readme updated with new info

README.md

@@ -14,17 +14,27 @@ pinned: false
 ## Milestone 3
 
 Click the link to access the HuggingFace Streamlit application:
-[Project webpage](https://huggingface.co/spaces/rk2546/csgy6613-project-rk2546).
+[Project webpage](https://huggingface.co/spaces/rk2546/csgy-6613-project-rk2546).
 
-The HuggingFace page includes the ability to check Milestone #2 as well.
+_The HuggingFace page includes the ability to check Milestone #2 as well._
 
+The fine-tuned models have been uploaded as HuggingFace models, public and open for anyone to use. You can find and use them here:
+
+- [https://huggingface.co/rk2546/uspto-patents-abstracts](https://huggingface.co/rk2546/uspto-patents-abstracts) 
+    - Purpose: Primary purpose is to evaluate whether a patent will be accepted or rejected. Trained from patent abstracts from the USPTO database.
+    - Model Name: `rk2546/uspto-patents-abstracts`
+- [https://huggingface.co/rk2546/uspto-patents-claims/tree/main](https://huggingface.co/rk2546/uspto-patents-claims/tree/main) 
+    - Purpose: Primary purpose is to evaluate whether a patent will be accepted or rejected. Trained from patent claims from the USPTO database.
+    - Model Name: `rk2546/uspto-patents-claims`
+
+Milestone #3's implementation performs a softmax evaluation using both models, then combines the responses based on a weight parameter. The weight parameter is adjustable in the HuggingFace application interface. By default, the output from the Claims model is weighed heavier (75%) over the Abstracts model (25%).
 
 ---
 
 ## Milestone 2
 
 Click the link to access the HuggingFace Streamlit application:
-[Project webpage](https://huggingface.co/spaces/rk2546/csgy6613-project-rk2546)
+[Project webpage](https://huggingface.co/spaces/rk2546/csgy-6613-project-rk2546)
 
 ---