File size: 1,607 Bytes
bf66e5a d1e4ec5 bf66e5a |
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 |
from typing import Dict, List, Any
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, StoppingCriteria, StoppingCriteriaList
class EndpointHandler():
def __init__(self, path=""):
# Preload all the elements you are going to need at inference.
tokenizer = AutoTokenizer.from_pretrained(path)
model = AutoModelForCausalLM.from_pretrained(path)
tokenizer.pad_token = tokenizer.eos_token
self.pipeline = pipeline('text-generation', model=model, tokenizer=tokenizer)
self.stopping_criteria = StoppingCriteriaList([StopAtPeriodCriteria(tokenizer)])
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
"""
data args:
inputs (:obj: `str` | `PIL.Image` | `np.array`)
kwargs
Return:
A :obj:`list` | `dict`: will be serialized and returned
"""
inputs = data.pop("inputs", data)
# Bad word: id 3070 corresponds to "(*", and we do not want to output a comment
prediction = self.pipeline(inputs, stopping_criteria=self.stopping_criteria, max_new_tokens=50, return_full_text=False, bad_words_ids=[[3070]])
return prediction
class StopAtPeriodCriteria(StoppingCriteria):
def __init__(self, tokenizer):
self.tokenizer = tokenizer
def __call__(self, input_ids, scores, **kwargs):
# Decode the last generated token to text
last_token_text = self.tokenizer.decode(input_ids[:, -1], skip_special_tokens=True)
# Check if the decoded text ends with a period
return '.' in last_token_text |