algorithm improved

ud.py

@@ -16,6 +16,8 @@ class BellmanFordTokenClassificationPipeline(TokenClassificationPipeline):
   def postprocess(self,model_outputs,**kwargs):
     if "logits" not in model_outputs:
       return self.postprocess(model_outputs[0],**kwargs)
+    return self.bellman_ford_token_classification(model_outputs,**kwargs)
+  def bellman_ford_token_classification(self,model_outputs,**kwargs):
     m=model_outputs["logits"][0].numpy()
     e=numpy.exp(m-numpy.max(m,axis=-1,keepdims=True))
     z=e/e.sum(axis=-1,keepdims=True)
@@ -56,28 +58,10 @@ class UniversalDependenciesPipeline(BellmanFordTokenClassificationPipeline):
         self.right_arc[v]=0
   def postprocess(self,model_outputs,**kwargs):
     import torch
+    kwargs["aggregation_strategy"]="simple"
     if "logits" not in model_outputs:
       return self.postprocess(model_outputs[0],**kwargs)
-    m=model_outputs["logits"][0].numpy()
-    e=numpy.exp(m-numpy.max(m,axis=-1,keepdims=True))
-    z=e/e.sum(axis=-1,keepdims=True)
-    for i in range(m.shape[0]-1,0,-1):
-      m[i-1]+=numpy.max(m[i]+self.transition,axis=1)
-    k=[numpy.argmax(m[0]+self.transition[0])]
-    for i in range(1,m.shape[0]):
-      k.append(numpy.argmax(m[i]+self.transition[k[-1]]))
-    w=[{"entity":self.model.config.id2label[j],"start":s,"end":e,"score":z[i,j]} for i,((s,e),j) in enumerate(zip(model_outputs["offset_mapping"][0].tolist(),k)) if s<e]
-    for i,t in reversed(list(enumerate(w))):
-      p=t.pop("entity")
-      if p.startswith("I-"):
-        w[i-1]["score"]=min(w[i-1]["score"],t["score"])
-        w[i-1]["end"]=w.pop(i)["end"]
-      elif p.startswith("B-"):
-        t["entity_group"]=p[2:]
-      else:
-        t["entity_group"]=p
-    for t in w:
-      t["text"]=model_outputs["sentence"][t["start"]:t["end"]]
+    w=self.bellman_ford_token_classification(model_outputs,**kwargs)
     off=[(t["start"],t["end"]) for t in w]
     for i,(s,e) in reversed(list(enumerate(off))):
       if s<e:
@@ -132,11 +116,11 @@ class UniversalDependenciesPipeline(BellmanFordTokenClassificationPipeline):
           e[i,i+j]=m[k]+self.right_arc
           k+=1
         k+=1
-    m,p=numpy.nanmax(e,axis=2),numpy.nanargmax(e,axis=2)
+    m,p=numpy.max(e,axis=2),numpy.argmax(e,axis=2)
     h=self.chu_liu_edmonds(m)
     z=[i for i,j in enumerate(h) if i==j]
     if len(z)>1:
-      k,h=z[numpy.nanargmax(m[z,z])],numpy.nanmin(m)-numpy.nanmax(m)
+      k,h=z[numpy.argmax(m[z,z])],numpy.min(m)-numpy.max(m)
       m[:,z]+=[[0 if j in z and (i!=j or i==k) else h for i in z] for j in range(m.shape[0])]
       h=self.chu_liu_edmonds(m)
     q=[self.model.config.id2label[p[j,i]].split("|") for i,j in enumerate(h)]
@@ -146,7 +130,7 @@ class UniversalDependenciesPipeline(BellmanFordTokenClassificationPipeline):
       u+="\t".join([str(i+1),t[s:e],"_",q[i][0],"_","_" if len(q[i])<3 else "|".join(q[i][1:-1]),str(0 if h[i]==i else h[i]+1),"root" if q[i][-1]=="root" else q[i][-1][2:],"_","_" if i+1<len(off) and e<off[i+1][0] else "SpaceAfter=No"])+"\n"
     return u+"\n"
   def chu_liu_edmonds(self,matrix):
-    h=numpy.nanargmax(matrix,axis=0)
+    h=numpy.argmax(matrix,axis=0)
     x=[-1 if i==j else j for i,j in enumerate(h)]
     for b in [lambda x,i,j:-1 if i not in x else x[i],lambda x,i,j:-1 if j<0 else x[j]]:
       y=[]
@@ -157,10 +141,10 @@ class UniversalDependenciesPipeline(BellmanFordTokenClassificationPipeline):
       if max(x)<0:
         return h
     y,x=[i for i,j in enumerate(x) if j==max(x)],[i for i,j in enumerate(x) if j<max(x)]
-    z=matrix-numpy.nanmax(matrix,axis=0)
-    m=numpy.block([[z[x,:][:,x],numpy.nanmax(z[x,:][:,y],axis=1).reshape(len(x),1)],[numpy.nanmax(z[y,:][:,x],axis=0),numpy.nanmax(z[y,y])]])
-    k=[j if i==len(x) else x[j] if j<len(x) else y[numpy.nanargmax(z[y,x[i]])] for i,j in enumerate(self.chu_liu_edmonds(m))]
+    z=matrix-numpy.max(matrix,axis=0)
+    m=numpy.block([[z[x,:][:,x],numpy.max(z[x,:][:,y],axis=1).reshape(len(x),1)],[numpy.max(z[y,:][:,x],axis=0),numpy.max(z[y,y])]])
+    k=[j if i==len(x) else x[j] if j<len(x) else y[numpy.argmax(z[y,x[i]])] for i,j in enumerate(self.chu_liu_edmonds(m))]
     h=[j if i in y else k[x.index(i)] for i,j in enumerate(h)]
-    i=y[numpy.nanargmax(z[x[k[-1]],y] if k[-1]<len(x) else z[y,y])]
+    i=y[numpy.argmax(z[x[k[-1]],y] if k[-1]<len(x) else z[y,y])]
     h[i]=x[k[-1]] if k[-1]<len(x) else i
     return h