added grid search to pipeline

tira-io · Jul 1, 2024 · 40ee264 · 40ee264
1 parent 25071bd
commit 40ee264
Showing 1 changed file with 35 additions and 7 deletions.
diff --git a/baseline-retrieval-system/tira-retrieval-system.ipynb b/baseline-retrieval-system/tira-retrieval-system.ipynb
@@ -94,9 +94,40 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 14,
    "metadata": {},
-   "outputs": [],
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "ir-lab-sose-2024/ir-acl-anthology-20240504-training documents:   0%|          | 0/126958 [00:00<?, ?it/s]/tmp/ipykernel_1917/3762444819.py:15: DeprecationWarning: specifying meta and meta_lengths in IterDictIndexer.index() is deprecated, use constructor instead\n",
+      "  index_ref = indexer.index(pt_dataset.get_corpus_iter(),\n",
+      "ir-lab-sose-2024/ir-acl-anthology-20240504-training documents:  71%|███████   | 90188/126958 [00:20<00:05, 6715.11it/s]"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "10:52:09.990 [ForkJoinPool-1-worker-3] WARN org.terrier.structures.indexing.Indexer - Adding an empty document to the index (2020.mir_conference-2020.1) - further warnings are suppressed\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "ir-lab-sose-2024/ir-acl-anthology-20240504-training documents: 100%|██████████| 126958/126958 [00:25<00:00, 4903.30it/s] \n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "10:52:22.976 [ForkJoinPool-1-worker-3] WARN org.terrier.structures.indexing.Indexer - Indexed 3 empty documents\n"
+     ]
+    }
+   ],
    "source": [
     "# The dataset: the union of the IR Anthology and the ACL Anthology\n",
     "# This line creates an IRDSDataset object and registers it under the name provided as an argument.\n",
@@ -198,8 +229,8 @@
     "df_sorted = df.sort_values('nDCG@10', ascending=False)\n",
     "print(df_sorted)\n",
     "best_configuration = df_sorted['run'].values[0]\n",
-    "best_b = 0.1\n",
-    "best_k_1 = 1.9"
+    "best_b = best_configuration.split(\"-\")[1].split(\"=\")[1]\n",
+    "best_k_1 = best_configuration.split(\"-\")[2].split(\"=\")[1]"
    ]
   },
   {
@@ -208,9 +239,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "best_b = 0.1\n",
-    "best_k_1 = 1.9\n",
-    "\n",
     "bm25 = pt.BatchRetrieve(index, wmodel=\"BM25\", controls= {\"bm25.b\" : best_b, \"bm25.k_1\": best_k_1})\n",
     "sdm = pt.rewrite.SDM()"
    ]