train disease-cell,gene,process,phenotype synthetic

UCDenver-ccp · Aug 29, 2024 · 251bb4a · 251bb4a
1 parent 396b7ed
commit 251bb4a
Show file tree

Hide file tree

Showing 15 changed files with 42,041 additions and 12,004 deletions.
diff --git a/.github/workflows/bert-models-cicd.yaml b/.github/workflows/bert-models-cicd.yaml
diff --git a/IMAGE_VERSIONS_PREDICT b/IMAGE_VERSIONS_PREDICT
@@ -15,7 +15,7 @@ BL_PROCESS_TO_ANATOMY=0.1
 BL_PROCESS_TO_COMPONENT_CELL=0.1
 BL_PROCESS_TO_CELL=0.1
 BL_DISEASE_TO_ANATOMY=0.3
-BL_DISEASE_TO_CELL=0.2
-BL_DISEASE_TO_GENE=0.1
-BL_DISEASE_TO_PHENOTYPE=0.1
-BL_DISEASE_TO_PROCESS=0.3
+BL_DISEASE_TO_CELL=0.3
+BL_DISEASE_TO_GENE=0.2
+BL_DISEASE_TO_PHENOTYPE=0.2
+BL_DISEASE_TO_PROCESS=0.4
diff --git a/MODEL_VERSIONS b/MODEL_VERSIONS
@@ -24,7 +24,7 @@ BL_PROCESS_TO_ANATOMY=0.1
 BL_PROCESS_TO_COMPONENT_CELL=0.1
 BL_PROCESS_TO_CELL=0.1
 BL_DISEASE_TO_ANATOMY=0.3
-BL_DISEASE_TO_CELL=0.2
-BL_DISEASE_TO_GENE=0.1
-BL_DISEASE_TO_PHENOTYPE=0.1
-BL_DISEASE_TO_PROCESS=0.3
+BL_DISEASE_TO_CELL=0.3
+BL_DISEASE_TO_GENE=0.2
+BL_DISEASE_TO_PHENOTYPE=0.2
+BL_DISEASE_TO_PROCESS=0.4
diff --git a/data/bl_disease_to_cell/CHANGELOG.txt b/data/bl_disease_to_cell/CHANGELOG.txt
@@ -1,3 +1,7 @@
+v0.3
+- trained on synthetic sentences created by GPT (disease-has_basis_in-cell.synthetic.bert.v1.random.tsv)
+- predicate is now has_basis_in
+
 v0.2
 - v0.1 had terrible performance; re-trained and observed expected performance
 

diff --git a/data/bl_disease_to_cell/data.tsv b/data/bl_disease_to_cell/data.tsv
diff --git a/data/bl_disease_to_cell/disease-has_basis_in-cell.synthetic.bert.v1.random.tsv b/data/bl_disease_to_cell/disease-has_basis_in-cell.synthetic.bert.v1.random.tsv
diff --git a/data/bl_disease_to_gene/CHANGELOG.txt b/data/bl_disease_to_gene/CHANGELOG.txt
@@ -1,2 +1,6 @@
+v0.2
+- trained on synthetic sentences created by GPT (disease-has_basis_in-gene.synthetic.bert.v1.random.tsv)
+- predicate is now has_basis_in
+
 v0.1
 - trained on synthetic sentences created by GPT (assoc-with-disease.gene.synthetic.bert.v1.tsv)
diff --git a/data/bl_disease_to_gene/data.tsv b/data/bl_disease_to_gene/data.tsv
diff --git a/data/bl_disease_to_gene/disease-has_basis_in-gene.synthetic.bert.v1.random.tsv b/data/bl_disease_to_gene/disease-has_basis_in-gene.synthetic.bert.v1.random.tsv
diff --git a/data/bl_disease_to_phenotype/CHANGELOG.txt b/data/bl_disease_to_phenotype/CHANGELOG.txt
@@ -1,2 +1,6 @@
+v0.2
+- trained on synthetic sentences created by GPT (disease-has_phenotype-phenotype.synthetic.bert.v1.random.tsv)
+- predicate is now has_phenotype
+
 v0.1
 - trained on synthetic sentences created by GPT (assoc-with-disease.phenotype.synthetic.bert.v1.random.tsv)
diff --git a/data/bl_disease_to_phenotype/data.tsv b/data/bl_disease_to_phenotype/data.tsv
diff --git a/data/bl_disease_to_phenotype/disease-has_phenotype-phenotype.synthetic.bert.v1.random.tsv b/data/bl_disease_to_phenotype/disease-has_phenotype-phenotype.synthetic.bert.v1.random.tsv
diff --git a/data/bl_disease_to_process/CHANGELOG.txt b/data/bl_disease_to_process/CHANGELOG.txt
@@ -1,3 +1,7 @@
+v0.4
+- trained on synthetic sentences created by GPT (disease-has_basis_in-biological_process.synthetic.bert.v1.random.tsv)
+- predicate is now has_basis_in
+
 v0.3
 - decent performance - no change in the data. Not sure why.
 

diff --git a/data/bl_disease_to_process/data.tsv b/data/bl_disease_to_process/data.tsv
diff --git a/...l_disease_to_process/disease-has_basis_in-biological_process.synthetic.bert.v1.random.tsv b/...l_disease_to_process/disease-has_basis_in-biological_process.synthetic.bert.v1.random.tsv