Spaces:
Runtime error
Runtime error
Jhoeel Luna
commited on
Commit
•
fc870a0
0
Parent(s):
Duplicate from NeoConsulting/Jarvis_QuAn_v01
Browse files- .gitattributes +35 -0
- README.md +15 -0
- app.py +145 -0
- olympics_sections_document_embeddings.csv +3 -0
- olympics_sections_text.csv +0 -0
- requirements.txt +2 -0
.gitattributes
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
4 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
5 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
11 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
12 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
13 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
14 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
15 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
16 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
17 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
19 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
20 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
21 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
22 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
23 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
24 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
25 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
26 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
27 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
28 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
29 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
30 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
31 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
32 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
33 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
34 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
35 |
+
olympics_sections_document_embeddings.csv filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
title: Jarvis QuAn v01
|
3 |
+
emoji: 🌖
|
4 |
+
colorFrom: gray
|
5 |
+
colorTo: yellow
|
6 |
+
sdk: gradio
|
7 |
+
sdk_version: 3.19.1
|
8 |
+
python_version: 3.11.2
|
9 |
+
app_file: app.py
|
10 |
+
pinned: false
|
11 |
+
license: openrail
|
12 |
+
duplicated_from: NeoConsulting/Jarvis_QuAn_v01
|
13 |
+
---
|
14 |
+
|
15 |
+
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
app.py
ADDED
@@ -0,0 +1,145 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import numpy as np
|
3 |
+
import openai
|
4 |
+
import pandas as pd
|
5 |
+
import tiktoken
|
6 |
+
import gradio as gr
|
7 |
+
|
8 |
+
COMPLETIONS_MODEL = "text-davinci-003"
|
9 |
+
EMBEDDING_MODEL = "text-embedding-ada-002"
|
10 |
+
|
11 |
+
openai.api_key = os.getenv("OPENAI_API_KEY")
|
12 |
+
|
13 |
+
# 1) Preprocess the document library
|
14 |
+
df = pd.read_csv("olympics_sections_text.csv")
|
15 |
+
df = df.set_index(["title", "heading"])
|
16 |
+
|
17 |
+
def get_embedding(text: str, model: str=EMBEDDING_MODEL) -> list[float]:
|
18 |
+
result = openai.Embedding.create(
|
19 |
+
model=model,
|
20 |
+
input=text
|
21 |
+
)
|
22 |
+
return result["data"][0]["embedding"]
|
23 |
+
|
24 |
+
# uncomment the below line to caculate embeddings from scratch. ========
|
25 |
+
|
26 |
+
#def compute_doc_embeddings(df: pd.DataFrame) -> dict[tuple[str, str], list[float]]:
|
27 |
+
# return {
|
28 |
+
# idx: get_embedding(r.content) for idx, r in df.iterrows()
|
29 |
+
# }
|
30 |
+
|
31 |
+
#document_embeddings = compute_doc_embeddings(df)
|
32 |
+
|
33 |
+
def load_embeddings(fname: str) -> dict[tuple[str, str], list[float]]:
|
34 |
+
"""
|
35 |
+
Read the document embeddings and their keys from a CSV.
|
36 |
+
|
37 |
+
fname is the path to a CSV with exactly these named columns:
|
38 |
+
"title", "heading", "0", "1", ... up to the length of the embedding vectors.
|
39 |
+
"""
|
40 |
+
|
41 |
+
df = pd.read_csv(fname, header=0)
|
42 |
+
max_dim = max([int(c) for c in df.columns if c != "title" and c != "heading"])
|
43 |
+
return {
|
44 |
+
(r.title, r.heading): [r[str(i)] for i in range(max_dim + 1)] for _, r in df.iterrows()
|
45 |
+
}
|
46 |
+
|
47 |
+
document_embeddings = load_embeddings("olympics_sections_document_embeddings.csv")
|
48 |
+
|
49 |
+
# 2) Find the most similar document embeddings to the question embedding
|
50 |
+
|
51 |
+
def vector_similarity(x: list[float], y: list[float]) -> float:
|
52 |
+
"""
|
53 |
+
Returns the similarity between two vectors.
|
54 |
+
|
55 |
+
Because OpenAI Embeddings are normalized to length 1, the cosine similarity is the same as the dot product.
|
56 |
+
"""
|
57 |
+
return np.dot(np.array(x), np.array(y))
|
58 |
+
|
59 |
+
def order_document_sections_by_query_similarity(query: str, contexts: dict[(str, str), np.array]) -> list[(float, (str, str))]:
|
60 |
+
"""
|
61 |
+
Find the query embedding for the supplied query, and compare it against all of the pre-calculated document embeddings
|
62 |
+
to find the most relevant sections.
|
63 |
+
|
64 |
+
Return the list of document sections, sorted by relevance in descending order.
|
65 |
+
"""
|
66 |
+
query_embedding = get_embedding(query)
|
67 |
+
|
68 |
+
document_similarities = sorted([
|
69 |
+
(vector_similarity(query_embedding, doc_embedding), doc_index) for doc_index, doc_embedding in contexts.items()
|
70 |
+
], reverse=True)
|
71 |
+
|
72 |
+
return document_similarities
|
73 |
+
|
74 |
+
# 3) Add the most relevant document sections to the query prompt
|
75 |
+
|
76 |
+
MAX_SECTION_LEN = 500
|
77 |
+
SEPARATOR = "\n* "
|
78 |
+
ENCODING = "gpt2" # encoding for text-davinci-003
|
79 |
+
|
80 |
+
encoding = tiktoken.get_encoding(ENCODING)
|
81 |
+
separator_len = len(encoding.encode(SEPARATOR))
|
82 |
+
|
83 |
+
def construct_prompt(question: str, context_embeddings: dict, df: pd.DataFrame) -> str:
|
84 |
+
"""
|
85 |
+
Fetch relevant
|
86 |
+
"""
|
87 |
+
most_relevant_document_sections = order_document_sections_by_query_similarity(question, context_embeddings)
|
88 |
+
|
89 |
+
chosen_sections = []
|
90 |
+
chosen_sections_len = 0
|
91 |
+
chosen_sections_indexes = []
|
92 |
+
|
93 |
+
for _, section_index in most_relevant_document_sections:
|
94 |
+
# Add contexts until we run out of space.
|
95 |
+
document_section = df.loc[section_index]
|
96 |
+
|
97 |
+
chosen_sections_len += document_section.tokens + separator_len
|
98 |
+
if chosen_sections_len > MAX_SECTION_LEN:
|
99 |
+
break
|
100 |
+
|
101 |
+
chosen_sections.append(SEPARATOR + document_section.content.replace("\n", " "))
|
102 |
+
chosen_sections_indexes.append(str(section_index))
|
103 |
+
|
104 |
+
header = """Answer the question as truthfully as possible using the provided context, and if the answer is not contained within the text below, say "I don't know."\n\nContext:\n"""
|
105 |
+
|
106 |
+
return header + "".join(chosen_sections) + "\n\n Q: " + question + "\n A:"
|
107 |
+
|
108 |
+
prompt = construct_prompt(
|
109 |
+
"Who won the 2020 Summer Olympics men's high jump?",
|
110 |
+
document_embeddings,
|
111 |
+
df
|
112 |
+
)
|
113 |
+
|
114 |
+
# 4) Answer the user's question based on the context.
|
115 |
+
|
116 |
+
COMPLETIONS_API_PARAMS = {
|
117 |
+
# We use temperature of 0.0 because it gives the most predictable, factual answer.
|
118 |
+
"temperature": 0.0,
|
119 |
+
"max_tokens": 300,
|
120 |
+
"model": COMPLETIONS_MODEL,
|
121 |
+
}
|
122 |
+
|
123 |
+
def answer_query_with_context(
|
124 |
+
query: str,
|
125 |
+
df: pd.DataFrame,
|
126 |
+
document_embeddings: dict[(str, str), np.array]
|
127 |
+
) -> str:
|
128 |
+
prompt = construct_prompt(
|
129 |
+
query,
|
130 |
+
document_embeddings,
|
131 |
+
df
|
132 |
+
)
|
133 |
+
|
134 |
+
response = openai.Completion.create(
|
135 |
+
prompt=prompt,
|
136 |
+
**COMPLETIONS_API_PARAMS
|
137 |
+
)
|
138 |
+
|
139 |
+
return response["choices"][0]["text"].strip(" \n")
|
140 |
+
|
141 |
+
def answer_question(query):
|
142 |
+
return answer_query_with_context(query, df, document_embeddings)
|
143 |
+
|
144 |
+
iface = gr.Interface(fn=answer_question, inputs="text", outputs="text")
|
145 |
+
iface.launch()
|
olympics_sections_document_embeddings.csv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fdf6cfaa80e7db13902fbd8064942ceb7ad4861aa8bc0348d5edd89bf7f971f3
|
3 |
+
size 118432915
|
olympics_sections_text.csv
ADDED
The diff for this file is too large to render.
See raw diff
|
|
requirements.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
openai
|
2 |
+
tiktoken
|