File size: 1,933 Bytes
35d140d |
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 |
import torch
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("Salesforce/codegen2-1B")
model = AutoModelForCausalLM.from_pretrained("sarah111/codegen_1b_tatqa", trust_remote_code=True)
import gradio as gr
def tableQA(Question, table):
answer = 0
program = ""
try :
table = table.to_csv(header=True, index=False).strip('\n').split('\n')
table = '\n'.join(table)
#print(table)
entree = 'Table :\n{0}\n\nQuestion : {1}\n\nProgram :'.format(table, Question)
print(entree)
model.to(device)
input_ids = tokenizer(entree, return_tensors="pt").input_ids.to(device)
gen_tokens = model.generate(input_ids.to(device),do_sample=True,temperature=0.9,max_length=400)
output = tokenizer.batch_decode(gen_tokens)[0]
program = output.split("Program :````Python\n",1)[1].split("<|endoftext|>",1)[0].split("answer=")[1]
print(program)
program = program.replace(",", "")
answer = eval(program)
print(answer)
except:
print('exception')
return program, answer
demo = gr.Interface(
fn = tableQA,
inputs = [
"text",
gr.Dataframe(
headers=["", "2019", "2018"],
datatype=["str", "str", "str"],
label="Table",
),
],
outputs=[gr.Textbox(label="Derivation"),gr.Textbox(label="Result")],
title ="Outil d’aide aux financiers",
description = "Ce prototype met en évidence une situation réelle où le système de question réponse est mis en place pour permettre à des financiers de poser des questions nécessitant un raisonnement arithmétique et portant sur une table issue d’un rapport financier.",
examples=ex,
)
demo.launch(share=True)
|