Update README.md

README.md

@@ -53,13 +53,13 @@ The table above provides a high-level summary of observed accuracy deltas across
 
 ### Generation Stability (EOS Behavior)
 
-| Model | Premature EOS Rate |
+| Model | EOS Non-Emission Rate |
 |------|--------------------|
 | Torch baseline | 6% |
 | Previous INT4 GPU ONNX model | 52% |
 | Updated QAT INT4 GPU ONNX model | **11%** |
 
-The updated model reduces premature EOS generation by approximately **5×** compared to the previous INT4 GPU ONNX release, resulting in more stable and complete generations while remaining close to Torch baseline behavior.
+The updated model reduces EOS non-emission by approximately 5× compared to the previous INT4 GPU ONNX release, as observed across a large set of randomly generated prompts, resulting in more reliable sequence termination and generation behavior closer to the Torch baseline.
 
 ## Hardware Supported
 The ONNX models are tested on: