State persistence across sessions in Agentic AI - ML Experiment: Train & Evaluate

Experiment - State persistence across sessions

Problem:You have an AI agent that learns from user interactions during a session. However, when the session ends, the agent loses all learned information and starts fresh next time.

Current Metrics:Session 1: Agent accuracy 85%, Session 2: Agent accuracy resets to 50% (random guess).

Issue:The agent does not save its learned state between sessions, causing loss of progress and poor performance in new sessions.

Your Task

Implement state persistence so the agent retains learned knowledge across sessions, improving accuracy in subsequent sessions to at least 80%.

You must keep the agent's learning algorithm unchanged.

You can only add mechanisms to save and load the agent's state.

Use simple file-based storage for persistence.

Hint 1

Hint 2

Hint 3

Solution

Agentic AI

import pickle

class SimpleAgent:
    def __init__(self):
        self.knowledge = {}

    def learn(self, data):
        for item in data:
            self.knowledge[item] = self.knowledge.get(item, 0) + 1

    def predict(self, item):
        return self.knowledge.get(item, 0) > 0

    def save_state(self, filename):
        with open(filename, 'wb') as f:
            pickle.dump(self.knowledge, f)

    def load_state(self, filename):
        try:
            with open(filename, 'rb') as f:
                self.knowledge = pickle.load(f)
        except FileNotFoundError:
            self.knowledge = {}


# Simulate session 1
agent = SimpleAgent()
agent.load_state('agent_state.pkl')
training_data_session1 = ['apple', 'banana', 'apple', 'orange']
agent.learn(training_data_session1)
agent.save_state('agent_state.pkl')

# Test prediction after session 1
print('Session 1 prediction for apple:', agent.predict('apple'))  # Expected: True

# Simulate session 2
agent2 = SimpleAgent()
agent2.load_state('agent_state.pkl')

# Test prediction at start of session 2
print('Session 2 prediction for apple:', agent2.predict('apple'))  # Expected: True

# Learn new data in session 2
training_data_session2 = ['banana', 'banana', 'grape']
agent2.learn(training_data_session2)
agent2.save_state('agent_state.pkl')

# Test prediction after session 2 learning
print('Session 2 prediction for grape:', agent2.predict('grape'))  # Expected: True

Added methods save_state and load_state to save and load the agent's knowledge dictionary using pickle.

Loaded saved state at the start of each session to retain learned knowledge.

Saved updated state at the end of each session to persist new learning.

Results Interpretation

Before: Each new session started with no knowledge, causing accuracy to drop to random guess levels (~50%).

After: The agent loads saved knowledge at session start, maintaining learned information and achieving over 80% accuracy in subsequent sessions.

Saving and loading the agent's state allows it to remember past learning, improving performance across sessions and demonstrating the importance of state persistence in AI systems.

Bonus Experiment

Try implementing state persistence using a database instead of file storage to handle larger or multiple agents.

💡 Hint

Use a simple key-value store like SQLite or Redis to save and retrieve the agent's knowledge dictionary.

Practice

1. What is the main purpose of state persistence in agentic AI systems?

easy

A. To increase the AI model size for better accuracy

B. To save AI memory so it can continue tasks across sessions

C. To speed up the AI training process by using GPUs

D. To prevent AI from accessing external data sources

Solution

1. Step 1: Understand what state persistence means

   State persistence means saving the AI's memory or data so it can be reused later.

2. Step 2: Connect state persistence to AI tasks

   This allows the AI to continue learning or interacting smoothly across different sessions.

3. Final Answer:

   To save AI memory so it can continue tasks across sessions -> Option B

4. Quick Check:

   State persistence = saving AI memory across sessions [OK]

Hint: State persistence means saving AI memory between sessions [OK]

Common Mistakes:

• Confusing state persistence with faster training
• Thinking it increases model size
• Assuming it blocks external data access

2. Which of the following is the correct Python syntax to save an AI agent's state to a file named state.pkl using the pickle module?

easy

A. pickle.write(agent_state, 'state.pkl')

B. pickle.load(agent_state, open('state.pkl', 'wb'))

C. pickle.save(agent_state, 'state.pkl')

D. pickle.dump(agent_state, open('state.pkl', 'wb'))

Solution

1. Step 1: Recall pickle syntax for saving data

   Pickle saves data using pickle.dump(object, file) with file opened in write-binary mode.

