Generative models of code, pretrained on large corpora of programs, have shown great success in translating natural language to code (Chen et al., 2021; Austin et al., 2021; Li et al., 2022, inter alia). While these models do not explicitly incorporate program semantics (i.e., execution results) during training, they are able to generate correct solutions for many problems. However, choosing a single correct program from a generated set for each problem remains challenging. In this work, we introduce execution result--based minimum Bayes risk decoding (MBR-EXEC) for program selection and show that it improves the few-shot performance of pretrained code models on natural-language-to-code tasks. We select output programs from a generated candidate set by marginalizing over program implementations that share the same semantics. Because exact equivalence is intractable, we execute each program on a small number of test inputs to approximate semantic equivalence. Across datasets, execution or simulated execution significantly outperforms the methods that do not involve program semantics. We find that MBR-EXEC consistently improves over all execution-unaware selection methods, suggesting it as an effective approach for natural language to code translation. We open-source our code at github.com/facebookresearch/mbr-exec and data at dl.fbaipublicfiles.com/mbr-exec/mbr-exec-release.zip
We present SymForce, a library for fast symbolic computation, code generation, and nonlinear optimization for robotics applications like computer vision, motion planning, and controls. SymForce combines the development speed and flexibility of symbolic math with the performance of autogenerated, highly optimized code in C++ or any target runtime language. SymForce provides geometry and camera types, Lie group operations, and branchless singularity handling for creating and analyzing complex symbolic expressions in Python, built on top of SymPy. Generated functions can be integrated as factors into our tangent-space nonlinear optimizer, which is highly optimized for real-time production use. We introduce novel methods to automatically compute tangent-space Jacobians, eliminating the need for bug-prone handwritten derivatives. This workflow enables faster runtime code, faster development time, and fewer lines of handwritten code versus the state-of-the-art. Our experiments demonstrate that our approach can yield order of magnitude speedups on computational tasks core to robotics. Code is available at https://github.com/symforce-org/symforce.
Code is seldom written in a single left-to-right pass and is instead repeatedly edited and refined. We introduce InCoder, a unified generative model that can perform program synthesis (via left-to-right generation) as well as editing (via infilling). InCoder is trained to generate code files from a large corpus of permissively licensed code, where regions of code have been randomly masked and moved to the end of each file, allowing code infilling with bidirectional context. Our model is the first generative model that is able to directly perform zero-shot code infilling, which we evaluate on challenging tasks such as type inference, comment generation, and variable re-naming. We find that the ability to condition on bidirectional context substantially improves performance on these tasks, while still performing comparably on standard program synthesis benchmarks in comparison to left-to-right only models pretrained at similar scale. The InCoder models and code are publicly released. https://sites.google.com/view/incoder-code-models
We apply preference modeling and reinforcement learning from human feedback (RLHF) to finetune language models to act as helpful and harmless assistants. We find this alignment training improves performance on almost all NLP evaluations, and is fully compatible with training for specialized skills such as python coding and summarization. We explore an iterated online mode of training, where preference models and RL policies are updated on a weekly cadence with fresh human feedback data, efficiently improving our datasets and models. Finally, we investigate the robustness of RLHF training, and identify a roughly linear relation between the RL reward and the square root of the KL divergence between the policy and its initialization. Alongside our main results, we perform peripheral analyses on calibration, competing objectives, and the use of OOD detection, compare our models with human writers, and provide samples from our models using prompts appearing in recent related work.
Program synthesis strives to generate a computer program as a solution to a given problem specification, expressed with input-output examples or natural language descriptions. The prevalence of large language models advances the state-of-the-art for program synthesis, though limited training resources and data impede open access to such models. To democratize this, we train and release a family of large language models up to 16.1B parameters, called CODEGEN, on natural language and programming language data, and open source the training library JAXFORMER. We show the utility of the trained model by demonstrating that it is competitive with the previous state-of-the-art on zero-shot Python code generation on HumanEval. We further investigate the multi-step paradigm for program synthesis, where a single program is factorized into multiple prompts specifying subproblems. To this end, we construct an open benchmark, Multi-Turn Programming Benchmark (MTPB), consisting of 115 diverse problem sets that are factorized into multi-turn prompts. Our analysis on MTPB shows that the same intent provided to CODEGEN in multi-turn fashion significantly improves program synthesis over that provided as a single turn. We make the training library JAXFORMER and model checkpoints available as open source contribution: https://github.com/salesforce/CodeGen.
