Remove header from output

.vscode/launch.json

@@ -4,6 +4,7 @@
     {
     "name": "Python: Attach",
     "type": "python",
+    "justMyCode": false,
     "request": "attach",
     "host": "localhost",
     "port": 5678,

docker/docker-compose.dev.yaml

@@ -15,32 +15,16 @@ services:
       - /media:/media
       - /tmp:/tmp
       - /mnt/data/model:/model
-    depends_on:
-      - elastic
+      - /mnt:/mnt
+    entrypoint:
+      - "/bin/sh"
+      - -ecx
+      - |
+        python -m debugpy --wait-for-client --listen 0.0.0.0:5678 \
+        /app/src/app.py retrieval --topic-nrb 2 \
+        --reranking word-mover-distance --reuse-unranked /mnt/data/initial_retrieval_dump.pkl test \
+        /media/training-datasets/"touche-task-1"/touche-task-1-2022-02-16 /tmp/test
 
 
-  elastic:
-    image: "docker.elastic.co/elasticsearch/elasticsearch:7.15.2"
-    restart: always
-    networks:
-      - tira
-    ports:
-      - "9200:9200"
-      - "9300:9300"
-    volumes:
-      - /mnt/data/elastic:/usr/share/elasticsearch/data
-      - ./conifg:/conifg
-    environment:
-      - discovery.type=single-node
-      - logger.level=DEBUG
-    healthcheck:
-      test:
-        [
-          "CMD",
-          "curl",
-          "-s",
-          "-f",
-          "http://localhost:9200/_cat/health"
-        ]
 networks:
   tira: null

python/src/retrieval/reranking.py

@@ -124,12 +124,15 @@ class Reranking:
                     stance_premise = premise['_source']['sentence_stance']
                     if stance_conclusion == stance_premise \
                             and premise['_id'] not in encountered_premise_ids:
+                        
+                        final_key = "final_reranking_score" if "final_reranking_score" in conclusion else "_score"
+
                         trec_style_rows.append(
                             (topic_nrb, conclusion['_source']['sentence_stance'],
                              f'{"__".join(premise["_id"].split("_"))},' +
                              f'{"__".join(conclusion["_id"].split("_"))}',
                              len(trec_style_rows),
-                             f'{conclusion["final_reranking_score"]:.2f}', self.run_name))
+                             f'{conclusion[final_key]:.2f}', self.run_name))
                         encountered_premise_ids.add(premise['_id'])
                         break
                 rows_per_topic[topic_nrb] = trec_style_rows

python/src/utils/parse_cli_args.py

@@ -80,11 +80,11 @@ def parse_cli_args() -> argparse.Namespace:
     parser_retrieval.add_argument('--lambda-conclusions',
                                   type=float,
                                   required=False,
-                                  default=0.5)
+                                  default=0.75)
     parser_retrieval.add_argument('--lambda-premises',
                                   type=float,
                                   required=False,
-                                  default=0.5)
+                                  default=0.75)
     parser_retrieval.add_argument('--wait-for-es',
                                 action='store_true')
 

python/src/utils/write_trec_file.py

@@ -10,7 +10,6 @@ def write_terc_file(result: list, output_path: Path) -> None:
     with open(str(output_path), 'w') as csvfile:
         fieldnames = ['qid', 'stance', 'pair', 'rank', 'score', 'tag']
         writer = DictWriter(csvfile, delimiter=' ', fieldnames=fieldnames)
-        writer.writeheader()
         for row in result:
             writer.writerow(dict(zip(fieldnames, row)))
             csvfile.flush()