File size: 1,385 Bytes
cd607b2 7b856a8 cd607b2 7b856a8 cd607b2 7b856a8 cd607b2 7b856a8 cd607b2 7b856a8 cd607b2 |
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 |
# + tags=["hide_inp"]
desc = """
# QA
Questions answering with embeddings. Adapted from [OpenAI Notebook](https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_embeddings.ipynb).
"""
# -
import datasets
import numpy as np
from minichain import EmbeddingPrompt, TemplatePrompt, show_log, start_chain
# We use Hugging Face Datasets as the database by assigning
# a FAISS index.
olympics = datasets.load_from_disk("olympics.data")
olympics.add_faiss_index("embeddings")
# Fast KNN retieval prompt
class KNNPrompt(EmbeddingPrompt):
def find(self, out, inp):
res = olympics.get_nearest_examples("embeddings", np.array(out), 3)
return {"question": inp, "docs": res.examples["content"]}
# QA prompt to ask question with examples
class QAPrompt(TemplatePrompt):
template_file = "qa.pmpt.tpl"
with start_chain("qa") as backend:
prompt = KNNPrompt(backend.OpenAIEmbed()).chain(QAPrompt(backend.OpenAI()))
question = "Who won the 2020 Summer Olympics men's high jump?"
gradio = prompt.to_gradio(fields=["query"],
examples=[question],
description=desc)
if __name__ == "__main__":
gradio.launch()
# # + tags=["hide_inp"]
# QAPrompt().show(
# {"question": "Who won the race?", "docs": ["doc1", "doc2", "doc3"]}, "Joe Bob"
# )
# # -
# show_log("qa.log")
|