Upload 9 files

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+lancedb/cbse.lance/data/615853d2-bbd1-453c-88c7-8df0edb43e5e.lance filter=lfs diff=lfs merge=lfs -text
+logo.png filter=lfs diff=lfs merge=lfs -text

backend/query_llm.py

@@ -0,0 +1,177 @@
+
+
+# import openai
+# import gradio as gr
+
+# from os import getenv
+# from typing import Any, Dict, Generator, List
+
+# from huggingface_hub import InferenceClient
+# from transformers import AutoTokenizer
+# from gradio_client import Client
+# #tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1")
+# #tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x7B-Instruct-v0.1")
+# #tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x22B-Instruct-v0.1")
+# tokenizer=''
+# temperature = 0.5
+# top_p = 0.7
+# repetition_penalty = 1.2
+
+# OPENAI_KEY = getenv("OPENAI_API_KEY")
+# HF_TOKEN = getenv("HUGGING_FACE_HUB_TOKEN")
+
+# # hf_client = InferenceClient(
+# #        "mistralai/Mistral-7B-Instruct-v0.1",
+# #        token=HF_TOKEN
+# #        )
+
+# client = Client("Qwen/Qwen1.5-110B-Chat-demo")
+# hf_client=''
+# # hf_client = InferenceClient(
+# #         "mistralai/Mixtral-8x7B-Instruct-v0.1",
+# #         token=HF_TOKEN
+# #         )
+# def format_prompt(message: str, api_kind: str):
+#     """
+#     Formats the given message using a chat template.
+
+#     Args:
+#         message (str): The user message to be formatted.
+
+#     Returns:
+#         str: Formatted message after applying the chat template.
+#     """
+
+#     # Create a list of message dictionaries with role and content
+#     messages: List[Dict[str, Any]] = [{'role': 'user', 'content': message}]
+
+#     if api_kind == "openai":
+#         return messages
+#     elif api_kind == "hf":
+#         return tokenizer.apply_chat_template(messages, tokenize=False)
+#     elif api_kind:
+#         raise ValueError("API is not supported")
+
+
+# def generate_hf(prompt: str, history: str, temperature: float = 0.5, max_new_tokens: int = 4000,
+#              top_p: float = 0.95, repetition_penalty: float = 1.0) -> Generator[str, None, str]:
+#     """
+#     Generate a sequence of tokens based on a given prompt and history using Mistral client.
+
+#     Args:
+#         prompt (str): The initial prompt for the text generation.
+#         history (str): Context or history for the text generation.
+#         temperature (float, optional): The softmax temperature for sampling. Defaults to 0.9.
+#         max_new_tokens (int, optional): Maximum number of tokens to be generated. Defaults to 256.
+#         top_p (float, optional): Nucleus sampling probability. Defaults to 0.95.
+#         repetition_penalty (float, optional): Penalty for repeated tokens. Defaults to 1.0.
+
+#     Returns:
+#         Generator[str, None, str]: A generator yielding chunks of generated text.
+#                                    Returns a final string if an error occurs.
+#     """
+
+#     temperature = max(float(temperature), 1e-2)  # Ensure temperature isn't too low
+#     top_p = float(top_p)
+    
+#     generate_kwargs = {
+#         'temperature': temperature,
+#         'max_new_tokens': max_new_tokens,
+#         'top_p': top_p,
+#         'repetition_penalty': repetition_penalty,
+#         'do_sample': True,
+#         'seed': 42,
+#         }
+    
+#     formatted_prompt = format_prompt(prompt, "hf")
+
+#     try:
+#         stream = hf_client.text_generation(formatted_prompt, **generate_kwargs,
+#                                             stream=True, details=True, return_full_text=False)
+#         output = ""
+#         for response in stream:
+#             output += response.token.text
+#             yield output
+
+#     except Exception as e:
+#         if "Too Many Requests" in str(e):
+#             print("ERROR: Too many requests on Mistral client")
+#             gr.Warning("Unfortunately Mistral is unable to process")
+#             return "Unfortunately, I am not able to process your request now."
+#         elif "Authorization header is invalid" in str(e):
+#             print("Authetification error:", str(e))
+#             gr.Warning("Authentication error: HF token was either not provided or incorrect")
+#             return "Authentication error"
+#         else:
+#             print("Unhandled Exception:", str(e))
+#             gr.Warning("Unfortunately Mistral is unable to process")
+#             return "I do not know what happened, but I couldn't understand you."
+
+# def generate_qwen(formatted_prompt: str, history: str):
+#     response = client.predict(
+#                     query=formatted_prompt,
+#                     history=[],
+#                     system='You are wonderful',
+#                     api_name="/model_chat"
+#                 )
+#     print('Response:',response)
+    
+#     #return output
+#     #return response[1][0][1]
+#     return response[1][0][1]
+    
+    
+
+# def generate_openai(prompt: str, history: str, temperature: float = 0.9, max_new_tokens: int = 256,
+#              top_p: float = 0.95, repetition_penalty: float = 1.0) -> Generator[str, None, str]:
+#     """
+#     Generate a sequence of tokens based on a given prompt and history using Mistral client.
+
+#     Args:
+#         prompt (str): The initial prompt for the text generation.
+#         history (str): Context or history for the text generation.
+#         temperature (float, optional): The softmax temperature for sampling. Defaults to 0.9.
+#         max_new_tokens (int, optional): Maximum number of tokens to be generated. Defaults to 256.
+#         top_p (float, optional): Nucleus sampling probability. Defaults to 0.95.
+#         repetition_penalty (float, optional): Penalty for repeated tokens. Defaults to 1.0.
+
+#     Returns:
+#         Generator[str, None, str]: A generator yielding chunks of generated text.
+#                                    Returns a final string if an error occurs.
+#     """
+
+#     temperature = max(float(temperature), 1e-2)  # Ensure temperature isn't too low
+#     top_p = float(top_p)
+    
+#     generate_kwargs = {
+#         'temperature': temperature,
+#         'max_tokens': max_new_tokens,
+#         'top_p': top_p,
+#         'frequency_penalty': max(-2., min(repetition_penalty, 2.)),
+#         }
+
+#     formatted_prompt = format_prompt(prompt, "openai")
+
+#     try:
+#         stream = openai.ChatCompletion.create(model="gpt-3.5-turbo-0301",
+#                                                 messages=formatted_prompt, 
+#                                                 **generate_kwargs, 
+#                                                 stream=True)
+#         output = ""
+#         for chunk in stream:
+#             output += chunk.choices[0].delta.get("content", "")
+#             yield output
+
+#     except Exception as e:
+#         if "Too Many Requests" in str(e):
+#             print("ERROR: Too many requests on OpenAI client")
+#             gr.Warning("Unfortunately OpenAI is unable to process")
+#             return "Unfortunately, I am not able to process your request now."
+#         elif "You didn't provide an API key" in str(e):
+#             print("Authetification error:", str(e))
+#             gr.Warning("Authentication error: OpenAI key was either not provided or incorrect")
+#             return "Authentication error"
+#         else:
+#             print("Unhandled Exception:", str(e))
+#             gr.Warning("Unfortunately OpenAI is unable to process")
+#             return "I do not know what happened, but I couldn't understand you."

