Spaces:
Sleeping
Sleeping
Commit
·
1f48fed
1
Parent(s):
2b64a07
update: retrieved chunks data structure
Browse files
medrag_multi_modal/retrieval/bm25s_retrieval.py
CHANGED
@@ -175,7 +175,7 @@ class BM25sRetriever(weave.Model):
|
|
175 |
results.documents.flatten().tolist(),
|
176 |
results.scores.flatten().tolist(),
|
177 |
):
|
178 |
-
retrieved_chunks.append({
|
179 |
return retrieved_chunks
|
180 |
|
181 |
@weave.op()
|
|
|
175 |
results.documents.flatten().tolist(),
|
176 |
results.scores.flatten().tolist(),
|
177 |
):
|
178 |
+
retrieved_chunks.append({**chunk, **{"score": score}})
|
179 |
return retrieved_chunks
|
180 |
|
181 |
@weave.op()
|
medrag_multi_modal/retrieval/contriever_retrieval.py
CHANGED
@@ -192,8 +192,8 @@ class ContrieverRetriever(weave.Model):
|
|
192 |
for score in scores:
|
193 |
retrieved_chunks.append(
|
194 |
{
|
195 |
-
|
196 |
-
"score": score["item"],
|
197 |
}
|
198 |
)
|
199 |
return retrieved_chunks
|
|
|
192 |
for score in scores:
|
193 |
retrieved_chunks.append(
|
194 |
{
|
195 |
+
**self._chunk_dataset[score["original_index"]],
|
196 |
+
**{"score": score["item"]},
|
197 |
}
|
198 |
)
|
199 |
return retrieved_chunks
|
medrag_multi_modal/retrieval/medcpt_retrieval.py
CHANGED
@@ -231,8 +231,8 @@ class MedCPTRetriever(weave.Model):
|
|
231 |
for score in scores:
|
232 |
retrieved_chunks.append(
|
233 |
{
|
234 |
-
|
235 |
-
"score": score["item"],
|
236 |
}
|
237 |
)
|
238 |
return retrieved_chunks
|
|
|
231 |
for score in scores:
|
232 |
retrieved_chunks.append(
|
233 |
{
|
234 |
+
**self._chunk_dataset[score["original_index"]],
|
235 |
+
**{"score": score["item"]},
|
236 |
}
|
237 |
)
|
238 |
return retrieved_chunks
|
medrag_multi_modal/retrieval/nv_embed_2.py
CHANGED
@@ -217,8 +217,8 @@ class NVEmbed2Retriever(weave.Model):
|
|
217 |
for score in scores:
|
218 |
retrieved_chunks.append(
|
219 |
{
|
220 |
-
|
221 |
-
"score": score["item"],
|
222 |
}
|
223 |
)
|
224 |
return retrieved_chunks
|
|
|
217 |
for score in scores:
|
218 |
retrieved_chunks.append(
|
219 |
{
|
220 |
+
**self._chunk_dataset[score["original_index"]],
|
221 |
+
**{"score": score["item"]},
|
222 |
}
|
223 |
)
|
224 |
return retrieved_chunks
|