Issue with LLM Reranker Output Evaluation: InvalidIndexError
Hello team, I have two rerankers, a BGE M3 and a LLM reranker before passing my documents to the LLM to generate an answer. I am interested in evaluating the output documents from the second reranker (LLM reranker) but I'm running into the following issue:
InvalidIndexError Traceback (most recent call last)
Cell In[3], line 3
1 from phoenix.trace.dsl import SpanQuery
----> 3 evaluation_df = pd.concat(
4 px.Client().query_spans(
5 SpanQuery().select(input="input.value", output="output.value").where("parent_id is None"),
6 SpanQuery()
7 .where("span_kind == 'RERANKER'")
8 .select(span_id="parent_id")
9 .concat(
10 "reranker.output_documents",
11 reference="document.content",
12 ),
13 ),
14 axis=1,
15 join="inner",
16 )
18 evaluation_df
File ~/Documents/DEV_ENV/venv/lib/python3.10/site-packages/pandas/core/reshape/concat.py:393, in concat(objs, axis, join, ignore_index, keys, levels, names, verify_integrity, sort, copy)
378 copy = False
380 op = _Concatenator(
381 objs,
382 axis=axis,
...
-> 3882 raise InvalidIndexError(self._requires_unique_msg)
3884 if len(target) == 0:
3885 return np.array([], dtype=np.intp)
InvalidIndexError: Reindexing only valid with uniquely valued Index objectsAny idea what is going on? Seems related to the fact that I have two reranking activities
