FEAT: new data source

app.py

@@ -108,12 +108,13 @@ option = st.selectbox("Choose type of input:",
                        "03 - JSON (example CA-OOD)",
                        "04 - JSON (example Elysai)",
                        "05 - Diagnostic mode",
-                       "06 - JSON (example Elysai - large)"])
+                       "06 - JSON (example Elysai - large)",
+                       "07 - Dialogue Breakdown Challenge"])
 
 progres_bar = st.progress(0.0, text="Inference")
 
-with st.form("input_text"):
-    if "01" in option:
+if "01" in option:
+    with st.form("input_text"):
         context = st.text_area("Insert context here (one turn per line):")
         actual_text = st.text_input("Insert current turn:")
         context = list(filter(lambda x: len(x.strip()) >= 1, context.split("\n")))
@@ -131,7 +132,9 @@ with st.form("input_text"):
             ax.pie([prop_follow, prop_not_follow], labels=["Probability - Follow", "Probability - Not Follow"],
                    autopct='%1.1f%%')
             st.pyplot(fig)
-    elif "02" in option or "03" in option or "04" in option or "06" in option:
+
+if "02" in option or "03" in option or "04" in option or "06" in option:
+    with st.form("input_text"):
         from data.example_data import ca_ood, elysai, elysai_large
 
         option: str
@@ -179,7 +182,9 @@ with st.form("input_text"):
             df = pandas.DataFrame(results, columns=["Context", "Query", "Human Label", "Probability (follow)",
                                                     "Probability (not-follow)"])
             st.dataframe(df)
-    elif "05" in option:
+
+if "05" in option:
+    with st.form("input_text"):
         context_size = 5
         context = st.text_area("Insert dialogue here (one turn per line):")
         submitted = st.form_submit_button("Submit")
@@ -203,6 +208,47 @@ with st.form("input_text"):
                 aggregated_result.append([line] + scores[idx].tolist())
             st.table(aggregated_result)
 
+if "07" in option:
+    from data.example_data import dbc
+    select_conversation = st.selectbox("Which dialogue to evaluate", list(range(len(dbc))), index=0)
+    context = st.text_area("Insert dialogue here (one turn per line):", value=json.dumps([dbc[int(select_conversation)]]))
+    st.markdown("# Formatted form")
+    context_json = json.loads(context)
+    output = ""
+    for conversation in context_json:
+        for utterance in conversation:
+            output += " * " + utterance["text"] + "\n"
+        output += "## ------------------------ "
+    st.markdown(output)
+    with st.form("input_text"):
+        context_size = 5
+        submitted = st.form_submit_button("Submit")
+        if submitted:
+            aggregated_result = []
+            for idx, conversation in enumerate(context_json):
+                data_for_evaluation = get_evaluation_data_from_dialogue([x["text"] for x in conversation])
+                lines = []
+                scores = np.zeros(shape=(len(data_for_evaluation), context_size))
+                for datapoint in data_for_evaluation:
+                    progres_bar.progress(idx / len(data_for_evaluation), text="Inference")
+                    for actual_sentence, contexts in datapoint.items():
+                        lines.append(actual_sentence)
+                        for c in contexts:
+                            input_tensor = inference_tokenizer.get_item(context=c, actual_sentence=actual_sentence)
+                            output_model = model(**input_tensor.data).logits
+                            output_model = torch.softmax(output_model, dim=-1).detach().numpy()[0]
+                            prop_follow = output_model[0]
+                            prop_not_follow = output_model[1]
+                            scores[len(lines) - 1][len(c) - 1] = prop_follow
+
+                for idx, line in enumerate(lines):
+                    NB = conversation[idx]["NB"]
+                    PB = conversation[idx]["PB"]
+                    B = conversation[idx]["B"]
+                    aggregated_result.append([line] + [f"{NB}/{PB}/{B}"] + scores[idx].tolist())
+                aggregated_result.append([["-"] * len(aggregated_result[-1])])
+            st.table(aggregated_result)
+
 st.markdown("## Description of models:")
 for x in sorted(models.values(), key=lambda x: x["model"]):
     st.write((str(x["model"] + " - " + x["description"])))