Update README.md

README.md

@@ -13,8 +13,31 @@ pipeline_tag: text-to-image
 
 For more information (including how to compress models yourself), check out https://huggingface.co/DFloat11
 
-This is my first time using DF11 to compress a model outside the Flux architecture. The reason why it is much easier for me to compress Flux-based models is that
+This is my first time using DF11 to compress a model outside the Flux architecture. The reason why the process for compressing Flux-based models is much easier is because the compression code requires a `pattern_dict` as input, but the original [example code](https://github.com/LeanModels/DFloat11/tree/master/examples/compress_flux1) only provides the correct `pattern_dict` for Flux, which means I have to learn the notation myself and modify it to fit other models. After a long wait, the compressed model seems to work fine on my RTX 4090, but do let me know if you run into any problems.
 
+The `pattern_dict` I used to compress this model is here:
+```python
+pattern_dict={
+    "transformer_blocks\.\d+": (
+        "norm1.linear_1",
+        "norm1.linear_2",
+        "attn1.to_q",
+        "attn1.to_k",
+        "attn1.to_v",
+        "attn1.to_out.0",
+        "norm2.linear_1",
+        "norm2.linear_2",
+        "attn2.to_q",
+        "attn2.to_k",
+        "attn2.to_v",
+        "attn2.to_out.0",
+        "norm3.linear_1",
+        "norm3.linear_2",
+        "ff.net.0.proj",
+        "ff.net.2"
+    )
+}
+```
 
 ### How to Use