Merge branch 'project-BasicRAG' into project-RAG-GUI

arjbingly · Mar 25, 2024 · 47b6e44 · 47b6e44
2 parents 58be4d8 + ee884c7
commit 47b6e44
Show file tree

Hide file tree

Showing 30 changed files with 1,822 additions and 271 deletions.
diff --git a/LICENSE b/LICENSE
diff --git a/llm_quantize/quantize.py b/llm_quantize/quantize.py
@@ -1,6 +1,6 @@
+import os
 import subprocess
 import sys
-import os
 
 
 def execute_commands(model_dir_path, quantization=None):
@@ -13,7 +13,7 @@ def execute_commands(model_dir_path, quantization=None):
         if quantization:
             model_file = f"llama.cpp/models/{model_dir_path}/ggml-model-f16.gguf"
             quantized_model_file = f"llama.cpp/models/{model_dir_path.split('/')[-1]}/ggml-model-{quantization}.gguf"
-            subprocess.run(["llama.cpp/llm_quantize", model_file, quantized_model_file, quantization], check=True)
+            subprocess.run(["llama.cpp/quantize", model_file, quantized_model_file, quantization], check=True)
 
     else:
         print("llama.cpp doesn't exist, check readme how to clone.")

diff --git a/projects/Basic-RAG/BasciRAG_CustomPrompt.py b/projects/Basic-RAG/BasciRAG_CustomPrompt.py
@@ -1,5 +1,5 @@
+from grag.components.prompt import Prompt
 from grag.rag.basic_rag import BasicRAG
-from grap.components.prompt import Prompt
 
 custom_prompt = Prompt(
     input_keys={"context", "question"},

diff --git a/projects/Basic-RAG/BasicRAG-ingest_data.py b/projects/Basic-RAG/BasicRAG-ingest_data.py
diff --git a/projects/Basic-RAG/BasicRAG_FewShotPrompt.py b/projects/Basic-RAG/BasicRAG_FewShotPrompt.py
@@ -1,5 +1,5 @@
+from grag.components.prompt import FewShotPrompt
 from grag.rag.basic_rag import BasicRAG
-from grap.components.prompt import FewShotPrompt
 
 custom_few_shot_prompt = FewShotPrompt(
     input_keys={"context", "question"},

diff --git a/projects/Basic-RAG/BasicRAG_ingest.py b/projects/Basic-RAG/BasicRAG_ingest.py
@@ -0,0 +1,14 @@
+from pathlib import Path
+
+from grag.components.multivec_retriever import Retriever
+from grag.components.vectordb.deeplake_client import DeepLakeClient
+
+# from grag.rag.basic_rag import BasicRAG
+
+client = DeepLakeClient(collection_name="test")
+retriever = Retriever(vectordb=client)
+
+dir_path = Path(__file__).parent / 'some_dir'
+
+retriever.ingest(dir_path)
+# rag = BasicRAG(doc_chain="refine")
diff --git a/projects/Basic-RAG/BasicRAG_refine.py b/projects/Basic-RAG/BasicRAG_refine.py
@@ -1,5 +1,9 @@
-from grag.grag.rag import BasicRAG
+from grag.components.multivec_retriever import Retriever
+from grag.components.vectordb.deeplake_client import DeepLakeClient
+from grag.rag.basic_rag import BasicRAG
 
+client = DeepLakeClient(collection_name="test")
+retriever = Retriever(vectordb=client)
 rag = BasicRAG(doc_chain="refine")
 
 if __name__ == "__main__":

diff --git a/projects/Basic-RAG/BasicRAG_stuff.py b/projects/Basic-RAG/BasicRAG_stuff.py
@@ -1,6 +1,10 @@
-from grag.grag.rag import BasicRAG
+from grag.components.multivec_retriever import Retriever
+from grag.components.vectordb.deeplake_client import DeepLakeClient
+from grag.rag.basic_rag import BasicRAG
 
-rag = BasicRAG(doc_chain="stuff")
+client = DeepLakeClient(collection_name="test")
+retriever = Retriever(vectordb=client)
+rag = BasicRAG(doc_chain="stuff", retriever=retriever)
 
 if __name__ == "__main__":
     while True:

diff --git a/projects/Basic-RAG/BasicRAG_v1_depricated.py b/projects/Basic-RAG/BasicRAG_v1_depricated.py
diff --git a/projects/Basic-RAG/BasicRAG_v2_depricated.py b/projects/Basic-RAG/BasicRAG_v2_depricated.py
diff --git a/projects/Retriver-GUI/retriever_app.py b/projects/Retriver-GUI/retriever_app.py
@@ -46,7 +46,7 @@ def render_search_results(self):
                     st.write(result.metadata)
 
     def check_connection(self):
-        response = self.app.retriever.client.test_connection()
+        response = self.app.retriever.vectordb.test_connection()
         if response:
             return True
         else:
@@ -55,14 +55,14 @@ def check_connection(self):
     def render_stats(self):
         st.write(f'''
         **Chroma Client Details:** \n
-            Host Address    : {self.app.retriever.client.host}:{self.app.retriever.client.port} \n
-            Collection Name : {self.app.retriever.client.collection_name} \n
-            Embeddings Type : {self.app.retriever.client.embedding_type} \n
-            Embeddings Model: {self.app.retriever.client.embedding_model} \n
-            Number of docs  : {self.app.retriever.client.collection.count()} \n
+            Host Address    : {self.app.retriever.vectordb.host}:{self.app.retriever.vectordb.port} \n
+            Collection Name : {self.app.retriever.vectordb.collection_name} \n
+            Embeddings Type : {self.app.retriever.vectordb.embedding_type} \n
+            Embeddings Model: {self.app.retriever.vectordb.embedding_model} \n
+            Number of docs  : {self.app.retriever.vectordb.collection.count()} \n
         ''')
         if st.button('Check Connection'):
-            response = self.app.retriever.client.test_connection()
+            response = self.app.retriever.vectordb.test_connection()
             if response:
                 st.write(':green[Connection Active]')
             else:

diff --git a/pyproject.toml b/pyproject.toml
@@ -42,6 +42,7 @@ dependencies = [
     "huggingface_hub>=0.20.2",
     "pydantic>=2.5.0",
     "rouge-score>=0.1.2",
+    "deeplake>=3.8.27"
 ]
 
 [project.urls]

diff --git a/src/config.ini b/src/config.ini
@@ -14,6 +14,12 @@ n_gpu_layers_cpp : -1
 std_out : True
 base_dir : ${root:root_path}/models
 
+[deeplake]
+collection_name : arxiv
+embedding_type : instructor-embedding
+embedding_model : hkunlp/instructor-xl
+store_path : ${data:data_path}/vectordb
+
 [chroma]
 host : localhost
 port : 8000
@@ -59,4 +65,7 @@ table_as_html : True
 data_path : ${root:root_path}/data
 
 [root]
-root_path : /home/ubuntu/CapStone/Capstone_5
+root_path : /home/ubuntu/volume_2k/Capstone_5
+
+[quantize]
+llama_cpp_path : ${root:root_path}