Upload README.md with huggingface_hub

README.md

@@ -68,7 +68,7 @@ chunks_table = "your_chunks_table"  # columns: chunk_id, chunk_txt
 QUERY = "What is the best-known landmark in Paris?"
 
 query = f"""
-SELECT chunk_id, chunk_txt, emb_1 AS relevance_score
+SELECT chunk_id, chunk_txt, emb_0 AS relevance_score
 FROM mldb.ONNXEmbeddings(
     ON (SELECT chunk_id, chunk_txt, '{QUERY}' || '</s></s>' || chunk_txt AS txt
          FROM {chunks_table}) AS InputTable
@@ -86,7 +86,7 @@ result = tdml.DataFrame.from_query(query).to_pandas().sort_values("relevance_sco
 
 `relevance_score` is the model's raw logit -- sorting by it directly already gives the correct
 ranking, since `sigmoid` is monotonic. If you need an actual `[0, 1]` probability rather than
-just a ranking, compute it in SQL instead of Python: `1 / (1 + EXP(-emb_1))`.
+just a ranking, compute it in SQL instead of Python: `1 / (1 + EXP(-emb_0))`.
 
 ## Combining with vector search for two-stage retrieval
 
@@ -103,9 +103,9 @@ CREATE VOLATILE TABLE shortlist AS (
         ON chunk_embeddings AS ReferenceTable DIMENSION
         USING
             TargetIDColumn('query_id')
-            TargetFeatureColumns('[emb_1:emb_384]')
+            TargetFeatureColumns('[emb_0:emb_383]')
             RefIDColumn('chunk_id')
-            RefFeatureColumns('[emb_1:emb_384]')
+            RefFeatureColumns('[emb_0:emb_383]')
             DistanceMeasure('cosine')
             TopK(20)
     ) AS dt