Daya Guo

Zhiyu Li, Shuai Lu, Daya Guo et al. · microsoft-research

Code review is an essential part to software development lifecycle since it aims at guaranteeing the quality of codes. Modern code review activities necessitate developers viewing, understanding and even running the programs to assess logic, functionality, latency, style and other factors. It turns out that developers have to spend far too much time reviewing the code of their peers. Accordingly, it is in significant demand to automate the code review process. In this research, we focus on utilizing pre-training techniques for the tasks in the code review scenario. We collect a large-scale dataset of real-world code changes and code reviews from open-source projects in nine of the most popular programming languages. To better understand code diffs and reviews, we propose CodeReviewer, a pre-trained model that utilizes four pre-training tasks tailored specifically for the code review scenario. To evaluate our model, we focus on three key tasks related to code review activities, including code change quality estimation, review comment generation and code refinement. Furthermore, we establish a high-quality benchmark dataset based on our collected data for these three tasks and conduct comprehensive experiments on it. The experimental results demonstrate that our model outperforms the previous state-of-the-art pre-training approaches in all tasks. Further analysis show that our proposed pre-training tasks and the multilingual pre-training dataset benefit the model on the understanding of code changes and reviews.

Daya Guo, Canwen Xu, Nan Duan et al.

In this paper, we introduce a new task for code completion that focuses on handling long code input and propose a sparse Transformer model, called LongCoder, to address this task. LongCoder employs a sliding window mechanism for self-attention and introduces two types of globally accessible tokens - bridge tokens and memory tokens - to improve performance and efficiency. Bridge tokens are inserted throughout the input sequence to aggregate local information and facilitate global interaction, while memory tokens are included to highlight important statements that may be invoked later and need to be memorized, such as package imports and definitions of classes, functions, or structures. We conduct experiments on a newly constructed dataset that contains longer code context and the publicly available CodeXGLUE benchmark. Experimental results demonstrate that LongCoder achieves superior performance on code completion tasks compared to previous models while maintaining comparable efficiency in terms of computational resources during inference. All the codes and data are available at https://github.com/microsoft/CodeBERT.

Daya Guo, Shuai Lu, Nan Duan et al.

Pre-trained models for programming languages have recently demonstrated great success on code intelligence. To support both code-related understanding and generation tasks, recent works attempt to pre-train unified encoder-decoder models. However, such encoder-decoder framework is sub-optimal for auto-regressive tasks, especially code completion that requires a decoder-only manner for efficient inference. In this paper, we present UniXcoder, a unified cross-modal pre-trained model for programming language. The model utilizes mask attention matrices with prefix adapters to control the behavior of the model and leverages cross-modal contents like AST and code comment to enhance code representation. To encode AST that is represented as a tree in parallel, we propose a one-to-one mapping method to transform AST in a sequence structure that retains all structural information from the tree. Furthermore, we propose to utilize multi-modal contents to learn representation of code fragment with contrastive learning, and then align representations among programming languages using a cross-modal generation task. We evaluate UniXcoder on five code-related tasks over nine datasets. To further evaluate the performance of code fragment representation, we also construct a dataset for a new task, called zero-shot code-to-code search. Results show that our model achieves state-of-the-art performance on most tasks and analysis reveals that comment and AST can both enhance UniXcoder.

Caihua Li, Lianghong Guo, Yanlin Wang et al. · tencent-ai

Issue resolution, a complex Software Engineering (SWE) task integral to real-world development, has emerged as a compelling challenge for artificial intelligence. The establishment of benchmarks like SWE-bench revealed this task as profoundly difficult for large language models, thereby significantly accelerating the evolution of autonomous coding agents. This paper presents a systematic survey of this emerging domain. We begin by examining data construction pipelines, covering automated collection and synthesis approaches. We then provide a comprehensive analysis of methodologies, spanning training-free frameworks with their modular components to training-based techniques, including supervised fine-tuning and reinforcement learning. Subsequently, we discuss critical analyses of data quality and agent behavior, alongside practical applications. Finally, we identify key challenges and outline promising directions for future research. An open-source repository is maintained at https://github.com/DeepSoftwareAnalytics/Awesome-Issue-Resolution to serve as a dynamic resource in this field.

Shuai Lu, Nan Duan, Hojae Han et al.

Code completion, which aims to predict the following code token(s) according to the code context, can improve the productivity of software development. Recent work has proved that statistical language modeling with transformers can greatly improve the performance in the code completion task via learning from large-scale source code datasets. However, current approaches focus only on code context within the file or project, i.e. internal context. Our distinction is utilizing "external" context, inspired by human behaviors of copying from the related code snippets when writing code. Specifically, we propose a retrieval-augmented code completion framework, leveraging both lexical copying and referring to code with similar semantics by retrieval. We adopt a stage-wise training approach that combines a source code retriever and an auto-regressive language model for programming language. We evaluate our approach in the code completion task in Python and Java programming languages, achieving a state-of-the-art performance on CodeXGLUE benchmark.

