Updated README.md, removed unnecessary prompt

README.md

@@ -11,3 +11,29 @@ license: mit
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Easy Dubai Asset Management form compilation
+
+Who doesn't like to fill in form for any little thing? **Me**,
+so I decided that the best use for thousand-dollars GPU in data
+centers was to help me do just that!
+
+Basically if you live in a community managed by Dubai Asset Management
+you have to fill in a form for every kind of work that you
+want to do in your unit, including furniture delivery.
+
+With this ChatGPT 3.5 tool you can just explain in natural language
+what you want to do and it should fill in the form for you 
+automatically. If you upload an image of your signature it 
+will use it to sign the form
+
+## Other reasons why this is interesting
+In this case I am using an LLM so that a computer can understand
+natural language, for example: if you write that the
+work you need to do will take place in two days, it
+should correctly calculate the date and pass it down
+in a way that the rest of my application can parse, which I think
+is quite neat
+
+## Why is the code related to the Intel NPU all commented out?
+As of today HF cannot build the wheel: I tried to upload my 
+wheel but I built it on Windows and I guess HF uses Linux

app.py

@@ -38,7 +38,6 @@ def use_streamlit():
         with st.status("initialising LLM"):
             repository.init()
         with st.status("waiting for LLM"):
-            repository.send_prompt(pm.ingest_user_answers(user_input))
             answer = repository.send_prompt(pm.verify_user_input_prompt(user_input))
         with st.status("Checking for missing answers"):
             st.session_state["answers"] = LlmParser.parse_verification_prompt_answers(answer['content'])

prompts/prompts_manager.py

@@ -28,7 +28,7 @@ class PromptsManager:
                 f"{self.verification_prompt}")
 
     def get_work_category(self, work_description: str) -> str:
-        return (f"The work to do is {work_description}. Given the following categories {", ".join(self.work_categories.values())} "
+        return (f"The work to do is {work_description}. Given the following categories {', '.join(self.work_categories.values())} "
                 f"which ones are the most relevant? Only return the categories, separated by a semicolon")
 
     def ingest_user_answers(self, user_prompt: str) -> str: