updated README

Author: PeaBrane

benchmarks/data_generator/README.md

@@ -26,10 +26,14 @@ The following tools help analyze and synthesize new data based on the [mooncake
 
 **Hash ID Generation:** Each new hash ID is the next consecutive integer after the last one used. Two `hash_ids` sharing the same integers represents the prefix overlap. To generate these increasing hash IDs from a list of texts, we provide the `texts_to_hashes` function in `hasher.py`.
 
+> [!note]The `hashes_to_texts` function can then be used to generate back random texts from these hash IDs sampling from Lorem Ipsum.
+
 **Timestamp:** The arrival time (in milliseconds) of the request since the first request, which can be the same for multiple requests arriving simultaneously.
 
 **Block Size and Hash IDs:** In this example, the `block_size` (the page size of the KV cache) is assumed to be 512. The length of the `hash_ids` array equals `input_length // block_size`.
 
+A general workflow can use `texts_to_hashes` to convert texts to hashes, then use `datagen synthesize` to generate new hashes, then use `hashes_to_texts` to convert them back to random texts.
+
 ## Prefix Analyzer
 
 The Prefix Analyzer provides statistics on a trace file, such as Input Sequence Length (ISL), Output Sequence Length (OSL), and theoretical cache hit rate.