Learning Timeline
Key Insights
Understanding How AI Processes Text
AI doesn't count words or characters in the traditional way. It uses 'Tokens'. For instance, a sentence might have 26 words but be counted as 38 tokens by an LLM.
Why AI Becomes 'Forgetful'
When the token usage indicator exceeds 100%, the AI begins to discard the earliest information in the conversation to make room for new tokens. This causes the AI to 'forget' the original topic.
The Impact of Increasing the Context Window
Increasing the Context Window (e.g., from 2048 to 4096) allows the AI to remember longer conversations, but it will consume more of your computer's memory (VRAM/RAM).
Prompts
AI Memory Test
Target:
LM Studio
Right now I'm reading a book called How to Take Smart Notes.
Prompt to Increase Token Usage
Target:
LM Studio
Tell me a story about cows.
Memory Re-check
Target:
LM Studio
What book am I reading right now?
Step by Step
How to Adjust the Context Window in LM Studio
- Open the LM Studio application on your computer.
- Select the local model you want to use from the list (e.g., GEMMA 3 4B).
- Locate the configuration settings panel on the right side of the screen.
- Find the 'Context Length' or 'Context Window' section.
- Enter your desired token value (e.g., change from 2048 to 4096 for larger memory capacity).
- Click the blue 'Load Model' button at the top.
- Click the 'AI Chat' tab (message icon) on the left sidebar to start a new session.
- Type your message in the chat box and press Enter.
- Monitor the 'Usage' indicator at the bottom of the screen to see the percentage of token capacity used.
- If the AI starts forgetting earlier information, click the 'Eject Model' button at the top.
- Repeat the steps to adjust the 'Context Length' to a higher value, then click 'Load Model' again.