Canwen Xu, Daya Guo, Nan Duan et al.

In this paper, we propose LaPraDoR, a pretrained dual-tower dense retriever that does not require any supervised data for training. Specifically, we first present Iterative Contrastive Learning (ICoL) that iteratively trains the query and document encoders with a cache mechanism. ICoL not only enlarges the number of negative instances but also keeps representations of cached examples in the same hidden space. We then propose Lexicon-Enhanced Dense Retrieval (LEDR) as a simple yet effective way to enhance dense retrieval with lexical matching. We evaluate LaPraDoR on the recently proposed BEIR benchmark, including 18 datasets of 9 zero-shot text retrieval tasks. Experimental results show that LaPraDoR achieves state-of-the-art performance compared with supervised dense retrieval models, and further analysis reveals the effectiveness of our training strategy and objectives. Compared to re-ranking, our lexicon-enhanced approach can be run in milliseconds (22.5x faster) while achieving superior performance.

Xiaonan Li, Daya Guo, Yeyun Gong et al.

Code contrastive pre-training has recently achieved significant progress on code-related tasks. In this paper, we present \textbf{SCodeR}, a \textbf{S}oft-labeled contrastive pre-training framework with two positive sample construction methods to learn functional-level \textbf{Code} \textbf{R}epresentation. Considering the relevance between codes in a large-scale code corpus, the soft-labeled contrastive pre-training can obtain fine-grained soft-labels through an iterative adversarial manner and use them to learn better code representation. The positive sample construction is another key for contrastive pre-training. Previous works use transformation-based methods like variable renaming to generate semantically equal positive codes. However, they usually result in the generated code with a highly similar surface form, and thus mislead the model to focus on superficial code structure instead of code semantics. To encourage SCodeR to capture semantic information from the code, we utilize code comments and abstract syntax sub-trees of the code to build positive samples. We conduct experiments on four code-related tasks over seven datasets. Extensive experimental results show that SCodeR achieves new state-of-the-art performance on all of them, which illustrates the effectiveness of the proposed pre-training method.

DeepSeek-AI, Daya Guo, Dejian Yang et al. · stanford, tsinghua

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrates remarkable reasoning capabilities. Through RL, DeepSeek-R1-Zero naturally emerges with numerous powerful and intriguing reasoning behaviors. However, it encounters challenges such as poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorporates multi-stage training and cold-start data before RL. DeepSeek-R1 achieves performance comparable to OpenAI-o1-1217 on reasoning tasks. To support the research community, we open-source DeepSeek-R1-Zero, DeepSeek-R1, and six dense models (1.5B, 7B, 8B, 14B, 32B, 70B) distilled from DeepSeek-R1 based on Qwen and Llama.

DeepSeek-AI, Aixin Liu, Bei Feng et al. · pku

We present DeepSeek-V2, a strong Mixture-of-Experts (MoE) language model characterized by economical training and efficient inference. It comprises 236B total parameters, of which 21B are activated for each token, and supports a context length of 128K tokens. DeepSeek-V2 adopts innovative architectures including Multi-head Latent Attention (MLA) and DeepSeekMoE. MLA guarantees efficient inference through significantly compressing the Key-Value (KV) cache into a latent vector, while DeepSeekMoE enables training strong models at an economical cost through sparse computation. Compared with DeepSeek 67B, DeepSeek-V2 achieves significantly stronger performance, and meanwhile saves 42.5% of training costs, reduces the KV cache by 93.3%, and boosts the maximum generation throughput to 5.76 times. We pretrain DeepSeek-V2 on a high-quality and multi-source corpus consisting of 8.1T tokens, and further perform Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) to fully unlock its potential. Evaluation results show that, even with only 21B activated parameters, DeepSeek-V2 and its chat versions still achieve top-tier performance among open-source models.

DeepSeek-AI, Xiao Bi, Deli Chen et al. · microsoft-research, pku

The rapid development of open-source large language models (LLMs) has been truly remarkable. However, the scaling law described in previous literature presents varying conclusions, which casts a dark cloud over scaling LLMs. We delve into the study of scaling laws and present our distinctive findings that facilitate scaling of large scale models in two commonly used open-source configurations, 7B and 67B. Guided by the scaling laws, we introduce DeepSeek LLM, a project dedicated to advancing open-source language models with a long-term perspective. To support the pre-training phase, we have developed a dataset that currently consists of 2 trillion tokens and is continuously expanding. We further conduct supervised fine-tuning (SFT) and Direct Preference Optimization (DPO) on DeepSeek LLM Base models, resulting in the creation of DeepSeek Chat models. Our evaluation results demonstrate that DeepSeek LLM 67B surpasses LLaMA-2 70B on various benchmarks, particularly in the domains of code, mathematics, and reasoning. Furthermore, open-ended evaluations reveal that DeepSeek LLM 67B Chat exhibits superior performance compared to GPT-3.5.

