michaelthwan
commented
1 year ago test searching using 10 texts with a search_text. Sequential call mode : 2.879sec Batch call mode : 0.609sec Result exact match
search_text = 'delicious beans'
texts = [
"Discover the world of delicious beans with our premium selection.",
"Try our savory bean soup recipe for a delicious and nutritious meal.",
"Our roasted coffee beans are carefully selected for their rich and delicious flavor.",
"Beans are not only delicious, but also a great source of protein and dietary fiber.",
"Looking for a delicious vegan meal? Try our spicy black bean burger recipe.",
"The sky is blue and the sun is shining today.",
"I need to go grocery shopping after work to pick up some milk and bread.",
"Did you hear about the new movie that just came out? It's supposed to be really good.",
"I'm planning a trip to Europe next summer and I'm so excited.",
"My cat keeps meowing at me for no reason and it's driving me crazy.",
]
C:\Users\MW\Anaconda3\envs\searchgpt\python.exe C:/github/!searchGPT/searchGPT/playground/test_OpenAI_Embedding.py
(10, 2)
Sequential call mode:
compute_embeddings() text: Discover the world of delicious beans with our premium selection.
compute_embeddings() text: Try our savory bean soup recipe for a delicious and nutritious meal.
compute_embeddings() text: Our roasted coffee beans are carefully selected for their rich and delicious flavor.
compute_embeddings() text: Beans are not only delicious, but also a great source of protein and dietary fiber.
compute_embeddings() text: Looking for a delicious vegan meal? Try our spicy black bean burger recipe.
compute_embeddings() text: The sky is blue and the sun is shining today.
compute_embeddings() text: I need to go grocery shopping after work to pick up some milk and bread.
compute_embeddings() text: Did you hear about the new movie that just came out? It's supposed to be really good.
compute_embeddings() text: I'm planning a trip to Europe next summer and I'm so excited.
compute_embeddings() text: My cat keeps meowing at me for no reason and it's driving me crazy.
search_similar() text: delicious beans
compute_embeddings() text: delicious beans
text ... similarities
0 Discover the world of delicious beans with our... ... 0.89247
1 Try our savory bean soup recipe for a deliciou... ... 0.88657
3 Beans are not only delicious, but also a great... ... 0.87353
[3 rows x 4 columns]
_ ._ __/__ _ _ _ _ _/_ Recorded: 20:39:21 Samples: 60
/_//_/// /_\ / //_// / //_'/ // Duration: 2.880 CPU time: 0.062
/ _/ v4.4.0
Program: C:/github/!searchGPT/searchGPT/playground/test_OpenAI_Embedding.py
2.879 <module> test_OpenAI_Embedding.py:1
├─ 2.660 Series.apply pandas\core\series.py:4661
│ [3 frames hidden] pandas
│ 2.660 SeriesApply.apply_standard pandas\core\apply.py:1159
│ └─ 2.660 <lambda> test_OpenAI_Embedding.py:89
│ └─ 2.660 compute_embeddings test_OpenAI_Embedding.py:29
│ └─ 2.660 Embedding.create openai\api_resources\embedding.py:14
│ [130 frames hidden] openai, requests, urllib3, http, sock...
└─ 0.210 search_similar test_OpenAI_Embedding.py:35
└─ 0.206 compute_embeddings test_OpenAI_Embedding.py:29
└─ 0.206 Embedding.create openai\api_resources\embedding.py:14
[25 frames hidden] openai, requests, urllib3, http, sock...
Batch call mode:
compute_embeddings_2() len(texts): 10
search_similar() text: delicious beans
compute_embeddings() text: delicious beans
text ... similarities
0 Discover the world of delicious beans with our... ... 0.892470
1 Try our savory bean soup recipe for a deliciou... ... 0.886590
3 Beans are not only delicious, but also a great... ... 0.873494
[3 rows x 4 columns]
_ ._ __/__ _ _ _ _ _/_ Recorded: 20:39:24 Samples: 21
/_//_/// /_\ / //_// / //_'/ // Duration: 0.609 CPU time: 0.000
/ _/ v4.4.0
Program: C:/github/!searchGPT/searchGPT/playground/test_OpenAI_Embedding.py
0.609 <module> test_OpenAI_Embedding.py:1
├─ 0.400 compute_embeddings_2 test_OpenAI_Embedding.py:43
│ └─ 0.400 Embedding.create openai\api_resources\embedding.py:14
│ [41 frames hidden] openai, requests, urllib3, http, sock...
├─ 0.199 search_similar test_OpenAI_Embedding.py:35
│ └─ 0.196 compute_embeddings test_OpenAI_Embedding.py:29
│ └─ 0.196 Embedding.create openai\api_resources\embedding.py:14
│ [26 frames hidden] openai, requests, urllib3, http, sock...
└─ 0.008 DataFrame.__repr__ pandas\core\frame.py:1054
[70 frames hidden] pandas
Process finished with exit code 0
pro1: We don't use a lot of faiss api pro2: Stupid duplicated call pro3: enable distance -> can tune the footnote
Difficulty: how to effectively batch call