Joshua 5.0: Sparser, Better, Faster, Server

نویسندگان

  • Matt Post
  • Juri Ganitkevitch
  • Luke Orland
  • Jonathan Weese
  • Yuan Cao
  • Chris Callison-Burch
چکیده

We describe improvements made over the past year to Joshua, an open-source translation system for parsing-based machine translation. The main contributions this past year are significant improvements in both speed and usability of the grammar extraction and decoding steps. We have also rewritten the decoder to use a sparse feature representation, enabling training of large numbers of features with discriminative training methods.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Compression by the Signs: Distributed Learning Is a Two-way Street

Training large neural networks requires distributing learning over multiple workers. The rate limiting step is often in sending gradients from workers to parameter server and back again. We present SIGNSGD with majority vote: the first gradient compression scheme to achieve 1-bit compression of worker-server communication in both directions with non-vacuous theoretical guarantees. To achieve th...

متن کامل

Finite Capacity Queuing System with Vacations and Server Breakdowns

This paper deals with finite capacity single server queuing system with vacations. Vacation starts at rate nu if the system is empty. Also the server takes another vacation if upon his arrival to the system, finds the system empty. Customers arrive in the system in Poisson fashion at rate lamda0 during vacation, faster rate lamdaf during active service and slower rate lamdas during the breakdow...

متن کامل

signSGD: compressed optimisation for non-convex problems

Training large neural networks requires distributing learning across multiple workers, where the cost of communicating gradients can be a significant bottleneck. SIGNSGD alleviates this problem by transmitting just the sign of each minibatch stochastic gradient. We prove that it can get the best of both worlds: compressed gradients and SGD-level convergence rate. SIGNSGD can exploit mismatches ...

متن کامل

Making sparse matrices sparser: Computational results

Many optimization algorithms involve repeated processing of a fixed set of linear constraints. If we pre-process the constraint matrix A to be sparser, then algebraic operations on A will become faster. We consider the problem of making a given matrix as sparse as possible, the Sparsity Problem (SP). In a companion paper with S. Frank Chang, we developed some theoretical algorithms for SP under...

متن کامل

Machine Repair Queueing System with with Non-Reliable Service Stations And Heterogeneous Service Discipline (RESEARCH NOTE)

This investigation deals with M/M/R/N machine repair problem with R non-reliable service stations which are subjected to unpredictable breakdown. 1here is provision of an additional server to reduce backlog in the case of heavy load of failed machines. 1he permanent service stations repair the failed machines at an identical rate m and switch to faster repair rate when all service stations are ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2013