From d71bd470a53f65b28565ab8faea3ca9d698f8edc Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:08:38 +0100 Subject: [PATCH 1/6] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index a6024ce..13b1e5e 100644 --- a/README.md +++ b/README.md @@ -201,7 +201,7 @@ To produce the exact numbers below, make sure you have the correct version of th Also, make sure you link your BLAS to OpenBLAS. We did observe different numbers when using different BLAS linkage. Example scripts for generating the config files for the experiments can be found at scp/batch_script/. -benchmark +benchmark 5 example models were saved at model/demo/ for reference. You can directly run testing or interaction on these models. From 782d96e8481f7263367e372723b1b3a8943ab423 Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:09:43 +0100 Subject: [PATCH 2/6] Create README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 13b1e5e..4c777cd 100644 --- a/README.md +++ b/README.md @@ -201,7 +201,7 @@ To produce the exact numbers below, make sure you have the correct version of th Also, make sure you link your BLAS to OpenBLAS. We did observe different numbers when using different BLAS linkage. Example scripts for generating the config files for the experiments can be found at scp/batch_script/. -benchmark +benchmark 5 example models were saved at model/demo/ for reference. You can directly run testing or interaction on these models. From 159cbe4458f8189569c293f9e0d7c1c17ac77f89 Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:11:00 +0100 Subject: [PATCH 3/6] Create README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 4c777cd..1de6b64 100644 --- a/README.md +++ b/README.md @@ -201,7 +201,7 @@ To produce the exact numbers below, make sure you have the correct version of th Also, make sure you link your BLAS to OpenBLAS. We did observe different numbers when using different BLAS linkage. Example scripts for generating the config files for the experiments can be found at scp/batch_script/. -benchmark +benchmark 5 example models were saved at model/demo/ for reference. You can directly run testing or interaction on these models. From 6f81640b1534a6abb4f911a9d08f7dafc60d741c Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:11:22 +0100 Subject: [PATCH 4/6] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 1de6b64..b9a2864 100644 --- a/README.md +++ b/README.md @@ -201,7 +201,7 @@ To produce the exact numbers below, make sure you have the correct version of th Also, make sure you link your BLAS to OpenBLAS. We did observe different numbers when using different BLAS linkage. Example scripts for generating the config files for the experiments can be found at scp/batch_script/. -benchmark +benchmark 5 example models were saved at model/demo/ for reference. You can directly run testing or interaction on these models. From d49104a3bcf95b72d08a99d6d190cf395ad120aa Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:17:38 +0100 Subject: [PATCH 5/6] Create README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index b9a2864..8b9699c 100644 --- a/README.md +++ b/README.md @@ -3,7 +3,7 @@ NNDial is an open source toolkit for building **end-to-end trainable task-oriented dialogue models**. It is released by **Tsung-Hsien (Shawn) Wen** from **Cambridge Dialogue Systems Group** under **Apache License 2.0**. # Requirement -Here are a list of packages in order to run the program with the suggested versions, +In order to run the program, here are a list of packages with the suggested versions, ``` - Theano 0.8.2 - Numpy 1.12.0 From 14b89cb74ca17ac3d4ac7f52960ab152af5dce2e Mon Sep 17 00:00:00 2001 From: Shawn Wen Date: Wed, 14 Jun 2017 11:23:04 +0100 Subject: [PATCH 6/6] Create README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 8b9699c..457b925 100644 --- a/README.md +++ b/README.md @@ -185,7 +185,7 @@ The selection criterion was success+0.5\*BLEU on the validation set. The hyperparameters we were searching are, ``` - Initial learning rate. For NDM we searched over [0.008,0.010,0.012] while for LIDM we searched over [0.002,0.004,0.006]. - The range of LIDM is smaller than NDM simply because NVI requires a smaller learning rate in the beginning for efficient inference. + The range of LIDM is lower than NDM simply because NVI requires a smaller learning rate in the beginning for efficient inference. - L2 regularisation. We searched over [0.0, 1e-4, 1e-5, 1e-6]. - Random seed. We searched over [1, 2, 3, 4, 5]. ```