While there has been a recent burgeoning of applications at the intersection of natural and programming languages, such as code generation and code summarization, these applications are usually English-centric. This creates a barrier for program developers who are not proficient in English. To mitigate this gap in technology development across languages, we propose a multilingual dataset, MCoNaLa, to benchmark code generation from natural language commands extending beyond English. Modeled off of the methodology from the English Code/Natural Language Challenge (CoNaLa) dataset, we annotated a total of 896 NL-code pairs in three languages: Spanish, Japanese, and Russian. We present a quantitative evaluation of performance on the MCoNaLa dataset by testing with state-of-the-art code generation systems. While the difficulties vary across these three languages, all systems lag significantly behind their English counterparts, revealing the challenges in adapting code generation to new languages.
Programming is a powerful and ubiquitous problem-solving tool. Developing systems that can assist programmers or even generate programs independently could make programming more productive and accessible, yet so far incorporating innovations in AI has proven challenging. Recent large-scale language models have demonstrated an impressive ability to generate code, and are now able to complete simple programming tasks. However, these models still perform poorly when evaluated on more complex, unseen problems that require problem-solving skills beyond simply translating instructions into code. For example, competitive programming problems which require an understanding of algorithms and complex natural language remain extremely challenging. To address this gap, we introduce AlphaCode, a system for code generation that can create novel solutions to these problems that require deeper reasoning. In simulated evaluations on recent programming competitions on the Codeforces platform, AlphaCode achieved on average a ranking of top 54.3% in competitions with more than 5,000 participants. We found that three key components were critical to achieve good and reliable performance: (1) an extensive and clean competitive programming dataset for training and evaluation, (2) large and efficient-to-sample transformer-based architectures, and (3) large-scale model sampling to explore the search space, followed by filtering based on program behavior to a small set of submissions.
In education applications, knowledge tracing refers to the problem of estimating students' time-varying concept/skill mastery level from their past responses to questions and predicting their future performance. One key limitation of most existing knowledge tracing methods is that they treat student responses to questions as binary-valued, i.e., whether they are correct or incorrect. Response correctness analysis/prediction ignores important information on student knowledge contained in the exact content of the responses, especially for open-ended questions. In this paper, we conduct the first exploration into open-ended knowledge tracing (OKT) by studying the new task of predicting students' exact open-ended responses to questions. Our work is grounded in the domain of computer science education with programming questions. We develop an initial solution to the OKT problem, a student knowledge-guided code generation approach, that combines program synthesis methods using language models with student knowledge tracing methods. We also conduct a series of quantitative and qualitative experiments on a real-world student code dataset to validate OKT and demonstrate its promise in educational applications.
Considering the seq2seq architecture of TranX for natural language to code translation, we identify four key components of importance: grammatical constraints, lexical preprocessing, input representations, and copy mechanisms. To study the impact of these components, we use a state-of-the-art architecture that relies on BERT encoder and a grammar-based decoder for which a formalization is provided. The paper highlights the importance of the lexical substitution component in the current natural language to code systems.
Writing software exploits is an important practice for offensive security analysts to investigate and prevent attacks. In particular, shellcodes are especially time-consuming and a technical challenge, as they are written in assembly language. In this work, we address the task of automatically generating shellcodes, starting purely from descriptions in natural language, by proposing an approach based on Neural Machine Translation (NMT). We then present an empirical study using a novel dataset (Shellcode_IA32), which consists of 3,200 assembly code snippets of real Linux/x86 shellcodes from public databases, annotated using natural language. Moreover, we propose novel metrics to evaluate the accuracy of NMT at generating shellcodes. The empirical analysis shows that NMT can generate assembly code snippets from the natural language with high accuracy and that in many cases can generate entire shellcodes with no errors.
Automatic code generation from natural language descriptions can be highly beneficial during the process of software development. In this work, we propose GAP-Gen, a Guided Automatic Python Code Generation method based on Python syntactic constraints and semantic constraints. We first introduce Python syntactic constraints in the form of Syntax-Flow, which is a simplified version of Abstract Syntax Tree (AST) reducing the size and high complexity of Abstract Syntax Tree but maintaining crucial syntactic information of Python code. In addition to Syntax-Flow, we introduce Variable-Flow which abstracts variable and function names consistently through out the code. In our work, rather than pretraining, we focus on modifying the finetuning process which reduces computational requirements but retains high generation performance on automatic Python code generation task. GAP-Gen fine-tunes the transformer based language models T5 and CodeT5 using the Code-to-Docstring datasets CodeSearchNet, CodeSearchNet AdvTest and Code-Docstring Corpus from EdinburghNLP. Our experiments show that GAP-Gen achieves better results on automatic Python code generation task than previous works.
