python ninja_setup.py download and installs the latest Ninja release. Ninja is highly recommended to use with CMake on Windows, Mac and Linux.
increase CI coverage
Contributors:Thomas Wolf, Lysandre Debut, Julien Chaumond, Victor SANH, Patrick von Platen, Aymeric Augustin, Rémi Louf, Funtowicz Morgan, Stefan Schweter, Denis, Sam Shleifer, erenup, Manuel Romero, Matt, Piero Molino, Grégory Châtel, Bram Vanroy, Tim Rault, Gunnlaugur Thor Briem, Julien Plu, Anthony MOI, Malte Pietsch, Catalin Voss, Bilal Khan, Fei Wang, Martin Malmsten, Louis Martin, Davide Fiocco, Clement, Ananya Harsh Jha
ELECTRA Model (@LysandreJik)
ELECTRA is a new method for self-supervised language representation learning. It can be used to pre-train transformer networks using relatively little compute. ELECTRA models are trained to distinguish "real" input tokens vs "fake" input tokens generated by another neural network, similar to the discriminator of a GAN. At small scale, ELECTRA achieves strong results even when trained on a single GPU. At large scale, ELECTRA achieves state-of-the-art results on the SQuAD 2.0 dataset.
This release comes with 6 ELECTRA checkpoints:
Models available in the community models
Thanks to the author @clarkkev for his help during the implementation.
Bad word filters in generate (@patrickvonplaten)
The generate method now has a bad word filter.
Fixes and improvements
Decoder input ids are not necessary for T5 training anymore (@patrickvonplaten)
Update encoder and decoder on set_input_embedding for BART (@sshleifer)
Using loaded checkpoint with --do_predict (instead of random init) for Pytorch-lightning scripts (@ethanjperez)
Clean summarization and translation example testing files for T5 and Bart (@patrickvonplaten)
Cleaner examples (@julien-c)
Extensive testing for T5 model (@patrickvonplaten)
Force models outputs to always have batch_size as their first dim (@patrickvonplaten)
Fix for continuing training in some scripts (@xeb)
Resizing embedding matrix before sending it to the optimizer (@ngarneau)
BertJapaneseTokenizer accept options for mecab (@tamuhey)
Speed up GELU computation with torch.jit (@mryab)
fix argument order of update_mems fn in TF version (@patrickvonplaten, @dmytyar)
Split generate test function into beam search, no beam search (@patrickvonplaten)
Contributors:Robin, Richard Ellison, Malcolm Morgan, Andrea Gilardi, mark padgham, Nikolai, Karthik Ram, Layik Hama, Steven (Siwei) Ye, Martin, ilanfri, Maëlle Salmon, Jakub Nowosad, Ali Abbas, Scott Chamberlain, Andrew Smith, seanolondon, Merlin Unterfinger, hxd1011, Jeroen Ooms, James, Edzer Pebesma, Colin Caine
Documentation fixes (#384)
A bug affecting route() function calls when pbapply was not installed has been fixed (#386)
oneway() has been deprecated in favour of faster and easier-to-maintain function od_oneway() (also see the in-development od package) (#387)
Various other changes have been made to accomodate the dev version of dplyr (#383)
Here we publish the testing code used in our "DUNCAN: Disentangled Unsupervised Cycle-consistent Adversarial Network for Retrospective Artifact Correction of Structural MRI" article.
This is the first release. Code loads pre-processed data and steps through PCA, SOM, regression, and reconstruction of streamflow to identify communities whose water supply will change the most by the loss of glacier ice in Alberta.
No description provided.
Contributors:Timothée Giraud, dieghernan, Nicolas LAMBERT, MBunel, Arthur Gailes, Alfonso, 黄湘云, Robin Cura
getTiles(): possibility to use Thunderforest tiles with the apikey arg (@dieghernan)
getTiles(): new cachedir and forceDownload args to control tiles caching (@arthurgailes & @dieghernan)
getTiles(): possibility to use a custom tiles server (using a list in type arg)
barscale(): new parameter unit to plot the scalebar in kilometers, miles or meters (@arthurgailes)
add hatchedLayer() and legendHatched() to represent hatched layers (@dieghernan)
add getPngLayer() and pngLayer() to import and display png images (@dieghernan)
add wordcloudLayer() to display words à la wordcloud (@dieghernan)
add waffleLayer() and legenWaffle() to plot waffle maps (several quantitatives variables on the same map)
getTiles() gets a lot of new tiles sources (around 60) (@dieghernan)
bug fix in getTiles when x!=POINT & nrow(x) == 1, see #59 (thanks to @drikkes-jue)
tilesLayer() is enhanced with the possibility to use args from raster::plotRGB()
getBreaks() is now a full wrap of classIntervals + few other methods.
remove unused files martinique.shp
chimFinder is designed to identify reads from RNA and DNA sequencing experiments which align to two distinct genomes. This tool was developed as part of the work on identification of fusion transcripts resulting from HIV type 1 integrations in human CD4+ T cells.