configuration. Our library allows researchers and practitioners to leverage adapter modularity through composition blocks, enabling the design of complex adapter setups. We demonstrate the library's efficacy by evaluating its performance against full fine-tuning on various NLP tasks. Adapters provides a powerful tool for addressing the challenges of conventional fine-tuning paradigms and promoting more efficient and modular transfer learning. The library is available via https://adapterhub.ml/adapters.

Daniil Orel, Dilshod Azizov, Indraneil Paul et al.

Large language models (LLMs) are increasingly capable of generating functional source code, raising concerns about authorship, accountability, and security. While detecting AI-generated code is critical, existing datasets and benchmarks are narrow, typically limited to binary human-machine classification under in-distribution settings. To bridge this gap, we introduce $\emph{AICD Bench}$, the most comprehensive benchmark for AI-generated code detection. It spans $\emph{2M examples}$, $\emph{77 models}$ across $\emph{11 families}$, and $\emph{9 programming languages}$, including recent reasoning models. Beyond scale, AICD Bench introduces three realistic detection tasks: ($\emph{i}$)~$\emph{Robust Binary Classification}$ under distribution shifts in language and domain, ($\emph{ii}$)~$\emph{Model Family Attribution}$, grouping generators by architectural lineage, and ($\emph{iii}$)~$\emph{Fine-Grained Human-Machine Classification}$ across human, machine, hybrid, and adversarial code. Extensive evaluation on neural and classical detectors shows that performance remains far below practical usability, particularly under distribution shift and for hybrid or adversarial code. We release AICD Bench as a $\emph{unified, challenging evaluation suite}$ to drive the next generation of robust approaches for AI-generated code detection. The data and the code are available at https://huggingface.co/AICD-bench}.

Vatsal Venkatkrishna, Indraneil Paul, Iryna Gurevych

Multi-domain thinking verifiers trained via Reinforcement Learning with Verifiable Rewards (RLVR) are a cornerstone of modern post-training. However, their adoption in code generation has lagged behind execution feedback due to the prohibitive costs of the full RLVR pipeline. In this work, we ablate three primary drivers of RLVR performance and cost: intermediate thinking traces, learning from negative samples, and on-policy training. We introduce Aletheia, a controlled, execution-grounded testbed to facilitate a contamination-free analysis of code verifiers across disparate model sizes and covariate shifts. Our analysis reveals that the optimal training recipe is scale-dependent: on-policy learning is the primary performance driver for small verifiers, whereas thinking traces become the most vital factor for larger sizes. Furthermore, we show that negative samples stabilize training at large sizes, and scaling inference-time compute cannot compensate for any core RLVR component. These findings provide a compute-optimal roadmap for practitioners, offering concrete strategies to simplify verifier training based on model size. Consequently, our work establishes a foundation for scalable supervision, enabling efficiently trained code verifiers to reliably supervise much larger code generation policies.

Indraneil Paul, Glavaš Glavas, Iryna Gurevych

Reward models (RMs) have become an indispensable fixture of the language model (LM) post-training playbook, enabling policy alignment and test-time scaling. Research on the application of RMs in code generation, however, has been comparatively sparse, with existing work largely focusing on execution feedback. This choice constrains post-training to optimizing functional correctness over self-contained executable code. In this work, we examine the training and evaluation of multilingual, multi-criteria code RMs. To this end, we first compile Themis-CodeRewardBench, a benchmark to evaluate code RMs across five preference dimensions (i.e., criteria) and eight programming languages, on which we profile 50+ code, math, and general-purpose RMs. Observing the limited proficiency of current RMs beyond scoring for functional correctness, we develop Themis-CodePreference, the largest open-source collection of code preferences to date (more than 350k preference pairs), and use it to train Themis-RM, a suite of multilingual code reward models for flexible multi-criteria scoring, ranging in size from 600M to 32B parameters. Our experiments and ablations demonstrate positive scaling trends, strong cross-lingual transfer when training on diverse preferences, and the importance of multi-criteria training for reliable code reward modeling.

Anton Lozhkov, Raymond Li, Loubna Ben Allal et al. · berkeley, ibm-research

The BigCode project, an open-scientific collaboration focused on the responsible development of Large Language Models for Code (Code LLMs), introduces StarCoder2. In partnership with Software Heritage (SWH), we build The Stack v2 on top of the digital commons of their source code archive. Alongside the SWH repositories spanning 619 programming languages, we carefully select other high-quality data sources, such as GitHub pull requests, Kaggle notebooks, and code documentation. This results in a training set that is 4x larger than the first StarCoder dataset. We train StarCoder2 models with 3B, 7B, and 15B parameters on 3.3 to 4.3 trillion tokens and thoroughly evaluate them on a comprehensive set of Code LLM benchmarks. We find that our small model, StarCoder2-3B, outperforms other Code LLMs of similar size on most benchmarks, and also outperforms StarCoderBase-15B. Our large model, StarCoder2- 15B, significantly outperforms other models of comparable size. In addition, it matches or outperforms CodeLlama-34B, a model more than twice its size. Although DeepSeekCoder- 33B is the best-performing model at code completion for high-resource languages, we find that StarCoder2-15B outperforms it on math and code reasoning benchmarks, as well as several low-resource languages. We make the model weights available under an OpenRAIL license and ensure full transparency regarding the training data by releasing the SoftWare Heritage persistent IDentifiers (SWHIDs) of the source code data.

Shaoxiong Ji, Zihao Li, Jaakko Paavola et al.

This paper investigates a critical design decision in the practice of massively multilingual continual pre-training -- the inclusion of parallel data. Specifically, we study the impact of bilingual translation data for massively multilingual language adaptation of the Llama3 family of models to 500 languages. To this end, we construct the MaLA bilingual translation corpus, containing data from more than 2,500 language pairs. Subsequently, we develop the EMMA-500 Llama 3 suite of four massively multilingual models -- continually pre-trained from the Llama 3 family of base models extensively on diverse data mixes up to 671B tokens -- and explore the effect of continual pre-training with or without bilingual translation data. Comprehensive evaluation across 7 tasks and 12 benchmarks demonstrates that bilingual data tends to enhance language transfer and performance, particularly for low-resource languages. We open-source the MaLA corpus, EMMA-500 Llama 3 suite artefacts, code, and model generations.

Indraneil Paul, Goran Glavaš, Iryna Gurevych

Code understanding and generation have fast become some of the most popular applications of language models (LMs). Nonetheless, research on multilingual aspects of Code-LMs (i.e., LMs for code generation) such as cross-lingual transfer between different programming languages, language-specific data augmentation, and post-hoc LM adaptation, alongside exploitation of data sources other than the original textual content, has been much sparser than for their natural language counterparts. In particular, most mainstream Code-LMs have been pre-trained on source code files alone. In this work, we investigate the prospect of leveraging readily available compiler intermediate representations (IR) - shared across programming languages - to improve the multilingual capabilities of Code-LMs and facilitate cross-lingual transfer. To this end, we first compile SLTrans, a parallel dataset consisting of nearly 4M self-contained source code files coupled with respective intermediate representations. Next, starting from various base Code-LMs (ranging in size from 1.1B to 7.3B parameters), we carry out continued causal language modelling training on SLTrans, forcing the Code-LMs to (1) learn the IR language and (2) align the IR constructs with respective constructs of various programming languages. Our resulting models, dubbed IRCoder, display sizeable and consistent gains across a wide variety of code generation tasks and metrics, including prompt robustness, multilingual code completion, code understanding, and instruction following.

Indraneil Paul, Haoyi Yang, Goran Glavaš et al.

Language models (LMs) have become a staple of the code-writing toolbox. Their pre-training recipe has, however, remained stagnant over recent years, barring the occasional changes in data sourcing and filtering strategies. In particular, research exploring modifications to Code-LMs' pre-training objectives, geared towards improving data efficiency and better disentangling between syntax and semantics, has been noticeably sparse, especially compared with corresponding efforts in natural language LMs. In this work, we examine grounding on obfuscated code as a means of helping Code-LMs look beyond the surface-form syntax and enhance their pre-training sample efficiency. To this end, we compile ObscuraX, a dataset of approximately 55M source and obfuscated code pairs in seven languages. Subsequently, we pre-train ObscuraCoder models, ranging in size from 255M to 2.8B parameters, on a 272B-token corpus that includes ObscuraX and demonstrate that our obfuscation-based pre-training recipe leads to consistent improvements in Code-LMs' abilities compared to both vanilla autoregressive pre-training as well as existing de-obfuscation (DOBF) objectives. ObscuraCoder demonstrates sizeable gains across multiple tests of syntactic and semantic code understanding, along with improved capabilities in multilingual code completion, multilingual code commit summarization, and multi-purpose library-oriented code generation.

