update readme&bash files

ImKeTT · May 19, 2022 · e85f453 · e85f453
1 parent 8b2e266
commit e85f453
Show file tree

Hide file tree

Showing 11 changed files with 833 additions and 480 deletions.
diff --git a/.DS_Store b/.DS_Store
diff --git a/.idea/bert_adapter.iml b/.idea/bert_adapter.iml
diff --git a/.idea/deployment.xml b/.idea/deployment.xml
diff --git a/.idea/inspectionProfiles/profiles_settings.xml b/.idea/inspectionProfiles/profiles_settings.xml
diff --git a/.idea/misc.xml b/.idea/misc.xml
diff --git a/.idea/workspace.xml b/.idea/workspace.xml
diff --git a/README.md b/README.md
@@ -1,4 +1,5 @@
 ## AdaVAE: Exploring Adaptive GPT-2s in VAEs for Language Modeling
 
-**[Repo In progress]** Official implementation for **AdaVAE**, check the paper on arxiv [https://arxiv.org/pdf/2205.05862.pdf](https://arxiv.org/pdf/2205.05862.pdf).
+**[Repo In Progress]** Official implementation for **AdaVAE**, check the paper on arxiv [https://arxiv.org/pdf/2205.05862.pdf](https://arxiv.org/pdf/2205.05862.pdf).
 
+*This repo takes some practices from [https://github.com/fangleai/TransformerCVAE](https://github.com/fangleai/TransformerCVAE)*. Many thanks !!
diff --git a/low_nlu/.DS_Store b/low_nlu/.DS_Store
diff --git a/src/.DS_Store b/src/.DS_Store
diff --git a/src/run0.sh b/src/run0.sh
@@ -1,4 +1,3 @@
-python adaVAE.py --batch-sizes 100 --dataset yelp_data --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 15000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --kl_rate 0.50 &&\
-python adaVAE.py --batch-sizes 100 --dataset yahoo_data --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --latent_gen linear --kl_rate 0.50 &&\
-python adaVAE.py --batch-sizes 100 --dataset yahoo_data --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --latent_gen linear --add_mem --kl_rate 0.50 &&\
-python adaVAE.py --batch-sizes 100 --dataset yahoo_data --max_length 32 --pre_enc_iter start --add_mem --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --latent_gen linear --kl_rate 0.50
+python adaVAE.py --batch-sizes 90 --dataset yelp_data --max_length 32 --pre_enc_iter start --add_attn --add_z2adapters --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none  --kl_rate 0.50 &&\
+python adaVAE.py --batch-sizes 90 --dataset yelp_data --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --latent_gen  mean_max_linear --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none  --kl_rate 0.50 &&\
+python adaVAE.py --batch-sizes 90 --dataset yelp_data --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --latent_gen linear --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none  --kl_rate 0.50
diff --git a/src/run1.sh b/src/run1.sh
@@ -1,2 +1 @@
-python adaVAE.py --batch-sizes 100 --dataset cola --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 15000 --weighted_sample --latent_size 768 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --kl_rate 0.10 &&\
-python adaVAE.py --batch-sizes 100 --dataset sst-2 --max_length 32 --pre_enc_iter start --add_attn --beta_0 1 --fb 1 --adapter_size 128 --iterations 15000 --weighted_sample --latent_size 768 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none --kl_rate 0.10
+python adaVAE.py --batch-sizes 90 --dataset yelp_data --max_length 32 --pre_enc_iter start --add_attn --add_z2adapters --beta_0 1 --fb 1 --adapter_size 128 --iterations 22000 --weighted_sample --latent_size 32 --encoder_n_layer 8 --decoder_n_layer 12 --adapter_init bert --attn_mode none  --kl_rate 0.50