add max new tokens to pipeline params

pages/3_🌱 Generate Dataset.py

@@ -116,14 +116,33 @@ domain_expert_num_generations = st.slider(
     "Number of generations for domain expert response", 1, 10, 2
 )
 
-st.markdown(
-    "Temperature is a hyperparameter that controls the randomness of the generated text. \
-        Lower temperatures will generate more deterministic text, while higher temperatures \
-        will add more variation to generations."
-)
+with st.expander("🔥 Advanced parameters"):
+    st.markdown(
+        "Temperature is a hyperparameter that controls the randomness of the generated text. \
+            Lower temperatures will generate more deterministic text, while higher temperatures \
+            will add more variation to generations."
+    )
+
+    self_instruct_temperature = st.slider(
+        "Temperature for self-instruction", 0.1, 1.0, 0.9
+    )
+    domain_expert_temperature = st.slider(
+        "Temperature for domain expert", 0.1, 1.0, 0.9
+    )
+
+    st.markdown(
+        "`max_new_tokens` is the maximum number of tokens (word like things) that can be generated by each model call. \
+            This is a way to control the length of the generated text. in some cases, you may want to increase this to \
+            generate longer responses. You should adapt this value to your model chice, but default of 2096 works \
+            in most cases."
+    )
 
-self_instruct_temperature = st.slider("Temperature for self-instruction", 0.1, 1.0, 0.9)
-domain_expert_temperature = st.slider("Temperature for domain expert", 0.1, 1.0, 0.9)
+    self_instruct_max_new_tokens = st.number_input(
+        "Max new tokens for self-instruction", value=2096
+    )
+    domain_expert_max_new_tokens = st.number_input(
+        "Max new tokens for domain expert", value=2096
+    )
 
 ###############################################################
 # ARGILLA API
@@ -178,11 +197,13 @@ if all(
                 "domain_expert_temperature": domain_expert_temperature,
                 "self_intruct_num_generations": self_intruct_num_generations,
                 "domain_expert_num_generations": domain_expert_num_generations,
+                "self_instruct_max_new_tokens": self_instruct_max_new_tokens,
+                "domain_expert_max_new_tokens": domain_expert_max_new_tokens,
             },
             hub_username=hub_username,
             hub_token=hub_token,
             project_name=project_name,
-        )   
+        )
 
     st.success(
         f"Pipeline configuration pushed to the dataset repo {hub_username}/{project_name} on the Hub."