Added test vocab notebook

mdmmn378 · Feb 2, 2024 · 2cffadb · 2cffadb
1 parent 3c12d44
commit 2cffadb
Show file tree

Hide file tree

Showing 6 changed files with 7,991 additions and 1,759 deletions.
diff --git a/model_artifacts/.gitignore b/model_artifacts/.gitignore
@@ -0,0 +1,3 @@
+*
+!.gitignore
+!.dvc
diff --git a/notebooks/1.dataset-builder.ipynb → notebooks/1. dataset-builder.ipynb b/notebooks/1.dataset-builder.ipynb → notebooks/1. dataset-builder.ipynb
@@ -400,7 +400,7 @@
     "split_config = {\n",
     "    \"train\": \"./datasets/correction_train.jsonl\",\n",
     "    \"test\": \"./datasets/correction_test.jsonl\",\n",
-    "    \"val\": \"./datasets/correction_val.jsonl\"\n",
+    "    \"val\": \"./datasets/correction_val.jsonl\",\n",
     "}"
    ]
   },
@@ -464,7 +464,9 @@
     }
    ],
    "source": [
-    "dataset = dataset.map(lambda x: {\"from\": \"00xx00\"+x[\"from\"], \"to\": \"00xx00\"+x[\"to\"]})"
+    "dataset = dataset.map(\n",
+    "    lambda x: {\"from\": \"00xx00\" + x[\"from\"], \"to\": \"00xx00\" + x[\"to\"]}\n",
+    ")"
    ]
   },
   {