bug: old version numpy.float type deprecated for numpy >= 1.24

moverscore_v2.py

@@ -159,8 +159,8 @@ def word_mover_score(refs, hyps, idf_dict_ref, idf_dict_hyp, stop_words=[], n_gr
         distance_matrix = batched_cdist_l2(raw, raw).double().cpu().numpy()
 
         for i in range(batch_size):  
-            c1 = np.zeros(raw.shape[1], dtype=np.float)
-            c2 = np.zeros(raw.shape[1], dtype=np.float)
+            c1 = np.zeros(raw.shape[1], dtype=float)
+            c2 = np.zeros(raw.shape[1], dtype=float)
             c1[:len(ref_idf[i])] = ref_idf[i]
             c2[len(ref_idf[i]):] = hyp_idf[i]
 
@@ -169,7 +169,7 @@ def word_mover_score(refs, hyps, idf_dict_ref, idf_dict_hyp, stop_words=[], n_gr
 
             dst = distance_matrix[i]
             _, flow = emd_with_flow(c1, c2, dst)
-            flow = np.array(flow, dtype=np.float32)
+            flow = np.array(flow, dtype=float)
             score = 1./(1. + np.sum(flow * dst))#1 - np.sum(flow * dst)
             preds.append(score)
 
@@ -198,8 +198,8 @@ def plot_example(is_flow, reference, translation, device='cuda:0'):
 
 
     i = 0
-    c1 = np.zeros(raw.shape[1], dtype=np.float)
-    c2 = np.zeros(raw.shape[1], dtype=np.float)
+    c1 = np.zeros(raw.shape[1], dtype=float)
+    c2 = np.zeros(raw.shape[1], dtype=float)
     c1[:len(ref_idf[i])] = ref_idf[i]
     c2[len(ref_idf[i]):] = hyp_idf[i]
 
@@ -210,7 +210,7 @@ def plot_example(is_flow, reference, translation, device='cuda:0'):
 
     if is_flow:        
         _, flow = emd_with_flow(c1, c2, dst)
-        new_flow = np.array(flow, dtype=np.float32)    
+        new_flow = np.array(flow, dtype=float)    
         res = new_flow[:len(ref_tokens[i]), len(ref_idf[i]): (len(ref_idf[i])+len(hyp_tokens[i]))]
     else:    
         res = 1./(1. + dst[:len(ref_tokens[i]), len(ref_idf[i]): (len(ref_idf[i])+len(hyp_tokens[i]))]) 

requirements.txt

@@ -1,3 +1,4 @@
 torch>=1.0.0
 pyemd==0.5.1
 pytorch-transformers>=1.1.0
+numpy>=1.24