Major update. Support for 15 LLMs, World Flora Online taxonomy validation, geolocation, 2 OCR methods, significant UI changes, stability improvements, consistent JSON parsing

api_status.yaml

@@ -0,0 +1,11 @@
+date: January 26, 2024
+missing_keys: []
+present_keys:
+- Google OCR (Valid)
+- OpenAI (Valid)
+- Azure OpenAI (Valid)
+- Palm2 (Valid)
+- Gemini (Valid)
+- Mistral (Valid)
+- HERE Geocode (Valid)
+- OpenCage Geocode (Valid)

app.py

@@ -2299,7 +2299,7 @@ st.set_page_config(layout="wide", page_icon='img/icon.ico', page_title='VoucherV
 
 # Parse the 'is_hf' argument and set it in session state
 if 'is_hf' not in st.session_state:
-    st.session_state['is_hf'] = True
+    st.session_state['is_hf'] = False
   
 
 #################################################################################################################################################

vouchervision/API_validation.py

@@ -1,5 +1,4 @@
 import os, io, openai, vertexai, json, tempfile
-import google.generativeai as genai
 from mistralai.client import MistralClient
 from mistralai.models.chat_completion import ChatMessage
 from langchain.schema import HumanMessage
@@ -242,8 +241,10 @@ class APIvalidation:
                 palm_api_key = os.getenv('PALM_API_KEY')
                 google_project_id = os.getenv('GOOGLE_PROJECT_ID')
                 google_location = os.getenv('GOOGLE_LOCATION')
-
-                vertexai.init(project=google_project_id, location=google_location)#, credentials=credentials)
+                os.environ['GOOGLE_API_KEY'] =  os.getenv('PALM_API_KEY')
+                vertexai.init(project=os.getenv('GOOGLE_PROJECT_ID'), location=os.getenv('GOOGLE_LOCATION'))
+                # genai.configure(api_key=palm_api_key)
+                # vertexai.init(project=google_project_id, location=google_location)#, credentials=credentials)
 
                 try:
                     model = TextGenerationModel.from_pretrained("text-bison@001")

vouchervision/LLM_GooglePalm2.py

@@ -2,6 +2,8 @@ import os, time
 import vertexai
 from vertexai.language_models import TextGenerationModel
 from vertexai.generative_models._generative_models import HarmCategory, HarmBlockThreshold
+from vertexai.language_models import TextGenerationModel
+from vertexai.preview.generative_models import GenerativeModel
 from langchain.output_parsers import RetryWithErrorOutputParser
 from langchain.schema import HumanMessage
 from langchain.prompts import PromptTemplate
@@ -56,7 +58,8 @@ class GooglePalm2Handler:
         self.config = {
                 "max_output_tokens": 1024,
                 "temperature": self.starting_temp,
-                "top_p": 1
+                "top_p": 1.0,
+                "top_k": 40,
             }
         self.safety_settings = {
             HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
@@ -81,9 +84,7 @@ class GooglePalm2Handler:
 
     def _build_model_chain_parser(self):
         # Instantiate the parser and the retry parser
-        self.llm_model = ChatGoogleGenerativeAI(model=self.model_name, 
-                                                max_output_tokens=self.config.get('max_output_tokens'),
-                                                top_p=self.config.get('top_p'))
+        self.llm_model = ChatGoogleGenerativeAI(model=self.model_name)
         
         self.retry_parser = RetryWithErrorOutputParser.from_llm(
                                                 parser=self.parser,
@@ -93,12 +94,16 @@ class GooglePalm2Handler:
         self.chain = self.prompt | self.call_google_palm2
 
     # Define a function to format the input for Google PaLM call
+    # https://cloud.google.com/vertex-ai/docs/generative-ai/migrate/migrate-palm-to-gemini?_ga=2.225326234.-1652490527.1705461451&_gac=1.186295771.1706291573.CjwKCAiAzc2tBhA6EiwArv-i6QCpx7xTP0yrBy9KKSwno3QXOWUe14mbp9RGZO0ShcbtFqyXii2PnRoCywgQAvD_BwE
     def call_google_palm2(self, prompt_text):
         model = TextGenerationModel.from_pretrained(self.model_name)
         response = model.predict(prompt_text.text,
                                 max_output_tokens=self.config.get('max_output_tokens'),
                                 temperature=self.config.get('temperature'),
                                 top_p=self.config.get('top_p'))
+        # model = GenerativeModel(self.model_name)
+
+        # response = model.generate_content(prompt_text.text,generation_config=self.config, safety_settings=self.safety_settings, stream=False)
         return response.text
 
 

vouchervision/utils_VoucherVision.py

@@ -462,22 +462,21 @@ class VoucherVision():
             self.has_key_google_project_location = google_project_location is not None
 
             genai_api_key = os.getenv('PALM_API_KEY')
-            google_project_id = os.getenv('GOOGLE_PROJECT_ID')
-            google_location = os.getenv('GOOGLE_LOCATION')
-
+            os.environ['GOOGLE_API_KEY'] =  os.getenv('PALM_API_KEY')
+            vertexai.init(project=os.getenv('GOOGLE_PROJECT_ID'), location=os.getenv('GOOGLE_LOCATION'))
             # try:
                 # if genai_api_key:
                     # genai.configure(api_key=genai_api_key)
             # except:
                 # print(f"failed genai.configure")
 
-            try:
-                if google_project_id and google_location:
-                    # clien, credentials = self.init_google_client('gemini-pro', 'v1')
-                    # print(credentials)
-                    vertexai.init(project=google_project_id, location=google_location)#, credentials=credentials)
-            except:
-                print(f"failed vertexai.init")
+            # try:
+            #     if google_project_id and google_location:
+            #         # clien, credentials = self.init_google_client('gemini-pro', 'v1')
+            #         # print(credentials)
+            #         vertexai.init(project=google_project_id, location=google_location)#, credentials=credentials)
+            # except:
+            #     print(f"failed vertexai.init")
             # if self.has_key_openai:
             #     openai.api_key = openai_api_key
 
@@ -548,7 +547,11 @@ class VoucherVision():
                 os.environ['GOOGLE_PROJECT_ID'] = self.cfg_private['google_palm']['project_id'] # gemini
                 os.environ['GOOGLE_LOCATION'] = self.cfg_private['google_palm']['location'] # gemini
                 # genai.configure(api_key=self.cfg_private['google_palm']['google_palm_api'])
+                os.environ['GOOGLE_API_KEY'] = self.cfg_private['google_palm']['google_palm_api']
                 vertexai.init(project=os.environ['GOOGLE_PROJECT_ID'], location=os.environ['GOOGLE_LOCATION'])
+
+
+                
                 # os.environ.pop("GOOGLE_APPLICATION_CREDENTIALS", None)
                 # os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = self.cfg_private['google_cloud']['path_json_file'] #### 
                 # os.environ['GOOGLE_API_KEY'] = self.cfg_private['google_palm']['google_palm_api']