Spaces:
Sleeping
Sleeping
File size: 4,025 Bytes
33080cc 2fe32bb 33080cc 2fe32bb 33080cc f0c8373 33080cc 2fe32bb 060a333 f0c8373 060a333 f0c8373 060a333 f0c8373 f1fa604 f0c8373 f1fa604 f0c8373 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 |
from huggingface_hub import InferenceClient
import os
class ServerlessInference:
def __init__(self, vector_store_text = None, vector_store_images = None):
self.model:str = "HuggingFaceH4/zephyr-7b-beta"
self.client = InferenceClient(api_key=os.getenv("HF_SERVELESS_API"))
self.vs_text = vector_store_text
self.vs_images = vector_store_images
def test(self, query:str) -> str:
'''Responds to query using llm'''
messages:list = [
{
"role": "user",
"content": query
}
]
completion = self.client.chat.completions.create(
model=self.model,
messages=messages,
max_tokens=500
)
return completion.choices[0].message.content
def perform_rag(self, query:str):
# First perform text search
# Retrieval
retrieved_docs = self.vs_text.similarity_search(query=query, k=5)
retrieved_docs_text = [doc.page_content for doc in retrieved_docs] # We only need the text of the documents
context = "\nExtracted documents:\n"
context += "".join([f"Document {str(i)}:::\n" + doc for i, doc in enumerate(retrieved_docs_text)])
# Augmented Generation
messages:list = [
{
"role": "system",
"content": """Using the information contained in the context,
give a comprehensive answer to the question.
Respond only to the question asked, response should be concise and relevant to the question.
If the answer cannot be deduced from the context, do not give an answer. Instead say `Theres lack of information in document source.`""",
},
{
"role": "user",
"content": """Context:
{context}
---
Now here is the question you need to answer.
Question: {question}""".format(context=context, question=query),
},
]
completion = self.client.chat.completions.create(
model=self.model,
messages=messages,
max_tokens=500
)
response_text = completion.choices[0].message.content
# Image retrieval
retrieved_image = self.vs_images.similarity_search(query=query, k=5)
retrieved_docs_text = [doc.page_content for doc in retrieved_image] # We only need the text of the documents
context = "\nExtracted Images:\n"
context += "".join([f"Document {str(i)}:::\n" + doc for i, doc in enumerate(retrieved_docs_text)])
messages:list = [
{
"role": "system",
"content": """Using the information contained in the context about the images stored in the database,
give a list of identifiers of the image that best represent the kind of information seeked by the question.
Respond only to the question asked. Provide only number(s) of the source images relevant to the question.
If the image is relevant to the question then output format should be a list [1, 3, 0]
otherwise just reply empty list that is []""",
},
{
"role": "user",
"content": """Context:
Extracted Images:
Document 0:::
Rahuls playing football
Document 1:::
Rahul recieving award in Archery.
---
Now here is the question you need to answer.
Question: Who is Rahul"""
},
{
"role": "assistant",
"content": "[1, ]"
},
{
"role": "user",
"content": """Context:
{context}
---
Now here is the question you need to answer.
Question: {question}""".format(context=context, question=query),
},
]
completion = self.client.chat.completions.create(
model=self.model,
messages=messages,
max_tokens=500
)
images_list = completion.choices[0].message.content
return response_text + str(images_list) |