Junjie Huang, Daya Guo, Chenglong Wang et al.

Data wrangling, the process of preparing raw data for further analysis in computational notebooks, is a crucial yet time-consuming step in data science. Code generation has the potential to automate the data wrangling process to reduce analysts' overhead by translating user intents into executable code. Precisely generating data wrangling code necessitates a comprehensive consideration of the rich context present in notebooks, including textual context, code context and data context. However, notebooks often interleave multiple non-linear analysis tasks into linear sequence of code blocks, where the contextual dependencies are not clearly reflected. Directly training models with source code blocks fails to fully exploit the contexts for accurate wrangling code generation. To bridge the gap, we aim to construct a high quality datasets with clear and rich contexts to help training models for data wrangling code generation tasks. In this work, we first propose an automated approach, CoCoMine to mine data-wrangling code generation examples with clear multi-modal contextual dependency. It first adopts data flow analysis to identify the code blocks containing data wrangling codes. Then, CoCoMine extracts the contextualized datawrangling code examples through tracing and replaying notebooks. With CoCoMine, we construct CoCoNote, a dataset containing 58,221 examples for Contextualized Data-wrangling Code generation in Notebooks. To demonstrate the effectiveness of our dataset, we finetune a range of pretrained code models and prompt various large language models on our task. Furthermore, we also propose DataCoder, which encodes data context and code&textual contexts separately to enhance code generation. Experiment results demonstrate the significance of incorporating data context in data-wrangling code generation and the effectiveness of our model. We release code and data at url...

Hang Zhang, Yeyun Gong, Xingwei He et al.

Most dense retrieval models contain an implicit assumption: the training query-document pairs are exactly matched. Since it is expensive to annotate the corpus manually, training pairs in real-world applications are usually collected automatically, which inevitably introduces mismatched-pair noise. In this paper, we explore an interesting and challenging problem in dense retrieval, how to train an effective model with mismatched-pair noise. To solve this problem, we propose a novel approach called Noisy Pair Corrector (NPC), which consists of a detection module and a correction module. The detection module estimates noise pairs by calculating the perplexity between annotated positive and easy negative documents. The correction module utilizes an exponential moving average (EMA) model to provide a soft supervised signal, aiding in mitigating the effects of noise. We conduct experiments on text-retrieval benchmarks Natural Question and TriviaQA, code-search benchmarks StaQC and SO-DS. Experimental results show that NPC achieves excellent performance in handling both synthetic and realistic noise.

Z. Z. Ren, Zhihong Shao, Junxiao Song et al. · stanford, tsinghua

We introduce DeepSeek-Prover-V2, an open-source large language model designed for formal theorem proving in Lean 4, with initialization data collected through a recursive theorem proving pipeline powered by DeepSeek-V3. The cold-start training procedure begins by prompting DeepSeek-V3 to decompose complex problems into a series of subgoals. The proofs of resolved subgoals are synthesized into a chain-of-thought process, combined with DeepSeek-V3's step-by-step reasoning, to create an initial cold start for reinforcement learning. This process enables us to integrate both informal and formal mathematical reasoning into a unified model. The resulting model, DeepSeek-Prover-V2-671B, achieves state-of-the-art performance in neural theorem proving, reaching 88.9% pass ratio on the MiniF2F-test and solving 49 out of 658 problems from PutnamBench. In addition to standard benchmarks, we introduce ProverBench, a collection of 325 formalized problems, to enrich our evaluation, including 15 selected problems from the recent AIME competitions (years 24-25). Further evaluation on these 15 AIME problems shows that the model successfully solves 6 of them. In comparison, DeepSeek-V3 solves 8 of these problems using majority voting, highlighting that the gap between formal and informal mathematical reasoning in large language models is substantially narrowing.

Junlong Li, Daya Guo, Dejian Yang et al.

Reasoning is a fundamental capability of Large Language Models. While prior research predominantly focuses on enhancing narrow skills like math or code generation, improving performance on many other reasoning tasks remains challenging due to sparse and fragmented training data. To address this issue, we propose CodeI/O, a novel approach that systematically condenses diverse reasoning patterns inherently embedded in contextually-grounded codes, through transforming the original code into a code input-output prediction format. By training models to predict inputs/outputs given code and test cases entirely in natural language as Chain-of-Thought (CoT) rationales, we expose them to universal reasoning primitives -- like logic flow planning, state-space searching, decision tree traversal, and modular decomposition -- while decoupling structured reasoning from code-specific syntax and preserving procedural rigor. Experimental results demonstrate CodeI/O leads to consistent improvements across symbolic, scientific, logic, math & numerical, and commonsense reasoning tasks. By matching the existing ground-truth outputs or re-executing the code with predicted inputs, we can verify each prediction and further enhance the CoTs through multi-turn revision, resulting in CodeI/O++ and achieving higher performance. Our data and models are available at https://github.com/hkust-nlp/CodeIO.

