probability + more test data

README.md

@@ -19,11 +19,13 @@ $ python bin/download.py
 
 ## TODO
 
-- [ ] properly load intents data from json
+- [x] properly load intents data from json
 - [ ] add type annotations
 - [ ] do some benchmarking (maybe)
-- [ ] implement simple user interaction
+- [x] implement simple user interaction
 - [ ] python linter
+- [ ] support regex in intents to generate more testing data
+- [ ] support queries inside utterances
 
 ## RESOURCES
 

data/intents.json

@@ -5,7 +5,12 @@
       "utterances": [
           "what is the weather currently",
           "how's the weather",
-          "what's it like outside"
+          "what's it like outside",
+          "weather",
+          "can you tell me the weather",
+          "weather report",
+          "current weather",
+          "what's the weather like"
       ],
       "responses": [
       ]
@@ -15,7 +20,9 @@
       "utterances": [
           "what time is it",
           "tell me the time",
-          "what's the time"
+          "what's the time",
+          "current time",
+          "time"
       ],
       "responses": [
       ]

main.py

@@ -1,5 +1,6 @@
 import sys
 import torch
+from pipeop import pipes
 
 from nn import NeuralNet
 import preprocess
@@ -25,16 +26,31 @@
 model.load_state_dict(model_state)
 model.eval()
 
-# preprocess query
-x = preprocess.tokenize(query)
-x = preprocess.stem(x)
-x = preprocess.bag_words(x, word_dict)
-x = x.reshape(1, x.shape[0])
-x = torch.from_numpy(x)
-
-output = model(x)
+@pipes
+def preprocess_query(query):
+    x = (query
+        >> preprocess.tokenize
+        >> preprocess.stem
+        >> preprocess.bag_words(word_dict)
+    )
+    x = x.reshape(1, x.shape[0])
+    return torch.from_numpy(x)
+
+
+preprocessed = preprocess_query(query)
+# TODO catch a tensor that is all zero
+print(preprocessed)
+output = model(preprocessed)
 _, predicted = torch.max(output, dim=1)
 tag = tags[predicted.item()]
 
-print(tag)
+probs = torch.softmax(output, dim=1)
+prob = probs[0][predicted.item()]
+
+confidence_threshold = 0.75
+if prob.item() > confidence_threshold:
+    print(f"[prob={prob.item():.4f}] {tag}")
+else:
+    print("query not understood")
+