Update README.md

README.md

@@ -33,7 +33,15 @@ dtype: bfloat16
 ## Quantization
 
 The quantization was applied using [LLM Compressor](https://github.com/vllm-project/llm-compressor) with 512 random examples from [anydef-kilt-tasks-v2](https://huggingface.co/datasets/daisd-ai/anydef-kilt-tasks-v2) dataset.
-We tested other number of examples, but did not see noticeable improvement with higher number of examples during quantization.
+We tested other numbers of examples, but did not see noticeable improvement with higher number of examples during quantization.
+
+The recipe for quantization:
+```python
+recipe = [
+    SmoothQuantModifier(smoothing_strength=0.8),
+    GPTQModifier(targets="Linear", scheme="W4A16", ignore=["lm_head"]),
+]
+```
 
 ## Inference