Daniil Orel, Indraneil Paul, Iryna Gurevych et al.

In this work, we compile $\textbf{$\texttt{DroidCollection}$}$, the most extensive open data suite for training and evaluating machine-generated code detectors, comprising over a million code samples, seven programming languages, outputs from 43 coding models, and over three real-world coding domains. Alongside fully AI-generated samples, our collection includes human-AI co-authored code, as well as adversarial samples explicitly crafted to evade detection. Subsequently, we develop $\textbf{$\texttt{DroidDetect}$}$, a suite of encoder-only detectors trained using a multi-task objective over $\texttt{DroidCollection}$. Our experiments show that existing detectors' performance fails to generalise to diverse coding domains and programming languages outside of their narrow training data. Additionally, we demonstrate that while most detectors are easily compromised by humanising the output distributions using superficial prompting and alignment approaches, this problem can be easily amended by training on a small amount of adversarial data. Finally, we demonstrate the effectiveness of metric learning and uncertainty-based resampling as means to enhance detector training on possibly noisy distributions.

Terry Yue Zhuo, Minh Chien Vu, Jenny Chim et al.

Task automation has been greatly empowered by the recent advances in Large Language Models (LLMs) via Python code, where the tasks ranging from software engineering development to general-purpose reasoning. While current benchmarks have shown that LLMs can solve tasks using programs like human developers, the majority of their evaluations are limited to short and self-contained algorithmic tasks or standalone function calls. Solving challenging and practical tasks requires the capability of utilizing diverse function calls as tools to efficiently implement functionalities like data analysis and web development. In addition, using multiple tools to solve a task needs compositional reasoning by accurately understanding complex instructions. Fulfilling both of these characteristics can pose a great challenge for LLMs.To assess how well LLMs can solve challenging and practical tasks via programs, we introduce BigCodeBench, a benchmark that challenges LLMs to invoke multiple function calls as tools from 139 libraries and 7 domains for 1,140 fine-grained tasks. To evaluate LLMs rigorously, each task encompasses 5.6 test cases with an average branch coverage of 99%. In addition, we propose a natural-language-oriented variant of BigCodeBench, BigCodeBench-Instruct, that automatically transforms the original docstrings into short instructions only with essential information. Our extensive evaluation of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%. The results underscore the need for further advancements in this area.

Indraneil Paul, Abhinav Khattar, Shaan Chopra et al.

Social network and publishing platforms, such as Twitter, support the concept of a secret proprietary verification process, for handles they deem worthy of platform-wide public interest. In line with significant prior work which suggests that possessing such a status symbolizes enhanced credibility in the eyes of the platform audience, a verified badge is clearly coveted among public figures and brands. What are less obvious are the inner workings of the verification process and what being verified represents. This lack of clarity, coupled with the flak that Twitter received by extending aforementioned status to political extremists in 2017, backed Twitter into publicly admitting that the process and what the status represented needed to be rethought. With this in mind, we seek to unravel the aspects of a user's profile which likely engender or preclude verification. The aim of the paper is two-fold: First, we test if discerning the verification status of a handle from profile metadata and content features is feasible. Second, we unravel the features which have the greatest bearing on a handle's verification status. We collected a dataset consisting of profile metadata of all 231,235 verified English-speaking users (as of July 2018), a control sample of 175,930 non-verified English-speaking users and all their 494 million tweets over a one year collection period. Our proposed models are able to reliably identify verification status (Area under curve AUC > 99%). We show that number of public list memberships, presence of neutral sentiment in tweets and an authoritative language style are the most pertinent predictors of verification status. To the best of our knowledge, this work represents the first attempt at discerning and classifying verification worthy users on Twitter.