diff --git a/docs/source/quick_start.md b/docs/source/quick_start.md
index fd689c8c5..470446a98 100644
--- a/docs/source/quick_start.md
+++ b/docs/source/quick_start.md
@@ -151,7 +151,7 @@ qeff_model.generate(prompts=["My name is"])
 End to End demo examples for various models are available in **notebooks** directory. Please check them out.
 
 ### Draft-Based Speculative Decoding
-Draft-based speculative decoding is the approach where a small Draft Language Model (DLM) makes `num_speculative_tokens` autoregressive speculations ahead of the Target Language Model (TLM). The objective is to predict what the TLM would have predicted if it would have been used instead of the DLM. This approach is beneficial when the autoregressive decode phase of the TLM is memory bound and thus, we can leverage the extra computing resources of our hardware by batching the speculations of the DLM as an input to TLM to validate the speculations.  
+Draft-based speculative decoding is a technique where a small Draft Language Model (DLM) makes `num_speculative_tokens` autoregressive speculations ahead of the Target Language Model (TLM). The objective is to predict what the TLM would have predicted if it would have been used instead of the DLM. This approach is beneficial when the autoregressive decode phase of the TLM is memory bound and thus, we can leverage the extra computing resources of our hardware by batching the speculations of the DLM as an input to TLM to validate the speculations.  
 
 To export and compile both DLM/TLM, add corresponding `is_tlm` and `num_speculative_tokens` for TLM and export DLM as you would any other QEfficient LLM model: