Spaces:
Runtime error
Runtime error
File size: 2,837 Bytes
7c3e9f4 642e116 7c3e9f4 e43ab69 642e116 d86afc2 7c3e9f4 d86afc2 7c3e9f4 |
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 |
import logging
from telegram import Update
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch
from telegram.ext import (
CallbackContext,
)
NAME = "Conversation"
DESCRIPTION = """
Useful for building up conversation.
Input: A normal chat text
Output: A text
"""
GET_CON = range(1)
class Conversation():
tokenizer = AutoTokenizer.from_pretrained(
"microsoft/GODEL-v1_1-large-seq2seq")
model = AutoModelForSeq2SeqLM.from_pretrained(
"microsoft/GODEL-v1_1-large-seq2seq")
# async def talk(self, message: str):
# logging.info(f"{message}")
# chat_history_ids = torch.tensor([], dtype=torch.long)
# new_user_input_ids = self.tokenizer.encode(message + self.tokenizer.eos_token, return_tensors='pt')
# bot_input_ids = torch.cat([chat_history_ids, new_user_input_ids], dim=-1)
# chat_history_ids =self.model.generate(bot_input_ids, max_length=1000, pad_token_id=self.tokenizer.eos_token_id)
# return "{}".format(self.tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True))
def predict(self, input, history=[]):
instruction = "Instruction: given a dialog context and related knowledge, you need to answer the question based on the knowledge."
knowledge = '[KNOWLEDGE] '
s = list(sum(history, ()))
s.append(input)
dialog = ' EOS '.join(s)
query = f"{instruction} [CONTEXT] {dialog} {knowledge}"
input_ids = self.tokenizer.encode(f"{query}", return_tensors='pt')
print(input, s)
output = self.model.generate(input_ids, max_length=128, min_length=8, top_p=0.9, do_sample=True).tolist()
response = self.tokenizer.decode(output[0], skip_special_tokens=True)
history.append((input, response))
return response
# def generate(self, instruction, knowledge, dialog):
# if knowledge != '':
# knowledge = '[KNOWLEDGE] ' + knowledge
# dialog = ' EOS '.join(dialog)
# query = f"{instruction} [CONTEXT] {dialog} {knowledge}"
# input_ids = self.tokenizer(f"{query}", return_tensors="pt").input_ids
# outputs = self.model.generate(
# input_ids, max_length=128,
# min_length=8,
# top_p=0.9,
# do_sample=True,
# )
# output = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
# return output
async def process_conversation(self, update: Update, context: CallbackContext) -> int:
message = update.message.text
# instruction = f'Instruction: given a dialog context, you need to response empathically.'
# knowledge = ''
# dialog = []
# dialog .append(message)
text = self.predict(message)
await update.message.reply_text(f'{text}')
|