Update README.md

README.md

@@ -167,6 +167,10 @@ Simply put, this downloads the file as a temp file, we load it in with `TextFile
 
 Why do we want to support streaming? What about streaming is important, or useful?
 
+This is useful for user-experience. Otherwise we have to wait for the entire response to complete and return which can lead to poor user-experience.
+Other uses include the ability for early stopping if an issue is detected (including the user themselves saying stop-generation) which provides some control over how many tokens are used which is potentially resource/cost-saving.
+
+
 ### On Chat Start:
 
 The next scope is where "the magic happens". On Chat Start is when a user begins a chat session. This will happen whenever a user opens a new chat window, or refreshes an existing chat window.
@@ -209,6 +213,11 @@ Now, we'll save that into our user session!
 
 Why are we using User Session here? What about Python makes us need to use this? Why not just store everything in a global variable?
 
+Isolating the environment for each user ensures that each user's data is isolated. If we store everything in global environments we risk a situation where there are concurrent users of the site and one user uploads a PDF which overrides the other user's PDF.
+From a privacy perspective this isolation ensures users don't have access to each other's data and cant see each other's chats
+Given Python has a Global
+I do not believe there is anything Python-specific that requires this. Using user sessions is considered good engineering design whether Node app or a Python app.
+
 ### On Message
 
 First, we load our chain from the user session:
@@ -330,10 +339,21 @@ Upload a PDF file of the recent DeepSeek-R1 paper and ask the following question
 
 1. What is RL and how does it help reasoning?
 2. What is the difference between DeepSeek-R1 and DeepSeek-R1-Zero?
-3. What is this paper about?
+3. What is this paper about? <-- fails
+
+![image/png](https://cdn-uploads.huggingface.co/production/uploads/6709e00db992dee2c33f85d2/8fJn8UQEUB2tQwpRymS3b.png)
+
+![image/png](https://cdn-uploads.huggingface.co/production/uploads/6709e00db992dee2c33f85d2/WKjrd4Kq0ltSxHYJaX_Vu.png)
+
+![image/png](https://cdn-uploads.huggingface.co/production/uploads/6709e00db992dee2c33f85d2/9co6kecsU2G1uJs7ZWGJj.png)
 
 Does this application pass your vibe check? Are there any immediate pitfalls you're noticing?
 
+- You cannot ask it questions where particular words do not exist in the PDF and this fails even if the concept exists in the PDF.
+  - E.g. "How does DeepSeek compare to SOTA" fails but "How does DeepSeek compare to State-of-the-art?" succeeds
+  - This is a very prevalent failure (as evident in that I cant even ask what the paper is about!) and severely restricts how well you can interact and ask questions.
+- From a UX perspective copy-paste doesn't work
+
 ## 🚧 CHALLENGE MODE 🚧
 
 For the challenge mode, please instead create a simple FastAPI backend with a simple React (or any other JS framework) frontend.