backend/semantic_search.py

@@ -0,0 +1,28 @@
+
+import logging
+import lancedb
+import os
+from pathlib import Path
+from sentence_transformers import SentenceTransformer
+#from FlagEmbedding import LLMEmbedder, FlagReranker # Al document present here https://github.com/FlagOpen/FlagEmbedding/tree/master
+#EMB_MODEL_NAME = "thenlper/gte-base"
+EMB_MODEL_NAME = 'BAAI/llm-embedder'
+task = "qa" # Encode for a specific task (qa, icl, chat, lrlm, tool, convsearch)
+#EMB_MODEL_NAME = LLMEmbedder('BAAI/llm-embedder', use_fp16=False) # Load model (automatically use GPUs)
+
+#reranker_model = FlagReranker('BAAI/bge-reranker-base', use_fp16=True) # use_fp16 speeds up computation with a slight performance degradation
+
+
+#EMB_MODEL_NAME = "thenlper/gte-base"
+#DB_TABLE_NAME = "Huggingface_docs"
+DB_TABLE_NAME = "cbse"
+# Setting up the logging
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+retriever = SentenceTransformer(EMB_MODEL_NAME)
+
+# db
+db_uri = os.path.join(Path(__file__).parents[1], "lancedb")
+print(f'DB URL is {db_uri}')
+db = lancedb.connect(db_uri)
+table = db.open_table(DB_TABLE_NAME)

lancedb/cbse.lance/_transactions/0-bf8e63bb-3e01-448f-95c6-f20996e7415f.txn

@@ -0,0 +1 @@
+$bf8e63bb-3e01-448f-95c6-f20996e7415f�Utext ���������
*string0
84vector
 ���������
*fixed_size_list:float:7680
8

lancedb/cbse.lance/data/615853d2-bbd1-453c-88c7-8df0edb43e5e.lance

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1611699c7fdfa82b6674466912c66b6b5dbf1f21fba597cf5fc01381aacb828b
+size 16009653