more dataset

configs/kinship-complex.sh

@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="complex"
+add_reversed_training_edges="True"
+group_examples_by_query="True"
+entity_dim=200
+relation_dim=200
+num_rollouts=1
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=500
+batch_size=512
+train_batch_size=512
+dev_batch_size=64
+learning_rate=0.003
+grad_norm=5
+emb_dropout_rate=0.3
+beam_size=100
+
+num_negative_samples=10
+margin=10

configs/kinship-conf.sh

@@ -0,0 +1,47 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="point.rs.conve"
+group_examples_by_query="False"
+use_action_space_bucketing="True"
+use_conf="True"
+support_times="True"
+path_search_policy="tree"
+decrease_step=15
+decrease_rate=0.9
+decrease_offline=0
+baseline="curriculum"
+
+bandwidth=400
+entity_dim=200
+relation_dim=200
+history_dim=200
+history_num_layers=3
+num_rollouts=20
+num_rollout_steps=2
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=100
+num_peek_epochs=2
+batch_size=128  
+train_batch_size=128
+dev_batch_size=32
+learning_rate=0.001
+baseline="curriculum"
+grad_norm=5
+emb_dropout_rate=0.3
+ff_dropout_rate=0.1
+action_dropout_rate=0.9
+action_dropout_anneal_interval=1000
+reward_shaping_threshold=0
+beta=0.05
+relation_only="False"
+beam_size=128
+
+distmult_state_dict_path="model/kinship-distmult-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+complex_state_dict_path="model/kinship-complex-RV-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+conve_state_dict_path="model/kinship-conve-RV-xavier-200-200-0.003-32-3-0.2-0.3-0.2-0.1/model_best.tar"
+checkpoint_path="model/kinship-point.rs.conve-xavier-n/a-200-200-3-0.001-0.3-0.1-0.9-400-0.05-useConf/model_best.tar"
+
+num_paths_per_entity=-1
+margin=-1

configs/kinship-conve.sh

@@ -0,0 +1,22 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="conve"
+add_reversed_training_edges="True"
+group_examples_by_query="True"
+entity_dim=200
+relation_dim=200
+num_rollouts=1
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=500
+batch_size=512
+train_batch_size=512
+dev_batch_size=64
+learning_rate=0.003
+grad_norm=0
+emb_dropout_rate=0.2
+beam_size=128
+
+num_negative_samples=20
+margin=0.5

configs/kinship-distmult.sh

@@ -0,0 +1,20 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="distmult"
+group_examples_by_query="True"
+entity_dim=200
+relation_dim=200
+num_rollouts=1
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=200
+batch_size=512
+train_batch_size=512
+dev_batch_size=64
+learning_rate=0.003
+grad_norm=5
+emb_dropout_rate=0.3
+beam_size=100
+num_negative_samples=10
+margin=10

configs/kinship-rs.sh

@@ -0,0 +1,44 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="point.rs.conve"
+group_examples_by_query="False"
+use_action_space_bucketing="True"
+support_times="True"
+path_search_policy="tree"
+decrease_step=0
+decrease_rate=0
+
+bandwidth=400
+entity_dim=200
+relation_dim=200
+history_dim=200
+history_num_layers=3
+num_rollouts=20
+num_rollout_steps=2
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=400
+num_peek_epochs=2
+batch_size=128  
+train_batch_size=128
+dev_batch_size=32
+learning_rate=0.001
+baseline="n/a"
+grad_norm=5
+emb_dropout_rate=0.3
+ff_dropout_rate=0.1
+action_dropout_rate=0.9
+action_dropout_anneal_interval=1000
+reward_shaping_threshold=0
+beta=0.05
+relation_only="False"
+beam_size=128
+decrease_offline=0
+distmult_state_dict_path="model/kinship-distmult-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+complex_state_dict_path="model/kinship-complex-RV-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+conve_state_dict_path="model/kinship-conve-RV-xavier-200-200-0.003-32-3-0.2-0.3-0.2-0.1/model_best.tar"
+checkpoint_path="model/kinship-point.rs.conve-xavier-n/a-200-200-3-0.001-0.3-0.1-0.9-400-0.05/model_best.tar"
+
+num_paths_per_entity=-1
+margin=-1

