Search Results for "high"

Megatron-LM is a GPU-optimized deep learning framework from NVIDIA designed to train extremely large transformer-based language models efficiently at scale. The repository provides both a reference training implementation and Megatron Core, a composable library of high-performance building blocks for custom large-model pipelines. It supports advanced parallelism strategies including tensor, pipeline, data, expert, and context parallelism, enabling training across massive multi-GPU and multi-node clusters. The framework includes mixed-precision training options such as FP16, BF16, FP8, and FP4 to maximize performance and memory efficiency on modern hardware. ...

...Download OpenSIMPLY at opensimply.org The project is free computer simulation software based on discrete-event approach. OpenSIMPLY is a science software with very high simulation performance. So, the simulation results can be obtained much faster. The project is suitable for people with any modeling experience, and can be used for both science and education. Some application of this modeling software are: - traffic simulation - network simulation, - emergency and evacuation ways simulation and much more. ...

AllenNLP makes it easy to design and evaluate new deep learning models for nearly any NLP problem, along with the infrastructure to easily run them in the cloud or on your laptop. AllenNLP includes reference implementations of high quality models for both core NLP problems (e.g. semantic role labeling) and NLP applications (e.g. textual entailment). AllenNLP supports loading "plugins" dynamically. A plugin is just a Python package that provides custom registered classes or additional allennlp subcommands. There is ecosystem of open source plugins, some of which are maintained by the AllenNLP team here at AI2, and some of which are maintained by the broader community. ...

...The new ParseUtils.FindFormula method allows searching any of possible formulas within some text. Parse components are intended for Delphi developers and perform high-speed (about 10 million evaluations per second) mathematics and boolean calculations. All components are available for Delphi 6, Delphi 7, Delphi 2005, Delphi 2006, Delphi 2007, Delphi 2009, Delphi 2010, Delphi XE, Delphi XE2, Delphi XE3, Delphi XE4, Delphi XE5, Delphi XE6, Delphi XE7, Delphi XE8, Delphi 10 Seattle, Delphi 10.1 Berlin. ...

The Library is for Finite Transforms such as the Number Theoretic Transform (NTT) and Finite Radon Transform (FRT). Current modules include NTTW for NTTs with high resolution (microsecond) timing, basic array and imaging. The transforms are optimised for performance.

...Besides, this framework properly handles multiple threads, their communication, allows to synchronise control systems to external events (e.g. variable timers or incoming network packages) and provides many other nice features. Because of a high-level schematic-description language -- in form of provided Scilab commands -- only a minimal implementation-effort (no error-prone C-programming) is required.

Esg2Net is a set of tools and libraries used to transfer large datasets over high-bandwidth networks. It introduces a new mechanism in which every operation involving in data movement can be profiled, tuned, and adjusted on the fly in order to eliminate end-host bottlenecks. It has mainly developed for the movement of ESG climate datasets for high-speed replication and data access over the network. Climate Data Mover is used for testing 100Gbps networks in Nov 2011 at Supercomputing Conference.

This engine is not an attemp to re-inventing the wheel. It is a more learning-oriented project than something intended to provide a wide set of features. The approach to the OpenGL in this engine is something very experimental born from my own thoughts and can be usefull in future. If you want to help for your portfolio just ask: the TODO list is very long :)

The project contains a matlab package for generating poised lattices and coarse-fine interpolation matrices for high-order adaptive mesh refinement. It accompanies the paper by Qinghai Zhang submitted to Comput. Methods Appl. Mech. Engrg.

Java-based Image Server engine to organise high resolution images. The intermediate images will be generated transparently and available through defined URLs. The aim is to extended to Digital Asset Management Server

Integration routine to be employed in evaluation of high dimensional integrals arising from problems in statistical mechanics, more specifically the partition function of classical Heisenberg spin-chains, and its resulting derivatives.