explain ag_news

data/explanations/.gitignore

@@ -2,3 +2,4 @@
 /wiki_pl
 /20_news
 /poleval
+/ag_news

dvc.lock

@@ -940,16 +940,16 @@ stages:
     cmd: PYTHONPATH=. python experiments/scripts/tag_dataset.py --dataset_name ag_news
     deps:
     - path: data/datasets/ag_news/
-      md5: c67ae5399a7bbbe3e25d3864b43c7b54.dir
-      size: 33946011
+      md5: 98d7192236c2f868e52f772e9a3aa01b.dir
+      size: 34020031
       nfiles: 3
     - path: experiments/scripts/tag_dataset.py
       md5: 19015179757440a5639ee263dcabfde3
       size: 5010
     outs:
     - path: data/preprocessed/ag_news/
-      md5: f87f6ac5c43e28205abec35a3f18f0e7.dir
-      size: 49924819
+      md5: 225096ec90dfb5085b6a8e3835681155.dir
+      size: 49998839
       nfiles: 3
   reduce@ag_news:
     cmd: PYTHONPATH=. python experiments/scripts/token_reduce.py --dataset_name ag_news
@@ -960,16 +960,16 @@ stages:
       size: 438958577
       nfiles: 6
     - path: data/preprocessed/ag_news/
-      md5: f87f6ac5c43e28205abec35a3f18f0e7.dir
-      size: 49924819
+      md5: 225096ec90dfb5085b6a8e3835681155.dir
+      size: 49998839
       nfiles: 3
     - path: experiments/scripts/token_reduce.py
       md5: 091c622fce2349978e61b8f6e4d22f27
       size: 4945
     outs:
     - path: data/reduced/ag_news
-      md5: 5ab65584e3b005a19cb527fd69dfaf2c.dir
-      size: 11648332
+      md5: adac9c05622da5a0895de488a2dc80fa.dir
+      size: 11715622
       nfiles: 1
   classify@ag_news:
     cmd: PYTHONPATH=. python experiments/scripts/classify.py --dataset_name ag_news
@@ -991,3 +991,23 @@ stages:
       md5: d9cabd3c3554172bf43b296079eec947.dir
       size: 11806224
       nfiles: 2
+  explain@ag_news:
+    cmd: PYTHONPATH=. python experiments/scripts/explain.py --dataset_name ag_news
+      --output_dir data/explanations/ag_news
+    deps:
+    - path: data/models/ag_news
+      md5: af9183a3115fd5f07ec2ba8e9086200b.dir
+      size: 438958577
+      nfiles: 6
+    - path: data/preprocessed/ag_news
+      md5: 225096ec90dfb5085b6a8e3835681155.dir
+      size: 49998839
+      nfiles: 3
+    - path: experiments/scripts/explain.py
+      md5: 2c1eca09f7cbdc5d93278b2cd27b126c
+      size: 4717
+    outs:
+    - path: data/explanations/ag_news/
+      md5: ae37f6468f0b44e40cf421e2544d2646.dir
+      size: 54861964
+      nfiles: 30405

experiments/scripts/explain.py

@@ -103,12 +103,14 @@ def main(
     os.makedirs(output_dir / "global", exist_ok=True)
     for class_id, class_name in model.config.id2label.items():
         importance_df = get_importance(shap_values[:, :, class_id].mean(0))
+        class_name = class_name.replace("/", "_")
         importance_df.to_json(
             output_dir / "global" / f"{class_name}__importance.json",
         )
 
     # LOCAL IMPORTANCE
     for class_id, class_name in model.config.id2label.items():
+        class_name = class_name.replace("/", "_")
         sub_dir = output_dir / "local" / "adversarial" /class_name
         os.makedirs(sub_dir, exist_ok=True)
         for shap_id, text_id in enumerate(test["id"]):