From 8fb6f40e160f716ada3c8cdfdc2328abf4097c9e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C4=B0lker=20SI=C4=9EIRCI?= <sigirci.ilker@gmail.com>
Date: Fri, 9 Aug 2024 13:39:33 +0300
Subject: [PATCH] score all endpoint returns lists instead of one element

---
 notebooks/playground.ipynb                    | 68 +++++++++++++++++++
 .../routers/score_calculation.py              | 11 ++-
 2 files changed, 76 insertions(+), 3 deletions(-)

diff --git a/notebooks/playground.ipynb b/notebooks/playground.ipynb
index d316332..a4b0081 100644
--- a/notebooks/playground.ipynb
+++ b/notebooks/playground.ipynb
@@ -22,6 +22,13 @@
     "load_dotenv()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Downloaded Model"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 2,
@@ -77,6 +84,67 @@
     "\n",
     "scorer = BERTScorer(model_type=model_name, lang=\"tr\", rescale_with_baseline=True)"
    ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Test the API"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import httpx\n",
+    "\n",
+    "\n",
+    "def get_bert_score(candidates: list[str], references: list[str]) -> dict:\n",
+    "    url = \"http://localhost:8888/score_calculation/all\"\n",
+    "\n",
+    "    data = {\n",
+    "        \"candidate\": candidates,\n",
+    "        \"reference\": references,\n",
+    "    }\n",
+    "\n",
+    "    response = httpx.post(url, json=data)\n",
+    "\n",
+    "    return response.json()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'precision': [0.8399680852890015, 0.5356022715568542],\n",
+       " 'recall': [0.8359646797180176, 0.5197705626487732],\n",
+       " 'f1': [0.8379616141319275, 0.5275676846504211]}"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "candidates = [\"The quick brown fox jumps over the lazy dog\", \"This is a sample test\"]\n",
+    "references = [\"A fast brown fox leaps over a lazy dog\", \"For example purposes\"]\n",
+    "\n",
+    "get_bert_score(candidates, references)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {
diff --git a/src/bert_score_api/routers/score_calculation.py b/src/bert_score_api/routers/score_calculation.py
index f14e045..00a6114 100644
--- a/src/bert_score_api/routers/score_calculation.py
+++ b/src/bert_score_api/routers/score_calculation.py
@@ -1,3 +1,4 @@
+import logging
 from typing import Annotated
 
 from fastapi import APIRouter, Depends, HTTPException
@@ -5,6 +6,7 @@
 from bert_score_api.deps import get_bert_scorer
 from bert_score_api.schemes import TextPair
 
+logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/score_calculation", tags=["calculate"])
 
 
@@ -24,10 +26,13 @@ async def calculate_bert_score(
             text_pair.reference,
             verbose=True,
         )
+
+        logger.debug(f"Precision type: {type(P)}")
+
         return {
-            "precision": P.mean().item(),
-            "recall": R.mean().item(),
-            "f1": F1.mean().item(),
+            "precision": P.tolist(),
+            "recall": R.tolist(),
+            "f1": F1.tolist(),
         }
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e)) from e