Tianyue Jiang, Yanli Wang, Yanlin Wang et al.

Repository-level code completion remains a challenging task for existing code large language models (code LLMs) due to their limited understanding of repository-specific context and domain knowledge. While retrieval-augmented generation (RAG) approaches have shown promise by retrieving relevant code snippets as cross-file context, they suffer from two fundamental problems: misalignment between the query and the target code in the retrieval process, and the inability of existing retrieval methods to effectively utilize the inference information. To address these challenges, we propose AlignCoder, a repository-level code completion framework that introduces a query enhancement mechanism and a reinforcement learning based retriever training method. Our approach generates multiple candidate completions to construct an enhanced query that bridges the semantic gap between the initial query and the target code. Additionally, we employ reinforcement learning to train an AlignRetriever that learns to leverage inference information in the enhanced query for more accurate retrieval. We evaluate AlignCoder on two widely-used benchmarks (CrossCodeEval and RepoEval) across five backbone code LLMs, demonstrating an 18.1% improvement in EM score compared to baselines on the CrossCodeEval benchmark. The results show that our framework achieves superior performance and exhibits high generalizability across various code LLMs and programming languages.

DeepSeek-AI, Aixin Liu, Aoxue Mei et al.

We introduce DeepSeek-V3.2, a model that harmonizes high computational efficiency with superior reasoning and agent performance. The key technical breakthroughs of DeepSeek-V3.2 are as follows: (1) DeepSeek Sparse Attention (DSA): We introduce DSA, an efficient attention mechanism that substantially reduces computational complexity while preserving model performance in long-context scenarios. (2) Scalable Reinforcement Learning Framework: By implementing a robust reinforcement learning protocol and scaling post-training compute, DeepSeek-V3.2 performs comparably to GPT-5. Notably, our high-compute variant, DeepSeek-V3.2-Speciale, surpasses GPT-5 and exhibits reasoning proficiency on par with Gemini-3.0-Pro, achieving gold-medal performance in both the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI). (3) Large-Scale Agentic Task Synthesis Pipeline: To integrate reasoning into tool-use scenarios, we developed a novel synthesis pipeline that systematically generates training data at scale. This methodology facilitates scalable agentic post-training, yielding substantial improvements in generalization and instruction-following robustness within complex, interactive environments.

Zhihong Shao, Peiyi Wang, Qihao Zhu et al.

Mathematical reasoning poses a significant challenge for language models due to its complex and structured nature. In this paper, we introduce DeepSeekMath 7B, which continues pre-training DeepSeek-Coder-Base-v1.5 7B with 120B math-related tokens sourced from Common Crawl, together with natural language and code data. DeepSeekMath 7B has achieved an impressive score of 51.7% on the competition-level MATH benchmark without relying on external toolkits and voting techniques, approaching the performance level of Gemini-Ultra and GPT-4. Self-consistency over 64 samples from DeepSeekMath 7B achieves 60.9% on MATH. The mathematical reasoning capability of DeepSeekMath is attributed to two key factors: First, we harness the significant potential of publicly available web data through a meticulously engineered data selection pipeline. Second, we introduce Group Relative Policy Optimization (GRPO), a variant of Proximal Policy Optimization (PPO), that enhances mathematical reasoning abilities while concurrently optimizing the memory usage of PPO.

Huajian Xin, Daya Guo, Zhihong Shao et al. · pku

Proof assistants like Lean have revolutionized mathematical proof verification, ensuring high accuracy and reliability. Although large language models (LLMs) show promise in mathematical reasoning, their advancement in formal theorem proving is hindered by a lack of training data. To address this issue, we introduce an approach to generate extensive Lean 4 proof data derived from high-school and undergraduate-level mathematical competition problems. This approach involves translating natural language problems into formal statements, filtering out low-quality statements, and generating proofs to create synthetic data. After fine-tuning the DeepSeekMath 7B model on this synthetic dataset, which comprises 8 million formal statements with proofs, our model achieved whole-proof generation accuracies of 46.3% with 64 samples and 52% cumulatively on the Lean 4 miniF2F test, surpassing the baseline GPT-4 at 23.0% with 64 samples and a tree search reinforcement learning method at 41.0%. Additionally, our model successfully proved 5 out of 148 problems in the Lean 4 Formalized International Mathematical Olympiad (FIMO) benchmark, while GPT-4 failed to prove any. These results demonstrate the potential of leveraging large-scale synthetic data to enhance theorem-proving capabilities in LLMs. Both the synthetic dataset and the model will be made available to facilitate further research in this promising field.

DeepSeek-AI, Qihao Zhu, Daya Guo et al.

We present DeepSeek-Coder-V2, an open-source Mixture-of-Experts (MoE) code language model that achieves performance comparable to GPT4-Turbo in code-specific tasks. Specifically, DeepSeek-Coder-V2 is further pre-trained from an intermediate checkpoint of DeepSeek-V2 with additional 6 trillion tokens. Through this continued pre-training, DeepSeek-Coder-V2 substantially enhances the coding and mathematical reasoning capabilities of DeepSeek-V2, while maintaining comparable performance in general language tasks. Compared to DeepSeek-Coder-33B, DeepSeek-Coder-V2 demonstrates significant advancements in various aspects of code-related tasks, as well as reasoning and general capabilities. Additionally, DeepSeek-Coder-V2 expands its support for programming languages from 86 to 338, while extending the context length from 16K to 128K. In standard benchmark evaluations, DeepSeek-Coder-V2 achieves superior performance compared to closed-source models such as GPT4-Turbo, Claude 3 Opus, and Gemini 1.5 Pro in coding and math benchmarks.

DeepSeek-AI, Aixin Liu, Bei Feng et al. · stanford, tsinghua

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.

Daya Guo, Qihao Zhu, Dejian Yang et al.

The rapid development of large language models has revolutionized code intelligence in software development. However, the predominance of closed-source models has restricted extensive research and development. To address this, we introduce the DeepSeek-Coder series, a range of open-source code models with sizes from 1.3B to 33B, trained from scratch on 2 trillion tokens. These models are pre-trained on a high-quality project-level code corpus and employ a fill-in-the-blank task with a 16K window to enhance code generation and infilling. Our extensive evaluations demonstrate that DeepSeek-Coder not only achieves state-of-the-art performance among open-source code models across multiple benchmarks but also surpasses existing closed-source models like Codex and GPT-3.5. Furthermore, DeepSeek-Coder models are under a permissive license that allows for both research and unrestricted commercial use.

Canwen Xu, Daya Guo, Nan Duan et al.

Chat models, such as ChatGPT, have shown impressive capabilities and have been rapidly adopted across numerous domains. However, these models are only accessible through a restricted API, creating barriers for new research and progress in the field. We propose a pipeline that can automatically generate a high-quality multi-turn chat corpus by leveraging ChatGPT to engage in a conversation with itself. Subsequently, we employ parameter-efficient tuning to enhance LLaMA, an open-source large language model. The resulting model, named Baize, demonstrates good performance in multi-turn dialogues with guardrails that minimize potential risks. Furthermore, we propose a new technique called Self-Distill with Feedback, to further improve the performance of the Baize models with feedback from ChatGPT. The Baize models and data are released for research purposes only at https://github.com/project-baize/baize-chatbot. An online demo is also available at https://huggingface.co/spaces/project-baize/chat-with-baize.

Yanli Wang, Yanlin Wang, Suiquan Wang et al.

Repository-level code translation refers to translating an entire code repository from one programming language to another while preserving the functionality of the source repository. Many benchmarks have been proposed to evaluate the performance of such code translators. However, previous benchmarks mostly provide fine-grained samples, focusing at either code snippet, function, or file-level code translation. Such benchmarks do not accurately reflect real-world demands, where entire repositories often need to be translated, involving longer code length and more complex functionalities. To address this gap, we propose a new benchmark, named RepoTransBench, which is a real-world repository-level code translation benchmark with an automatically executable test suite. We conduct experiments on RepoTransBench to evaluate the translation performance of 11 advanced LLMs. We find that the Success@1 score (test success in one attempt) of the best-performing LLM is only 7.33%. To further explore the potential of LLMs for repository-level code translation, we provide LLMs with error-related feedback to perform iterative debugging and observe an average 7.09% improvement on Success@1. However, even with this improvement, the Success@1 score of the best-performing LLM is only 21%, which may not meet the need for reliable automatic repository-level code translation. Finally, we conduct a detailed error analysis and highlight current LLMs' deficiencies in repository-level code translation, which could provide a reference for further improvements.

Daya Guo, Zhaoyang Zeng

Video advertisement content structuring aims to segment a given video advertisement and label each segment on various dimensions, such as presentation form, scene, and style. Different from real-life videos, video advertisements contain sufficient and useful multi-modal content like caption and speech, which provides crucial video semantics and would enhance the structuring process. In this paper, we propose a multi-modal encoder to learn multi-modal representation from video advertisements by interacting between video-audio and text. Based on multi-modal representation, we then apply Boundary-Matching Network to generate temporal proposals. To make the proposals more accurate, we refine generated proposals by scene-guided alignment and re-ranking. Finally, we incorporate proposal located embeddings into the introduced multi-modal encoder to capture temporal relationships between local features of each proposal and global features of the whole video for classification. Experimental results show that our method achieves significantly improvement compared with several baselines and Rank 1 on the task of Multi-modal Ads Video Understanding in ACM Multimedia 2021 Grand Challenge. Ablation study further shows that leveraging multi-modal content like caption and speech in video advertisements significantly improve the performance.

