Alexander Sanchez commited on
Commit
aad1d52
·
1 Parent(s): f1d274f

caligraphy metadata added

Browse files
Files changed (3) hide show
  1. corpus_loader.py +1 -0
  2. rag_corrector.py +1 -0
  3. vector_store.py +2 -0
corpus_loader.py CHANGED
@@ -98,6 +98,7 @@ class CorpusLoader:
98
  "type": str(row.get("type", "desconocido")),
99
  "region": str(row.get("region", "desconocida")),
100
  "date": str(row.get("date", "")),
 
101
  "source": source,
102
  "corrections": row.get("corrections", []),
103
  }
 
98
  "type": str(row.get("type", "desconocido")),
99
  "region": str(row.get("region", "desconocida")),
100
  "date": str(row.get("date", "")),
101
+ "caligrafia": str(row.get("caligrafia", "desconocida")),
102
  "source": source,
103
  "corrections": row.get("corrections", []),
104
  }
rag_corrector.py CHANGED
@@ -119,6 +119,7 @@ class RAGCorrector:
119
  corr = "; ".join(doc["corrections"]) if doc["corrections"] else "—"
120
  examples.append(
121
  f"Ejemplo {i} [{doc['type']}, {doc['region']}, {doc['date']}]"
 
122
  f" (similitud={doc['score']}):\n"
123
  f" HTR: \"{doc['htr']}\"\n"
124
  f" GT: \"{doc['gt']}\"\n"
 
119
  corr = "; ".join(doc["corrections"]) if doc["corrections"] else "—"
120
  examples.append(
121
  f"Ejemplo {i} [{doc['type']}, {doc['region']}, {doc['date']}]"
122
+ f"caligrafia: {doc['caligrafia']}"
123
  f" (similitud={doc['score']}):\n"
124
  f" HTR: \"{doc['htr']}\"\n"
125
  f" GT: \"{doc['gt']}\"\n"
vector_store.py CHANGED
@@ -88,6 +88,7 @@ class VectorStore:
88
  "type": p.get("type", ""),
89
  "region": p.get("region", ""),
90
  "date": p.get("date", ""),
 
91
  "corrections": json.dumps(
92
  p.get("corrections", []), ensure_ascii=False
93
  ),
@@ -131,6 +132,7 @@ class VectorStore:
131
  "type": meta.get("type", ""),
132
  "region": meta.get("region", ""),
133
  "date": meta.get("date", ""),
 
134
  "corrections": json.loads(meta.get("corrections", "[]")),
135
  "score": round(1 - dist, 4), # cosine similarity
136
  })
 
88
  "type": p.get("type", ""),
89
  "region": p.get("region", ""),
90
  "date": p.get("date", ""),
91
+ "caligrafia": p.get("caligrafia", "desconocida"),
92
  "corrections": json.dumps(
93
  p.get("corrections", []), ensure_ascii=False
94
  ),
 
132
  "type": meta.get("type", ""),
133
  "region": meta.get("region", ""),
134
  "date": meta.get("date", ""),
135
+ "caligrafia": meta.get("caligrafia", ""),
136
  "corrections": json.loads(meta.get("corrections", "[]")),
137
  "score": round(1 - dist, 4), # cosine similarity
138
  })