Molbap HF Staff commited on
Commit
c0a0c28
·
1 Parent(s): ee16843
Files changed (1) hide show
  1. modular_graph_and_candidates.py +6 -3
modular_graph_and_candidates.py CHANGED
@@ -156,11 +156,14 @@ def embedding_similarity_clusters(models_root: Path, missing: List[str], thr: fl
156
  sims_mat = embeddings @ embeddings.T
157
 
158
  out = {}
159
- for i in range(len(names)):
160
- for j in range(i + 1, len(names)):
 
 
 
161
  s = float(sims_mat[i, j])
162
  if s >= thr:
163
- out[(names[i], names[j])] = s
164
  return out
165
 
166
 
 
156
  sims_mat = embeddings @ embeddings.T
157
 
158
  out = {}
159
+ matrix_size = embeddings.shape[0] # Actual number of embeddings we have
160
+ processed_names = names[:matrix_size] # Only use names that have embeddings
161
+
162
+ for i in range(matrix_size):
163
+ for j in range(i + 1, matrix_size):
164
  s = float(sims_mat[i, j])
165
  if s >= thr:
166
+ out[(processed_names[i], processed_names[j])] = s
167
  return out
168
 
169