Daya Guo, Alexey Svyatkovskiy, Jian Yin et al.

Code completion is usually cast as a language modelling problem, i.e., continuing an input in a left-to-right fashion. However, in practice, some parts of the completion (e.g., string literals) may be very hard to predict, whereas subsequent parts directly follow from the context. To handle this, we instead consider the scenario of generating code completions with "holes" inserted in places where a model is uncertain. We develop Grammformer, a Transformer-based model that guides code generation by the programming language grammar, and compare it to a variety of more standard sequence models. We train the models on code completion for C# and Python given partial code context. To evaluate models, we consider both ROUGE as well as a new metric RegexAcc that measures success of generating completions matching long outputs with as few holes as possible. In our experiments, Grammformer generates 10-50% more accurate completions compared to traditional generative models and 37-50% longer sketches compared to sketch-generating baselines trained with similar techniques.

Wanjun Zhong, Siyuan Wang, Duyu Tang et al.

Analytical reasoning is an essential and challenging task that requires a system to analyze a scenario involving a set of particular circumstances and perform reasoning over it to make conclusions. In this paper, we study the challenge of analytical reasoning of text and introduce a new dataset consisting of questions from the Law School Admission Test from 1991 to 2016. We analyze what knowledge understanding and reasoning abilities are required to do well on this task. Furthermore, to address this reasoning challenge, we design two different baselines: (1) a Transformer-based method which leverages the state-of-the-art pre-trained language models and (2) Analytical Reasoning Machine (ARM), a logical-level reasoning framework extracting symbolic knowledge (e.g, participants, facts, logical functions) to deduce legitimate solutions. In our experiments, we find that the Transformer-based models struggle to solve this task as their performance is close to random guess and ARM achieves better performance by leveraging symbolic knowledge and interpretable reasoning steps. Results show that both methods still lag far behind human performance, which leave further space for future research.

Shuai Lu, Daya Guo, Shuo Ren et al.

Benchmark datasets have a significant impact on accelerating research in programming language tasks. In this paper, we introduce CodeXGLUE, a benchmark dataset to foster machine learning research for program understanding and generation. CodeXGLUE includes a collection of 10 tasks across 14 datasets and a platform for model evaluation and comparison. CodeXGLUE also features three baseline systems, including the BERT-style, GPT-style, and Encoder-Decoder models, to make it easy for researchers to use the platform. The availability of such data and baselines can help the development and validation of new methods that can be applied to various program understanding and generation problems.

Zenan Xu, Daya Guo, Duyu Tang et al.

We study the problem of leveraging the syntactic structure of text to enhance pre-trained models such as BERT and RoBERTa. Existing methods utilize syntax of text either in the pre-training stage or in the fine-tuning stage, so that they suffer from discrepancy between the two stages. Such a problem would lead to the necessity of having human-annotated syntactic information, which limits the application of existing methods to broader scenarios. To address this, we present a model that utilizes the syntax of text in both pre-training and fine-tuning stages. Our model is based on Transformer with a syntax-aware attention layer that considers the dependency tree of the text. We further introduce a new pre-training task of predicting the syntactic distance among tokens in the dependency tree. We evaluate the model on three downstream tasks, including relation classification, entity typing, and question answering. Results show that our model achieves state-of-the-art performance on six public benchmark datasets. We have two major findings. First, we demonstrate that infusing automatically produced syntax of text improves pre-trained models. Second, global syntactic distances among tokens bring larger performance gains compared to local head relations between contiguous tokens.

Shuo Ren, Daya Guo, Shuai Lu et al.

Evaluation metrics play a vital role in the growth of an area as it defines the standard of distinguishing between good and bad models. In the area of code synthesis, the commonly used evaluation metric is BLEU or perfect accuracy, but they are not suitable enough to evaluate codes, because BLEU is originally designed to evaluate the natural language, neglecting important syntactic and semantic features of codes, and perfect accuracy is too strict thus it underestimates different outputs with the same semantic logic. To remedy this, we introduce a new automatic evaluation metric, dubbed CodeBLEU. It absorbs the strength of BLEU in the n-gram match and further injects code syntax via abstract syntax trees (AST) and code semantics via data-flow. We conduct experiments by evaluating the correlation coefficient between CodeBLEU and quality scores assigned by the programmers on three code synthesis tasks, i.e., text-to-code, code translation, and code refinement. Experimental results show that our proposed CodeBLEU can achieve a better correlation with programmer assigned scores compared with BLEU and accuracy.

Daya Guo, Shuo Ren, Shuai Lu et al.