configs/kinship.sh

@@ -0,0 +1,34 @@
+#!/usr/bin/env bash
+
+data_dir="data/kinship"
+model="point"
+group_examples_by_query="False"
+use_action_space_bucketing="True"
+
+bandwidth=400
+entity_dim=200
+relation_dim=200
+history_dim=200
+history_num_layers=3
+num_rollouts=20
+num_rollout_steps=2
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=400
+num_peek_epochs=5
+batch_size=128
+train_batch_size=128
+dev_batch_size=64
+learning_rate=0.001
+baseline="n/a"
+grad_norm=5
+emb_dropout_rate=0.3
+ff_dropout_rate=0.1
+action_dropout_rate=0.9
+action_dropout_anneal_interval=1000
+beta=0.05
+relation_only="False"
+beam_size=128
+
+num_paths_per_entity=-1
+margin=-1

configs/umls-complex.sh

@@ -0,0 +1,21 @@
+#!/usr/bin/env bash
+
+data_dir="data/umls"
+model="complex"
+add_reversed_training_examples="False"
+group_examples_by_query="True"
+entity_dim=200
+relation_dim=200
+num_rollouts=1
+bucket_interval=10
+num_epochs=2000
+num_wait_epochs=500
+batch_size=64
+train_batch_size=64
+dev_batch_size=64
+learning_rate=0.003
+grad_norm=5
+emb_dropout_rate=0.3
+num_negative_samples=20
+margin=10
+beam_size=137

configs/umls-conf.sh

@@ -0,0 +1,47 @@
+#!/usr/bin/env bash
+
+data_dir="data/umls"
+model="point.rs.conve"
+group_examples_by_query="False"
+use_action_space_bucketing="True"
+use_conf="True"
+relation_only="False"
+support_times="True"
+path_search_policy="tree"
+decrease_step=10
+decrease_rate=0.9
+decrease_offline=0
+baseline="curriculum"
+
+bandwidth=400
+entity_dim=200
+relation_dim=200
+history_dim=200
+history_num_layers=3
+num_rollouts=20
+num_rollout_steps=2
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=100
+num_peek_epochs=2
+batch_size=128
+train_batch_size=128
+dev_batch_size=64
+learning_rate=0.001
+grad_norm=5
+emb_dropout_rate=0.1
+ff_dropout_rate=0
+action_dropout_rate=0.95
+action_dropout_anneal_interval=1000
+reward_shaping_threshold=0
+beta=0.05
+beam_size=128
+
+# pre-trained embedding based model checkpoints
+distmult_state_dict_path="model/umls-distmult-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+complex_state_dict_path="model/umls-complex-xavier-200-200-0.003-0.3-0.1/model_best.tar"
+conve_state_dict_path="model/umls-conve-RV-xavier-200-200-0.003-32-3-0.3-0.3-0.2-0.1/model_best.tar"
+checkpoint_path="model/umls-point.rs.conve-xavier-curriculum-200-200-3-0.001-0.1-0.0-0.95-400-0.05-20.0-0.81-0.0-useConf/model_best.tar"
+
+num_paths_per_entity=-1
+margin=-1

configs/umls-conve.sh

@@ -0,0 +1,21 @@
+#!/usr/bin/env bash
+
+data_dir="data/umls"
+model="conve"
+group_examples_by_query="True"
+add_reversed_training_edges="True"
+entity_dim=200
+relation_dim=200
+num_rollouts=1
+bucket_interval=10
+num_epochs=1000
+num_wait_epochs=500
+batch_size=512
+train_batch_size=512
+dev_batch_size=64
+learning_rate=0.003
+grad_norm=0
+emb_dropout_rate=0.3
+num_negative_samples=20
+margin=0.05
+beam_size=137