wassname/pytorch-ts
1
0
Fork 0
mirror of https://github.com/wassname/pytorch-ts.git synced 2026-08-04 13:13:57 +08:00
Code Issues Packages Projects Releases Wiki Activity
337 Commits 1 Branch 0 Tags
66c81482b64b7324cfe69665d8fe1b62bbe4eef2
Commit Graph
71 Commits
Author SHA1 Message Date
Kashif Rasul 1d341ac515 fix NB test 2020-07-06 13:20:27 +02:00
Kashif Rasul f0d64126c6 remove initialization 2020-07-04 01:10:22 +02:00
Kashif Rasul ead4609f91 use logits for neg. bin. 2020-07-04 01:08:46 +02:00
Dr. Kashif Rasul bab3716819 formatting 2020-06-18 22:13:18 +02:00
Ingmar SchusterandKashif Rasul 5a06d3406f First go at IndependentDistributionOutput (#16) 
* First go at IndependentDistributionOutput, subclassed by NormalOutput and NegativeBinomialOutput for now

* Multivariate test for new implementation of NormalOutput

* adding scaling parameter to NormalOutput

* IndependentNormalOutput now is an alias of NormalOutput with a DeprecatedWarning. Some more univariate distributions now inherit from IndependentDistributionOutput

* IndependentNormalOutput now is an alias of NormalOutput with a DeprecatedWarning. Some more univariate distributions now inherit from IndependentDistributionOutput
 2020-06-18 17:08:44 +02:00
Dr. Kashif Rasul 9ac375a89d log_abs_det doesnt change int he forward 2020-06-04 10:34:58 +02:00
Dr. Kashif Rasul 6efcd6c5b9 upstream fixes to evaluator and NB output 2020-06-03 11:01:32 +02:00
Dr. Kashif Rasul e5f66530b1 fix serialization 2020-05-22 15:15:05 +02:00
Dr. Kashif Rasul 2f65617be0 fix name clash 2020-05-15 13:57:16 +02:00
Dr. Kashif Rasul cd5b061b9b fix scaling of mixtures 2020-05-15 13:21:52 +02:00
Dr. Kashif Rasul af0b129b21 fixed typos 2020-05-15 13:12:44 +02:00
Kashif RasulandGitHub 085ee835a5 Mixture output (#13) 
* get dataset returns shuffled training data by default

* added Gaussian mixture output distribution

* added StudentTMixtureOutput
 2020-05-15 13:04:45 +02:00
Dr. Kashif Rasul 86e6828091 NB multiply 2nd moment by scale 2020-05-07 16:20:14 +02:00
Kashif RasulandGitHub Enterprise c5fac32bb2 initial lstnet multivariate point forecasting model (#9) 
* initial lstnet

* lstnet network

* fixed forward

* fix splitter

* fix prediction

* rename argument to what it is i.e. time_first

* fixed scaling and some default values

* scaler can now take time_first=False tensors
 2020-04-17 17:13:44 +02:00
Dr. Kashif Rasul 2d8f6d31f0 revert back change and added comments 2020-04-07 11:05:17 +02:00
Dr. Kashif Rasul 5b57396cea scale alpha so that sigma scales by scale as well 2020-04-07 09:41:22 +02:00
Dr. Kashif Rasul f615be41f0 Revert "scale alpha so that the var stays the same" 
This reverts commit 3c6bbf28d7.
 2020-04-06 12:09:38 +02:00
Dr. Kashif Rasul 3c6bbf28d7 scale alpha so that the var stays the same 2020-04-06 11:38:35 +02:00
Dr. Kashif Rasul 192f9d3991 fix typo 2020-03-31 11:04:56 +02:00
Dr. Kashif Rasul cb012405bf optimized the imports 2020-03-30 13:01:08 +02:00
Dr. Kashif Rasul 47063478b7 scale alpha by 1/scale so that var scales by scale factor 2020-03-27 11:34:12 +01:00
Dr. Kashif Rasul 55644c4671 added NormalOutput 2020-03-25 13:34:08 +01:00
Dr. Kashif Rasul 6bc281e526 fix scaling of alpha 2020-03-21 17:47:35 +01:00
Vahe HakobyanandKashif Rasul 1b123ef152 Model serialization (#6) 
* wip: serialization ran successfully

* wip: deserialization ran successfully
 2020-03-12 11:32:46 +01:00
Kashif Rasul 6bc8de7edb call __init__ before registering buffers 2020-03-05 15:37:45 +01:00
Kashif RasulandGitHub Enterprise 7a16872e26 fix the final log_abs_det_jacobian when scaling (#3) 
* fix the final log_abs_det_jacobian

* fix argument name

* subclass from Flow class

* move buffer for base dist to Flow class as well

* formatting
 2020-02-26 11:02:16 +01:00
Dr. Kashif Rasul b311aded13 exchange the forward and inverse in Real NVP 2020-01-22 10:48:45 +01:00
Kashif Rasul f09153cf5a fix RealNVP forward and inverse 2020-01-21 22:16:52 +01:00
Dr. Kashif Rasul ea2466f8a0 add scale variable 2020-01-17 13:14:31 +01:00
Dr. Kashif Rasul e472f57c1c fix import 2020-01-17 11:51:45 +01:00
Dr. Kashif Rasul 3b46a293c5 added Masked Linear Layer 2020-01-17 11:47:35 +01:00
Dr. Kashif Rasul 744eb22ec4 added MAF 2020-01-17 11:36:30 +01:00
Dr. Kashif Rasul 6ee817ab72 dont set scale if scaling is false 2020-01-15 15:24:11 +01:00
Dr. Kashif Rasul c1970eae9b revert 2020-01-15 13:24:57 +01:00
Dr. Kashif Rasul 888139203a detach scale tensor just in case and mult. in forward 2020-01-15 13:19:05 +01:00
Kashif RasulandGitHub Enterprise 9938121a42 Test conditioning (#1) 
* explicitly set the conditioning to flow

* try ELUs

* return first element of tuple

* typo

* revert ELU
 2020-01-15 11:39:10 +01:00
Kashif Rasul c429003932 cleanup 2020-01-13 20:31:05 +01:00
Kashif Rasul f8bf093fee typo 2020-01-13 20:15:31 +01:00
Kashif Rasul 5c0e84c2fd fix shape 2020-01-13 20:14:20 +01:00
Kashif Rasul 8cb33f20c6 shape has to be a tuple 2020-01-13 19:52:47 +01:00
Dr. Kashif Rasul ca2918b02e fix comparison 2020-01-13 16:59:07 +01:00
Dr. Kashif Rasul 7cea995d10 use scale 2020-01-13 16:46:52 +01:00
Dr. Kashif Rasul 52fbc77f21 added tempFlow model 2020-01-13 11:28:31 +01:00
Kashif Rasul f186223aaa typos 2020-01-12 14:28:30 +01:00
Kashif Rasul d41fc5d915 intial realnvp output distribution class 2020-01-12 13:46:13 +01:00
Kashif Rasul 5b5a71a358 added conditioning flows 2020-01-12 11:42:03 +01:00
Dr. Kashif Rasul e263bd5f5c fix past_observed_values 2020-01-09 13:46:56 +01:00
Dr. Kashif Rasul d2d570265e MultivariateNormal expects named arguments 2020-01-06 15:18:26 +01:00
Dr. Kashif Rasul 98717c2e8f added multivariate gaussian output 2020-01-06 14:49:40 +01:00
Dr. Kashif Rasul dc1951b19e added independent normal output 2020-01-04 20:47:04 +01:00