Auto-scheduling for tensor programs is a process where a search algorithm automatically explores candidate schedules (program transformations) for a given program on a target hardware platform to improve its performance. However this can be a very time consuming process depending on the complexity of the tensor program and the capacity of the target device, with often many thousands of program variants being explored. To address this, in this paper we introduce the idea of transfer-tuning, a novel approach to identify and reuse auto-schedules between tensor programs. We demonstrate this concept using Deep Neural Networks (DNNs), taking sets of auto-schedules from pre-tuned DNNs and using them to reduce the inference time of a new DNN. We compare transfer-tuning against the state-of-the-art Ansor auto-scheduler, defining the maximum possible speedup for a given DNN model as what Ansor achieves using its recommended full tuning time. On a server-class CPU and across 11 widely used DNN models, we observe that transfer-tuning achieves up to $88.41\%$ ($49.13\%$ on average) of this maximum speedup, while Ansor requires $6.5\times$ more search time on average to match it. We also evaluate transfer-tuning on a constrained edge CPU and observe that the differences in search time are exacerbated, with Ansor requiring $10.8\times$ more time on average to match transfer-tuning's speedup, which further demonstrates its value. Our code is available at https://www.github.com/gicLAB/transfer-tuning
Machine learning is shifting towards general-purpose pretrained generative models, trained in a self-supervised manner on large amounts of data, which can then be applied to solve a large number of tasks. However, due to their generic training methodology, these models often fail to meet some of the downstream requirements (e.g., hallucinations in abstractive summarization or style violations in code generation). This raises the important question of how to adapt pre-trained generative models to meet all requirements without destroying their general capabilities ("catastrophic forgetting"). Recent work has proposed to solve this problem by representing task-specific requirements through energy-based models (EBMs) and approximating these EBMs using distributional policy gradients (DPG). Despite its effectiveness, this approach is however limited to unconditional distributions. In this paper, we extend DPG to conditional tasks by proposing Conditional DPG (CDPG). We evaluate CDPG on four different control objectives across three tasks (translation, summarization and code generation) and two pretrained models (T5 and GPT-Neo). Our results show that fine-tuning using CDPG robustly moves these pretrained models closer towards meeting control objectives and -- in contrast with baseline approaches -- does not result in catastrophic forgetting.
Pre-trained models for Natural Languages (NL) like BERT and GPT have been recently shown to transfer well to Programming Languages (PL) and largely benefit a broad set of code-related tasks. Despite their success, most current methods either rely on an encoder-only (or decoder-only) pre-training that is suboptimal for generation (resp. understanding) tasks or process the code snippet in the same way as NL, neglecting the special characteristics of PL such as token types. We present CodeT5, a unified pre-trained encoder-decoder Transformer model that better leverages the code semantics conveyed from the developer-assigned identifiers. Our model employs a unified framework to seamlessly support both code understanding and generation tasks and allows for multi-task learning. Besides, we propose a novel identifier-aware pre-training task that enables the model to distinguish which code tokens are identifiers and to recover them when they are masked. Furthermore, we propose to exploit the user-written code comments with a bimodal dual generation task for better NL-PL alignment. Comprehensive experiments show that CodeT5 significantly outperforms prior methods on understanding tasks such as code defect detection and clone detection, and generation tasks across various directions including PL-NL, NL-PL, and PL-PL. Further analysis reveals that our model can better capture semantic information from code. Our code and pre-trained models are released at https: //github.com/salesforce/CodeT5 .
Recently, neural techniques have been used to generate source code automatically. While promising for declarative languages, these approaches achieve much poorer performance on datasets for imperative languages. Since a declarative language is typically embedded in an imperative language (i.e., the turducken-style programming) in real-world software development, the promising results on declarative languages can hardly lead to significant reduction of manual software development efforts. In this paper, we define a new code generation task: given a natural language comment, this task aims to generate a program in a base imperative language with an embedded declarative language. To our knowledge, this is the first turducken-style code generation task. For this task, we present Lyra: a dataset in Python with embedded SQL. This dataset contains 2,000 carefully annotated database manipulation programs from real-world projects. Each program is paired with both a Chinese comment and an English comment. In our experiment, we adopted Transformer, BERT-style, and GPT-style models as baselines. In the best setting, the generation performance of GPT-style models is better than others, where the AST exact matching accuracy is 24% and 25.5% when using Chinese and English comments, respectively. Therefore, we believe that Lyra provides a new challenge for code generation. Yet, overcoming this challenge may significantly boost the applicability of code generation techniques for real-world software development.
