Improve Makefile Compatibility and Update Model Version

.pre-commit-config.yaml

@@ -15,7 +15,7 @@ repos:
     -   id: add-trailing-comma
         args: [--py36-plus]
 -   repo: https://github.com/asottile/pyupgrade
-    rev: v3.15.2
+    rev: v3.16.0
     hooks:
     -   id: pyupgrade
         args: [--py37-plus]
@@ -33,6 +33,6 @@ repos:
     -   id: nbqa-isort
         args: ["--float-to-top"]
 -   repo: https://github.com/PyCQA/flake8
-    rev: 7.0.0
+    rev: 7.1.0
     hooks:
     -   id: flake8

Makefile

@@ -9,11 +9,11 @@ install:
 	@$(MAKE) install-kernel
 
 ensure-poetry:
-	@if [ "$(shell which poetry)" = "" ]; then \
+	@if [ -z "$$(which poetry)" ]; then \
 		echo "Did you activate the outer conda environment? Run: conda activate llm-math-education"; \
 		exit 1; \
 	else \
-		echo "Found existing Poetry installation at $(shell which poetry)."; \
+		echo "Found existing Poetry installation at $$(which poetry)"; \
 	fi
 	@poetry install
 

src/llm_math_education/prompt_utils.py

@@ -17,7 +17,7 @@ def __init__(self, intro_prompt_dict: dict):
         self.intro_prompt_dict = intro_prompt_dict
         self.pretty_name_to_id_map = {
             t[1]["pretty_name"] if "pretty_name" in t[1] else f"Prompt {i}": t[0]
-            for i, t in enumerate(self.intro_prompt_dict)
+            for i, t in enumerate(self.intro_prompt_dict.items())
         }
 
     def get_intro_prompt_pretty_names(self):
@@ -28,45 +28,38 @@ def get_intro_prompt_pretty_names(self):
         return pretty_name_list
 
     def get_intro_prompt_message_lists(self) -> list[dict[str, str]]:
-        message_lists = []
-        for prompt_info in self.intro_prompt_dict.values():
-            message_lists.append(prompt_info["messages"])
-        return message_lists
+        return [prompt_info["messages"] for prompt_info in self.intro_prompt_dict.values()]
 
     def get_default_intro_prompt(self) -> dict[str]:
         return self.intro_prompt_dict[next(iter(self.intro_prompt_dict.keys()))]
 
+    @staticmethod
     def convert_conversation_to_string(messages):
         conversation_string = ""
         for message in messages:
             conversation_string += message["role"].upper() + ":\n"
             conversation_string += message["content"] + "\n"
         return conversation_string
 
+    @staticmethod
     def convert_string_to_conversation(conversation_string: str) -> list[dict[str, str]]:
         """Given a string representing a conversation, convert into the expected messages list format.
 
-        Follows a pretty basic convention, defined in this implementation.
-
         Args:
             conversation_string (str): String representing a conversation.
 
         Returns:
             list[dict[str, str]]: List of messages, each with a "role" and "content".
         """
         messages = []
-        message = {
-            "content": "",
-        }
+        message = {"content": ""}
         for line in conversation_string.split("\n"):
             possible_role = line[:-1].lower()
             if possible_role in VALID_ROLES:
                 if "role" in message:
                     message["content"] = message["content"].strip()
                     messages.append(message)
-                    message = {
-                        "content": "",
-                    }
+                    message = {"content": ""}
                 message["role"] = possible_role
             else:
                 message["content"] += line + "\n"
@@ -127,12 +120,9 @@ def build_query(
         if previous_messages is None:
             previous_messages = self.stored_messages
         if len(previous_messages) == 0:
-            # this is a new query
             messages = [message.copy() for message in self.intro_messages]
             self.stored_messages.extend(messages)
         else:
-            # not a new query,
-            # so include the previous messages as context
             messages = [message.copy() for message in previous_messages]
         if user_query is not None:
             user_message = {
@@ -157,18 +147,17 @@ def build_query(
                 self.recent_slot_fill_dict.append(slot_fill_dict)
                 assert len(slot_fill_dict) == len(expected_slots), "Unexpected fill provided."
                 if "user_query" in slot_fill_dict and user_query is not None:
-                    # special case: fill user_query slots with the current user_query
                     slot_fill_dict["user_query"] = user_query
                     should_remove_user_query_message = True
                 try:
                     message["content"] = message["content"].format(**slot_fill_dict)
-                except KeyError:
-                    raise KeyError(f"Failed to fill {expected_slots} with {slot_fill_dict}.")
+                except KeyError as e:
+                    raise KeyError(f"Failed to fill {expected_slots} with {slot_fill_dict}. Missing key: {e}")
+                except ValueError as e:
+                    raise ValueError(f"Formatting error: {e}")
             else:
                 self.recent_slot_fill_dict.append({})
             if query_for_retrieval_context == "" and message["role"] == "user":
-                # use as retrieval context the most recent user message
-                # TODO rethink this, providing a more flexible way to specify the retrieval context
                 query_for_retrieval_context = message["content"]
         self.recent_slot_fill_dict = self.recent_slot_fill_dict[::-1]
         if should_remove_user_query_message:
@@ -182,11 +171,10 @@ def compute_stored_token_counts(self) -> int:
         total_token_count = sum(token_counts)
         return total_token_count
 
+    @staticmethod
     def identify_slots(prompt_string: str) -> list[str]:
         """Uses a regex to identify missing slots in a prompt_string.
 
-        More advanced slot formatting is not supported.
-
         Args:
             prompt_string (str): The prompt itself, with format-style slots to fill e.g. "This is a prompt with a slot: {slot_to_fill}"
 

src/streamlit_app/auth_utils.py

@@ -70,7 +70,7 @@ def validate_openai_api_key(openai_api_key: str) -> bool:
         # temporarily override api key used
         openai.api_key = openai_api_key
         completion = openai.ChatCompletion.create(
-            model="gpt-3.5-turbo-0613",
+            model="gpt-4o-mini",
             messages=[{"role": "user", "content": "Hi"}],
             request_timeout=10,
             max_tokens=1,

src/usage_demo.py

@@ -71,7 +71,7 @@
 import openai
 
 completion = openai.ChatCompletion.create(
-    model="gpt-3.5-turbo-0613",
+    model="gpt-4o-mini",
     messages=messages,
 )
 assistant_message = completion["choices"][0]["message"]

src/🤖_Math_QA.py

@@ -71,7 +71,7 @@ def process_user_query(user_query: str):
                         s += "#### " + key + ":\n\n" + value.replace("\n", "\n\n") + "\n\n\n"
                     st.markdown(s)
             completion = openai.ChatCompletion.create(
-                model="gpt-3.5-turbo-0613",
+                model="gpt-4o-mini",
                 messages=messages,
                 temperature=st.session_state.temperature,
                 request_timeout=20,