Pre-trained models for programming language have achieved dramatic empirical improvements on a variety of code-related tasks such as code search, code completion, code summarization, etc. However, existing pre-trained models regard a code snippet as a sequence of tokens, while ignoring the inherent structure of code, which provides crucial code semantics and would enhance the code understanding process. We present GraphCodeBERT, a pre-trained model for programming language that considers the inherent structure of code. Instead of taking syntactic-level structure of code like abstract syntax tree (AST), we use data flow in the pre-training stage, which is a semantic-level structure of code that encodes the relation of "where-the-value-comes-from" between variables. Such a semantic-level structure is neat and does not bring an unnecessarily deep hierarchy of AST, the property of which makes the model more efficient. We develop GraphCodeBERT based on Transformer. In addition to using the task of masked language modeling, we introduce two structure-aware pre-training tasks. One is to predict code structure edges, and the other is to align representations between source code and code structure. We implement the model in an efficient way with a graph-guided masked attention function to incorporate the code structure. We evaluate our model on four tasks, including code search, clone detection, code translation, and code refinement. Results show that code structure and newly introduced pre-training tasks can improve GraphCodeBERT and achieves state-of-the-art performance on the four downstream tasks. We further show that the model prefers structure-level attentions over token-level attentions in the task of code search.

Daya Guo, Duyu Tang, Nan Duan et al.

Generating inferential texts about an event in different perspectives requires reasoning over different contexts that the event occurs. Existing works usually ignore the context that is not explicitly provided, resulting in a context-independent semantic representation that struggles to support the generation. To address this, we propose an approach that automatically finds evidence for an event from a large text corpus, and leverages the evidence to guide the generation of inferential texts. Our approach works in an encoder-decoder manner and is equipped with a Vector Quantised-Variational Autoencoder, where the encoder outputs representations from a distribution over discrete variables. Such discrete representations enable automatically selecting relevant evidence, which not only facilitates evidence-aware generation, but also provides a natural way to uncover rationales behind the generation. Our approach provides state-of-the-art performance on both Event2Mind and ATOMIC datasets. More importantly, we find that with discrete representations, our model selectively uses evidence to generate different inferential texts.

Shangwen Lv, Yuechen Wang, Daya Guo et al.

Pre-training text representations has recently been shown to significantly improve the state-of-the-art in many natural language processing tasks. The central goal of pre-training is to learn text representations that are useful for subsequent tasks. However, existing approaches are optimized by minimizing a proxy objective, such as the negative log likelihood of language modeling. In this work, we introduce a learning algorithm which directly optimizes model's ability to learn text representations for effective learning of downstream tasks. We show that there is an intrinsic connection between multi-task pre-training and model-agnostic meta-learning with a sequence of meta-train steps. The standard multi-task learning objective adopted in BERT is a special case of our learning algorithm where the depth of meta-train is zero. We study the problem in two settings: unsupervised pre-training and supervised pre-training with different pre-training objects to verify the generality of our approach.Experimental results show that our algorithm brings improvements and learns better initializations for a variety of downstream tasks.

Daya Guo, Akari Asai, Duyu Tang et al.

We study the problem of generating inferential texts of events for a variety of commonsense like \textit{if-else} relations. Existing approaches typically use limited evidence from training examples and learn for each relation individually. In this work, we use multiple knowledge sources as fuels for the model. Existing commonsense knowledge bases like ConceptNet are dominated by taxonomic knowledge (e.g., \textit{isA} and \textit{relatedTo} relations), having a limited number of inferential knowledge. We use not only structured commonsense knowledge bases, but also natural language snippets from search-engine results. These sources are incorporated into a generative base model via key-value memory network. In addition, we introduce a meta-learning based multi-task learning algorithm. For each targeted commonsense relation, we regard the learning of examples from other relations as the meta-training process, and the evaluation on examples from the targeted relation as the meta-test process. We conduct experiments on Event2Mind and ATOMIC datasets. Results show that both the integration of multiple knowledge sources and the use of the meta-learning algorithm improve the performance.

Zhangyin Feng, Daya Guo, Duyu Tang et al.

We present CodeBERT, a bimodal pre-trained model for programming language (PL) and nat-ural language (NL). CodeBERT learns general-purpose representations that support downstream NL-PL applications such as natural language codesearch, code documentation generation, etc. We develop CodeBERT with Transformer-based neural architecture, and train it with a hybrid objective function that incorporates the pre-training task of replaced token detection, which is to detect plausible alternatives sampled from generators. This enables us to utilize both bimodal data of NL-PL pairs and unimodal data, where the former provides input tokens for model training while the latter helps to learn better generators. We evaluate CodeBERT on two NL-PL applications by fine-tuning model parameters. Results show that CodeBERT achieves state-of-the-art performance on both natural language code search and code documentation generation tasks. Furthermore, to investigate what type of knowledge is learned in CodeBERT, we construct a dataset for NL-PL probing, and evaluate in a zero-shot setting where parameters of pre-trained models are fixed. Results show that CodeBERT performs better than previous pre-trained models on NL-PL probing.

