local mode and model choosing done

datachad/database.py

@@ -1,4 +1,3 @@
-import os
 import re
 
 import deeplake
@@ -16,12 +15,14 @@ def get_dataset_path() -> str:
     # to get a string that can be used to create a new dataset
     dataset_name = re.sub(r"\W+", "-", st.session_state["data_source"])
     dataset_name = re.sub(r"--+", "- ", dataset_name).strip("-")
+    # we need to differntiate between differently chunked datasets
+    dataset_name += (
+        f"-{st.session_state['chunk_size']}-{st.session_state['chunk_overlap']}"
+    )
     if st.session_state["mode"] == MODES.LOCAL:
-        if not os.path.exists(DATA_PATH):
-            os.makedirs(DATA_PATH)
         dataset_path = str(DATA_PATH / dataset_name)
     else:
-        dataset_path = f"hub://{st.session_state['activeloop_org_name']}/{dataset_name}-{st.session_state['chunk_size']}"
+        dataset_path = f"hub://{st.session_state['activeloop_org_name']}/{dataset_name}"
     return dataset_path
 
 

datachad/models.py

@@ -59,33 +59,34 @@ class MODELS(Enum):
 
 
 def get_model() -> BaseLanguageModel:
-    match st.session_state["model"].name:
-        case MODELS.GPT35TURBO.name:
-            model = ChatOpenAI(
-                model_name=st.session_state["model"].name,
-                temperature=st.session_state["temperature"],
-                openai_api_key=st.session_state["openai_api_key"],
-            )
-        case MODELS.GPT4.name:
-            model = ChatOpenAI(
-                model_name=st.session_state["model"].name,
-                temperature=st.session_state["temperature"],
-                openai_api_key=st.session_state["openai_api_key"],
-            )
-        case MODELS.GPT4ALL.name:
-            model = GPT4All(
-                model=st.session_state["model"].path,
-                n_ctx=st.session_state["model_n_ctx"],
-                backend="gptj",
-                temp=st.session_state["temperature"],
-                verbose=True,
-            )
-        # Added models need to be cased here
-        case _default:
-            msg = f"Model {st.session_state['model']} not supported!"
-            logger.error(msg)
-            st.error(msg)
-            exit
+    with st.spinner("Loading Model..."):
+        match st.session_state["model"].name:
+            case MODELS.GPT35TURBO.name:
+                model = ChatOpenAI(
+                    model_name=st.session_state["model"].name,
+                    temperature=st.session_state["temperature"],
+                    openai_api_key=st.session_state["openai_api_key"],
+                )
+            case MODELS.GPT4.name:
+                model = ChatOpenAI(
+                    model_name=st.session_state["model"].name,
+                    temperature=st.session_state["temperature"],
+                    openai_api_key=st.session_state["openai_api_key"],
+                )
+            case MODELS.GPT4ALL.name:
+                model = GPT4All(
+                    model=st.session_state["model"].path,
+                    n_ctx=st.session_state["model_n_ctx"],
+                    backend="gptj",
+                    temp=st.session_state["temperature"],
+                    verbose=True,
+                )
+            # Added models need to be cased here
+            case _default:
+                msg = f"Model {st.session_state['model']} not supported!"
+                logger.error(msg)
+                st.error(msg)
+                exit
     return model
 
 

requirements.txt

@@ -1,8 +1,8 @@
 streamlit==1.22.0
 streamlit-chat==0.0.2.2
-deeplake==3.4.1
+deeplake==3.5.2
 openai==0.27.6
-langchain==0.0.173
+langchain==0.0.178
 tiktoken==0.4.0
 unstructured==0.6.5
 pdf2image==1.16.3
@@ -11,4 +11,4 @@ beautifulsoup4==4.12.2
 bs4==0.0.1
 python-dotenv==1.0.0
 sentence-transformers==2.2.2
-pygpt4all==1.1.0
+pygpt4all==0.2.3