Merge pull request #74 from arjbingly/project-BasicRAG

Added badge push to Jenkins

ci/modify_test_status.py

@@ -0,0 +1,22 @@
+import argparse
+
+if __name__ == '__main__':
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-f', '--fail', action='store_true')
+    args = parser.parse_args()
+    readme_path = 'README.md'  # Path to your README file
+
+    with open(readme_path, 'r') as file:
+        lines = file.readlines()
+
+    for i, line in enumerate(lines):
+        if 'img.shields.io' in line and ('passing' in line or 'failing' in line):
+            if args.f:
+                lines[i] = line.replace('passing', 'failing').replace('darggreen', 'red')
+            else:
+                lines[i] = line.replace('failing', 'passing').replace('red', 'darggreen')
+            print('README file has been updated.')
+
+    with open(readme_path, 'w') as file:
+        file.writelines(lines)

cookbook/Basic-RAG/BasicRAG_ingest.py

@@ -1,15 +1,23 @@
 """A cookbook demonstrating how to ingest pdf files for use with Basic RAG."""
 
+import asyncio
 from pathlib import Path
 
 from grag.components.multivec_retriever import Retriever
 from grag.components.vectordb.deeplake_client import DeepLakeClient
 
 # from grag.components.vectordb.chroma_client import ChromaClient
 
+SYNC = True  # Run synchronously (slow)
+ASYNC = True  # Run asynchronously 
+
 client = DeepLakeClient(collection_name="ci_test")
 # client = ChromaClient(collection_name="ci_test")
 retriever = Retriever(vectordb=client)
 
 dir_path = Path(__file__).parents[2] / "data/test/pdfs/new_papers"
-retriever.ingest(dir_path)
+
+if SYNC:
+    retriever.ingest(dir_path)
+elif ASYNC:
+    asyncio.run(retriever.aingest(dir_path))

src/grag/components/llm.py

@@ -2,6 +2,7 @@
 
 import os
 from pathlib import Path
+from typing import Optional, Union
 
 import torch
 from langchain.callbacks.manager import CallbackManager
@@ -38,18 +39,18 @@ class LLM:
 
     def __init__(
         self,
-        model_name=llm_conf["model_name"],
-        device_map=llm_conf["device_map"],
-        task=llm_conf["task"],
-        max_new_tokens=llm_conf["max_new_tokens"],
-        temperature=llm_conf["temperature"],
-        n_batch=llm_conf["n_batch_gpu_cpp"],
-        n_ctx=llm_conf["n_ctx_cpp"],
-        n_gpu_layers=llm_conf["n_gpu_layers_cpp"],
-        std_out=llm_conf["std_out"],
-        base_dir=llm_conf["base_dir"],
-        quantization=llm_conf["quantization"],
-        pipeline=llm_conf["pipeline"],
+        model_name: str = llm_conf["model_name"],
+        device_map: str = llm_conf["device_map"],
+        task: str = llm_conf["task"],
+        max_new_tokens: str = llm_conf["max_new_tokens"],
+        temperature: str = llm_conf["temperature"],
+        n_batch: str = llm_conf["n_batch_gpu_cpp"],
+        n_ctx: str = llm_conf["n_ctx_cpp"],
+        n_gpu_layers: str = llm_conf["n_gpu_layers_cpp"],
+        std_out: Union[bool, str] = llm_conf["std_out"],
+        base_dir: str = llm_conf["base_dir"],
+        quantization: str = llm_conf["quantization"],
+        pipeline: str = llm_conf["pipeline"],
     ):
         """Initialize the LLM class using the given parameters."""
         self.base_dir = Path(base_dir)
@@ -66,7 +67,7 @@ def __init__(
         if std_out:
             self.callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
         else:
-            self.callback_manager = None
+            self.callback_manager = None  # type: ignore
 
     @property
     def model_name(self):
@@ -85,7 +86,7 @@ def model_path(self):
             self.base_dir / self.model_name / f"ggml-model-{self.quantization}.gguf"
         )
 
-    def hf_pipeline(self, is_local=False):
+    def hf_pipeline(self, is_local: Optional[bool] = False):
         """Loads the model using Hugging Face transformers.
 
         Args:
@@ -161,7 +162,8 @@ def llama_cpp(self):
         return llm
 
     def load_model(
-        self, model_name=None, pipeline=None, quantization=None, is_local=None
+        self, model_name: Optional[str] = None, pipeline: Optional[str] = None, quantization: Optional[str] = None,
+        is_local: Optional[bool] = None
     ):
         """Loads the model based on the specified pipeline and model name.