NielsRogge/transformers

Pull Requests Commits

Imrpovements: model_path renamed pretrained_model, tokenizer loaded from pretrained_model, pretrained_model set to discriminator's when discrim is specified, sample = False by default but cli parameter introduced. To obtain identical samples call the cli with --sample

w4nderlust committed 6 years ago

f10b9250

Removed global variable device

w4nderlust committed 6 years ago

75904dae

Added support for generic discriminators

piero committed 6 years ago

7fd54b55

Added a +1 to epoch when saving weights

piero committed 6 years ago

b0eaff36

Added tqdm to preprocessing

piero committed 6 years ago

611961ad

Now run_pplm works on cpu. Identical output as before (when using gpu).

piero committed 6 years ago

afc7dcd9

Cleaned perturb_past. Identical output as before.

piero committed 6 years ago

61399e5a

Fix for making unditioned generation work. Identical output as before.

piero committed 6 years ago

ffc29354

Cleaned generate_text_pplm. Identical output as before.

piero committed 6 years ago

9f693a0c

Renamed SmallConst to SMALL_CONST and introduced BIG_CONST. Identical output as before.

piero committed 6 years ago

61a12f79

Removed commented code. Identical output as before.

piero committed 6 years ago

ef47b2c0

Removed commented code. Identical output as before.

piero committed 6 years ago

7ea12db3

Cleaned full_text_generation. Identical output as before.

piero committed 6 years ago

08c6e456

More cleanup for run_model. Identical output as before.

piero committed 6 years ago

6c9c1317

Improved device specification

piero committed 6 years ago

7ffe47c8

First cleanup step, changing function names and passing parameters all the way through without using args. Identical output as before.

piero committed 6 years ago

4f2164e4

piero committed 6 years ago

821de121

Fixed minor bug when running training on cuda

w4nderlust committed 6 years ago

7469d03b

Added script for training a discriminator for pplm to use

piero committed 6 years ago

0b51fba2

Let's make PPLM great again

Piero Molino committed 6 years ago

34a83faa

tokenizer white space: revert to previous behavior

julien-c committed 6 years ago

d5faa74c

rm extraneous import

julien-c committed 6 years ago

0b77d66a

fix the loss backward issue

mimosavvy committed 6 years ago

83b1e6ac

PPLM (squashed)

julien-c committed 6 years ago

572c24cf

Merge pull request #1903 from valohai/master

thomwolf committed 6 years ago

Verified f19a78a6

Merge pull request #2014 from aaugustin/mark-tf-auto-model-test-as-slow

thomwolf committed 6 years ago

Verified d100ad99

Change ref to original GLUE downloader script

Juha Kiili committed 6 years ago

66fc8d25

Remove annoying tokenization message

LysandreJik committed 6 years ago

fbaf05bd

Fix ALBERT exports with pretraining + sp classifier; Fix naming for ALBERT TF models

LysandreJik committed 6 years ago

e85855f2

Reorganize ALBERT conversion script

LysandreJik committed 6 years ago

b3d834ae

Newer Older