compress_texts
Function
Compresses text based on question (prompt), context (long text), and compress_rate (compression rate).
Prototype
def compress_texts(context, question, compress_rate, context_reorder)
Parameters
Parameter |
Data Type |
Required/Optional |
Description |
|---|---|---|---|
context |
String |
Required |
Long text to be summarized. Its length range is [1, 16 MB]. |
question |
String |
Required |
Prompt for summarizing the long text, which is used to calculate the relevance with the context chunk. Its length range is [1, 1000 × 1000]. |
compress_rate |
Float |
Optional |
Compression rate. The default value is 0.6. The value range is (0, 1). |
context_reorder |
Bool |
Optional |
Whether to perform re-ranking based on the score. The default value is False. If the value is True, after the relevance is calculated, chunks with low relevance are preferentially retained based on the compression rate. |
Return Value
Data Type |
Description |
|---|---|
String |
Compressed text |