Constants§
- JSON_
BEGIN_ 🔒MARKER - JSON markers for LLM prompt
- JSON_
END_ 🔒MARKER - MAX_
CONTEXT_ UTILIZATION - Maximum percentage of context window to use in a single request
- MAX_
CONTEXT_ WINDOW - Maximum context window size for LLM in tokens
- REQUEST_
TEMPERATURE - Temperature for requests, low for deterministic results
- SYSTEM_
PROMPT 🔒 - System prompt for converting course material to markdown
- USER_
PROMPT_ 🔒START - User prompt for converting course material to markdown
Functions§
- append_
markdown_ with_ separator - Appends markdown content to a result string with proper newline separators
- block_
to_ 🔒json_ string - Converts a block to JSON string, removing any private_spec fields recursively
- blocks_
to_ 🔒json_ string - Converts a vector of blocks to JSON string, removing any private_spec fields recursively
- calculate_
safe_ token_ limit - Calculate the safe token limit based on context window and utilization
- convert_
material_ blocks_ to_ markdown_ with_ llm - Cleans content by converting the material blocks to clean markdown using an LLM
- prepare_
llm_ messages - Prepare messages for the LLM request
- process_
block_ 🔒chunk - Process a subset of blocks in a single LLM request
- process_
chunks 🔒 - Process all chunks and combine the results
- remove_
private_ 🔒spec_ recursive - Recursively removes all fields named “private_spec” from a JSON value
- split_
blocks_ into_ chunks - Split blocks into chunks that fit within token limits
- split_
oversized_ 🔒block - Splits an oversized block into smaller string chunks