 wassname
|
3e7fdd5a3b
|
More examples
|
2018-11-05 08:15:23 +08:00 |
|
|
wassname
|
ed0c76ce82
|
Save/load
|
2018-11-04 22:45:47 +08:00 |
|
|
wassname
|
c9f8cb90ff
|
readme, clean up
|
2018-11-04 19:19:27 +08:00 |
|
|
wassname
|
1db17076fe
|
using language model from horoscope_language_model
https://github.com/rodgzilla/pytorch-openai-transformer-lm/blob/horoscope_language_model
|
2018-11-04 17:38:07 +08:00 |
|
|
wassname
|
2f9faab044
|
with decoding language model
|
2018-11-04 15:57:28 +08:00 |
|
|
wassname
|
87b738c089
|
clean
|
2018-11-04 13:23:29 +08:00 |
|
|
wassname
|
208daec592
|
working with gutenberg data set but multiheaded
|
2018-11-04 13:22:49 +08:00 |
|
|
Thomas Wolf
|
eafc28abdf
|
Merge pull request #37 from rodgzilla/removing_unused_decoder
Removing unused decoder for TransformedModel class.
|
2018-08-27 13:25:52 +02:00 |
|
|
Grégory Châtel
|
91b84b0acd
|
Removing unused decoder for TransformedModel class.
|
2018-08-27 13:13:58 +02:00 |
|
|
Thomas Wolf
|
a287899b79
|
Merge pull request #27 from sharpsy/sharpsy/add-position-info-comment
Add comment explaining the encoding of the position information.
|
2018-07-31 15:46:14 +02:00 |
|
|
Marko Pranjic
|
4e73dc3e00
|
Add comment explaining the encoding of the position information.
|
2018-07-18 14:35:17 +02:00 |
|
|
Thomas Wolf
|
561d4096be
|
Merge pull request #25 from rodgzilla/multiple_choice_head
Simplifying the use of the model to perform different tasks
|
2018-07-18 10:43:44 +02:00 |
|
|
Thomas Wolf
|
d9142286d3
|
Merge pull request #23 from rodgzilla/master
Removing unused field of the Transformer class.
|
2018-07-18 10:35:42 +02:00 |
|
|
Grégory Châtel
|
cbccdb0004
|
Fixing a syntax problem in SimilarityHead.
|
2018-07-16 11:18:52 +02:00 |
|
|
Grégory Châtel
|
cf65569610
|
Adding TODO and fixing indentation.
|
2018-07-16 11:09:37 +02:00 |
|
|
Grégory Châtel
|
08b83b4036
|
Moving loss function to a new module.
|
2018-07-16 10:40:46 +02:00 |
|
|
Grégory Châtel
|
e1adeda208
|
Fixing indentation error.
|
2018-07-16 10:24:55 +02:00 |
|
|
Grégory Châtel
|
df88759637
|
Adding inference option to DoubleHeadModel constructor.
|
2018-07-13 17:51:54 +02:00 |
|
|
Grégory Châtel
|
87b4901a81
|
Modifying the code of DoubleHeadModel to allow different task heads.
|
2018-07-13 17:27:33 +02:00 |
|
|
Grégory Châtel
|
ac2250881a
|
Adding the addition of the two sentence orderings in similarity head.
|
2018-07-13 17:26:32 +02:00 |
|
|
Grégory Châtel
|
ea7f5006d5
|
Fixing similarity head to output a (batch, 3) dimensional tensor.
|
2018-07-13 17:25:26 +02:00 |
|
|
Grégory Châtel
|
4e6775287d
|
Small changes to task head code.
|
2018-07-13 16:45:27 +02:00 |
|
|
Grégory Châtel
|
7270ba812e
|
Similarity Head for the transformer.
|
2018-07-13 16:45:02 +02:00 |
|
|
Grégory Châtel
|
e049859323
|
More generic DoubleHeadModel (WIP).
|
2018-07-12 14:43:11 +02:00 |
|
|
Grégory Châtel
|
5f3348430d
|
New ClfHead module.
|
2018-07-12 14:34:22 +02:00 |
|
|
Grégory Châtel
|
ed8bb28b50
|
Renaming ClfHead in MultipleChoiceHead.