Tao Shen, Xiubo Geng, Tao Qin et al.

We consider the problem of conversational question answering over a large-scale knowledge base. To handle huge entity vocabulary of a large-scale knowledge base, recent neural semantic parsing based approaches usually decompose the task into several subtasks and then solve them sequentially, which leads to following issues: 1) errors in earlier subtasks will be propagated and negatively affect downstream ones; and 2) each subtask cannot naturally share supervision signals with others. To tackle these issues, we propose an innovative multi-task learning framework where a pointer-equipped semantic parsing model is designed to resolve coreference in conversations, and naturally empower joint learning with a novel type-aware entity detection model. The proposed framework thus enables shared supervisions and alleviates the effect of error propagation. Experiments on a large-scale conversational question answering dataset containing 1.6M question answering pairs over 12.8M entities show that the proposed framework improves overall F1 score from 67% to 79% compared with previous state-of-the-art work.

Shangwen Lv, Daya Guo, Jingjing Xu et al.

Commonsense question answering aims to answer questions which require background knowledge that is not explicitly expressed in the question. The key challenge is how to obtain evidence from external knowledge and make predictions based on the evidence. Recent works either learn to generate evidence from human-annotated evidence which is expensive to collect, or extract evidence from either structured or unstructured knowledge bases which fails to take advantages of both sources. In this work, we propose to automatically extract evidence from heterogeneous knowledge sources, and answer questions based on the extracted evidence. Specifically, we extract evidence from both structured knowledge base (i.e. ConceptNet) and Wikipedia plain texts. We construct graphs for both sources to obtain the relational structures of evidence. Based on these graphs, we propose a graph-based approach consisting of a graph-based contextual word representation learning module and a graph-based inference module. The first module utilizes graph structural information to re-define the distance between words for learning better contextual word representations. The second module adopts graph convolutional network to encode neighbor information into the representations of nodes, and aggregates evidence with graph attention mechanism for predicting the final answer. Experimental results on CommonsenseQA dataset illustrate that our graph-based approach over both knowledge sources brings improvement over strong baselines. Our approach achieves the state-of-the-art accuracy (75.3%) on the CommonsenseQA leaderboard.

Daya Guo, Duyu Tang, Nan Duan et al.

In this paper, we present an approach to incorporate retrieved datapoints as supporting evidence for context-dependent semantic parsing, such as generating source code conditioned on the class environment. Our approach naturally combines a retrieval model and a meta-learner, where the former learns to find similar datapoints from the training data, and the latter considers retrieved datapoints as a pseudo task for fast adaptation. Specifically, our retriever is a context-aware encoder-decoder model with a latent variable which takes context environment into consideration, and our meta-learner learns to utilize retrieved datapoints in a model-agnostic meta-learning paradigm for fast adaptation. We conduct experiments on CONCODE and CSQA datasets, where the context refers to class environment in JAVA codes and conversational history, respectively. We use sequence-to-action model as the base semantic parser, which performs the state-of-the-art accuracy on both datasets. Results show that both the context-aware retriever and the meta-learning strategy improve accuracy, and our approach performs better than retrieve-and-edit baselines.

Yibo Sun, Daya Guo, Duyu Tang et al.

Machine reading comprehension (MRC) requires reasoning about both the knowledge involved in a document and knowledge about the world. However, existing datasets are typically dominated by questions that can be well solved by context matching, which fail to test this capability. To encourage the progress on knowledge-based reasoning in MRC, we present knowledge-based MRC in this paper, and build a new dataset consisting of 40,047 question-answer pairs. The annotation of this dataset is designed so that successfully answering the questions requires understanding and the knowledge involved in a document. We implement a framework consisting of both a question answering model and a question generation model, both of which take the knowledge extracted from the document as well as relevant facts from an external knowledge base such as Freebase/ProBase/Reverb/NELL. Results show that incorporating side information from external KB improves the accuracy of the baseline question answer system. We compare it with a standard MRC model BiDAF, and also provide the difficulty of the dataset and lay out remaining challenges.

Daya Guo, Yibo Sun, Duyu Tang et al.

We study how to learn a semantic parser of state-of-the-art accuracy with less supervised training data. We conduct our study on WikiSQL, the largest hand-annotated semantic parsing dataset to date. First, we demonstrate that question generation is an effective method that empowers us to learn a state-of-the-art neural network based semantic parser with thirty percent of the supervised training data. Second, we show that applying question generation to the full supervised training data further improves the state-of-the-art model. In addition, we observe that there is a logarithmic relationship between the accuracy of a semantic parser and the amount of training data.