TF-Coder: Program Synthesis for Tensor Manipulations
نویسندگان
چکیده
The success and popularity of deep learning is on the rise, partially due to powerful frameworks such as TensorFlow PyTorch, which make it easier develop models. However, these libraries also come with steep curves, since programming in quite different from traditional imperative explicit loops conditionals. In this work, we present a tool called TF-Coder for by example TensorFlow. uses bottom-up weighted enumerative search, value-based pruning equivalent expressions flexible type- filtering ensure that adhere various requirements imposed library. We train models predict operations features input output tensors natural language descriptions tasks prioritize relevant during search. solves 63 70 real-world within 5 minutes, sometimes finding simpler solutions less time compared experienced human programmers.
منابع مشابه
MathNMR: spin and spatial tensor manipulations in Mathematica.
Spin and spatial tensor manipulations are frequently required to describe the theory of NMR experiments. A Mathematica package is presented here, which provides some of the most common functions for these calculations. Examples are the calculation of matrix representations of operators, commutators, projections, rotations, Redfield matrix elements, matrix decomposition into basis operators, cha...
متن کاملUsing Program Synthesis for Program Analysis
In this paper, we identify a fragment of second-order logic with restricted quantification that is expressive enough to capture numerous static analysis problems (e.g. safety proving, bug finding, termination and non-termination proving, superoptimisation). We call this fragment the synthesis fragment. Satisfiability of a formula in the synthesis fragment is decidable over finite domains; speci...
متن کاملMaternal nutritional manipulations program adipose tissue dysfunction in offspring
Based on the concept of Developmental Origin of Health and Disease, both human and animal studies have demonstrated a close link between nutrient supply perturbations in the fetus or neonate (i.e., maternal undernutrition, obesity, gestational diabetes and/or rapid catch-up growth) and increased risk of adult-onset obesity. Indeed, the adipose tissue has been recognized as a key target of devel...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: ACM Transactions on Programming Languages and Systems
سال: 2022
ISSN: ['1558-4593', '0164-0925']
DOI: https://doi.org/10.1145/3517034