|
2018-07-12 14:23:09 +02:00 |
|
|
Grégory Châtel
|
93522a3b59
|
Removing unused field of the Transformer class.
|
2018-07-12 14:14:17 +02:00 |
|
|
Thomas Wolf
|
ea08605319
|
Merge pull request #22 from rodgzilla/noise_shape_dropout
Noise shape dropout
|
2018-07-12 11:18:36 +02:00 |
|
|
Grégory Châtel
|
e40f8811fc
|
Adding commentaries and fixing a syntax mistake.
|
2018-07-11 18:01:48 +02:00 |
|
|
Grégory Châtel
|
1c6c802545
|
Replacating the noise shape behavior.
|
2018-07-11 17:57:15 +02:00 |
|
|
Thomas Wolf
|
253ca422bb
|
Merge pull request #20 from rodgzilla/fixing_travis
Removing python 2.7 check from Travis.
|
2018-07-09 14:04:21 +02:00 |
|
|
Grégory Châtel
|
90bbbf055e
|
Removing python 2.7 check from Travis.
|
2018-07-09 13:57:48 +02:00 |
|
|
Thomas Wolf
|
55ba4d7840
|
Merge pull request #17 from rodgzilla/data_encoding_cleaning
Cleaning the data encoding code.
|
2018-07-05 15:43:07 +02:00 |
|
|
Thomas Wolf
|
7adb2a0091
|
Merge pull request #16 from rodgzilla/master
Updating README.md
|
2018-07-05 15:42:52 +02:00 |
|
|
Grégory Châtel
|
c88d4656a1
|
Cleaning the data encoding code.
|
2018-07-05 14:46:18 +02:00 |
|
|
Grégory Châtel
|
c149818fba
|
Cleaner version of the encoding code.
|
2018-07-05 14:32:00 +02:00 |
|
|
Grégory Châtel
|
a1c3b384ac
|
Updating refenrences from model_py.py to model_pytorch.py.
|
2018-07-05 11:09:38 +02:00 |
|
|
Grégory Châtel
|
615ac6c05d
|
Updating README.md with new module and class names.
|
2018-07-05 11:07:04 +02:00 |
|
|
Thomas Wolf
|
ec0c86da2e
|
Merge pull request #15 from rodgzilla/master
Fixing issue #13 and #14.
|
2018-07-04 13:57:01 +02:00 |
|
|
Grégory Châtel
|
be407cdd37
|
Solving missing variable issue.
|
2018-07-04 13:50:19 +02:00 |
|
|
Grégory Châtel
|
2b7e97e307
|
Creating the model no matter what the number of GPU is.
|
2018-07-04 13:49:50 +02:00 |
|
|
thomwolf
|
37e77aff19
|
small fix cpu only version
|
2018-06-28 14:47:01 +02:00 |
|
|
thomwolf
|
03c9e74b1a
|
clean up multi gpu logic
|
2018-06-28 13:41:15 +02:00 |
|
|
thomwolf
|
aded2b058a
|
Clean up multi-processing logic -- Switch to PyTorch 0.4 style
|
2018-06-28 10:43:15 +02:00 |
|
|
thomwolf
|
49ff9b5355
|
Merge pull request #10 from nottombrown/multi-gpu
[WIP] Add multi-gpu support
|
2018-06-27 21:11:20 +02:00 |
|
|
thomwolf
|
6a40c7d296
|
Merge pull request #9 from nottombrown/master
Remove globals to make code easier to follow
|
2018-06-27 09:03:34 +02:00 |
|
|
nottombrown
|
db2f819d2a
|
Complete todo in readme \o/
|
2018-06-26 20:31:47 -07:00 |
|
|
nottombrown
|
09d28722a7
|
Multi-GPU fine-tuning works correctly
|
2018-06-26 20:23:56 -07:00 |
|
|
nottombrown
|
3906aa6801
|
rm setup.py
|
2018-06-26 19:51:48 -07:00 |
|
|
nottombrown
|
a2820f3970
|
rm extra README.md
|
2018-06-26 19:51:02 -07:00 |
|