Software developers write a lot of source code and documentation during software development. Intrinsically, developers often recall parts of source code or code summaries that they had written in the past while implementing software or documenting them. To mimic developers' code or summary generation behavior, we propose a retrieval augmented framework, REDCODER, that retrieves relevant code or summaries from a retrieval database and provides them as a supplement to code generation or summarization models. REDCODER has a couple of uniqueness. First, it extends the state-of-the-art dense retrieval technique to search for relevant code or summaries. Second, it can work with retrieval databases that include unimodal (only code or natural language description) or bimodal instances (code-description pairs). We conduct experiments and extensive analysis on two benchmark datasets of code generation and summarization in Java and Python, and the promising results endorse the effectiveness of our proposed retrieval augmented framework.
We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of our problems with 100 samples per problem. Careful investigation of our model reveals its limitations, including difficulty with docstrings describing long chains of operations and with binding operations to variables. Finally, we discuss the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics.
We introduce a new type of programming challenge called programming puzzles, as an objective and comprehensive evaluation of program synthesis, and release an open-source dataset of Python Programming Puzzles (P3). Each puzzle is defined by a short Python program $f$, and the goal is to find an input which makes $f$ return True. The puzzles are objective in that each one is specified entirely by the source code of its verifier $f$, so evaluating $f$ is all that is needed to test a candidate solution. They do not require an answer key or input/output examples, nor do they depend on natural language understanding. The dataset is comprehensive in that it spans problems of a range of difficulties and domains, ranging from trivial string manipulation problems, to classic programming puzzles (e.g., Tower of Hanoi), to interview/competitive-programming problems (e.g., dynamic programming), to longstanding open problems in algorithms and mathematics (e.g., factoring). We develop baseline enumerative program synthesis, GPT-3 and Codex solvers that are capable of solving puzzles -- even without access to any reference solutions -- by learning from their own past solutions. Codex performs best, solving up to 18% of 397 test problems with a single try and 80% of the problems with 1,000 tries per problem. In a small user study, we find a positive correlation between puzzle-solving performance and coding experience, and between the puzzle difficulty for humans and AI solvers. Therefore, further improvements on P3 could have a significant impact on many program synthesis areas.
Answering a programming question using only its title is difficult as salient contextual information is omitted. Based on this observation, we present a corpus of over 40,000 StackOverflow question texts to be used in conjunction with their corresponding intents from the CoNaLa dataset (Yin et al., 2018). Using both the intent and question body, we use BART to establish a baseline BLEU score of 34.35 for this new task. We find further improvements of $2.8\%$ by combining the mined CoNaLa data with the labeled data to achieve a 35.32 BLEU score. We evaluate prior state-of-the-art CoNaLa models with this additional data and find that our proposed method of using the body and mined data beats the BLEU score of the prior state-of-the-art by $71.96\%$. Finally, we perform ablations to demonstrate that BART is an unsupervised multimodal learner and examine its extractive behavior. The code and data can be found https://github.com/gabeorlanski/stackoverflow-encourages-cheating.
Due to the great potential in facilitating software development, code generation has attracted increasing attention recently. Generally, dominant models are Seq2Tree models, which convert the input natural language description into a sequence of tree-construction actions corresponding to the pre-order traversal of an Abstract Syntax Tree (AST). However, such a traversal order may not be suitable for handling all multi-branch nodes. In this paper, we propose to equip the Seq2Tree model with a context-based Branch Selector, which is able to dynamically determine optimal expansion orders of branches for multi-branch nodes. Particularly, since the selection of expansion orders is a non-differentiable multi-step operation, we optimize the selector through reinforcement learning, and formulate the reward function as the difference of model losses obtained through different expansion orders. Experimental results and in-depth analysis on several commonly-used datasets demonstrate the effectiveness and generality of our approach. We have released our code at https://github.com/DeepLearnXMU/CG-RL.