2. Step 2: Match syntax to options

   pickle.dump(agent_state, open('state.pkl', 'wb')) correctly uses pickle.dump(agent_state, open('state.pkl', 'wb')).

3. Final Answer:

   pickle.dump(agent_state, open('state.pkl', 'wb')) -> Option D

4. Quick Check:

   pickle.dump + 'wb' mode = save state [OK]

Hint: Use pickle.dump with 'wb' mode to save state [OK]

Common Mistakes:

• Using pickle.load instead of dump to save
• Using non-existent pickle.save or pickle.write
• Opening file in wrong mode like 'wb' for loading

3. Given this code snippet for loading AI state:

import pickle
with open('state.pkl', 'rb') as f:
    agent_state = pickle.load(f)
print(agent_state)

What will be the output if state.pkl contains the dictionary {'score': 42, 'level': 3}?

medium

A. None

B. 42

C. {'score': 42, 'level': 3}

D. Error: file not found

Solution

1. Step 1: Understand pickle.load behavior

   pickle.load reads the saved object exactly as it was saved, here a dictionary.

2. Step 2: Predict print output

   Printing agent_state will show the dictionary {'score': 42, 'level': 3}.

3. Final Answer:

   {'score': 42, 'level': 3} -> Option C

4. Quick Check:

   pickle.load returns saved object = dict printed [OK]

Hint: pickle.load returns saved object exactly [OK]

Common Mistakes:

• Expecting only one value instead of full dict
• Assuming file not found error without checking
• Thinking pickle.load returns None

4. You wrote this code to save AI state but it raises an error:

import pickle
agent_state = {'score': 10}
file = open('state.pkl', 'r')
pickle.dump(agent_state, file)
file.close()

What is the main error causing the failure?

medium

A. File opened in read mode 'r' instead of write-binary 'wb'

B. pickle.dump requires a string, not a dict

C. Missing import statement for pickle

D. File not closed before dumping

Solution

1. Step 1: Check file open mode for saving

   Saving with pickle.dump requires file opened in write-binary mode 'wb', not 'r'.

2. Step 2: Identify error cause

   Opening file in 'r' mode causes error because it is read-only, so dump fails.

3. Final Answer:

   File opened in read mode 'r' instead of write-binary 'wb' -> Option A

4. Quick Check:

   File mode must be 'wb' to save with pickle.dump [OK]

Hint: Open file with 'wb' mode to save pickle data [OK]

Common Mistakes:

• Using 'r' mode instead of 'wb' for saving
• Thinking pickle.dump needs string input
• Forgetting to import pickle
• Closing file before dumping

5. You want your AI agent to remember user preferences across sessions and update them dynamically. Which approach best ensures state persistence and smooth updates?

hard

A. Save preferences to a database after each change and load at start

B. Store preferences only in memory during runtime without saving

C. Save preferences once at the first session and never update

D. Write preferences to a text file without structured format

Solution

1. Step 1: Understand need for persistence and updates

   To remember and update preferences, data must be saved after each change and loaded when AI restarts.

2. Step 2: Evaluate options for persistence

   Saving to a database supports dynamic updates and reliable loading, unlike memory-only or one-time saves.

3. Final Answer:

   Save preferences to a database after each change and load at start -> Option A

4. Quick Check:

   Database save + load = persistent, updateable state [OK]

Hint: Save and load state dynamically using a database [OK]

Common Mistakes:

• Not saving updates leads to lost changes
• Using memory only loses data on restart
• Saving once prevents updates
• Unstructured text files cause data errors