MistralLima / handler.py
Ngit's picture
setup inference
8d5025a
import torch
from typing import Dict, List, Any
from transformers import AutoTokenizer, BitsAndBytesConfig
from peft import AutoPeftModelForCausalLM
def parse_output(text):
marker = "### Response:"
if marker in text:
pos = text.find(marker) + len(marker)
else:
pos = 0
return text[pos:].replace("<pad>", "").replace("</s>", "").strip()
class EndpointHandler:
def __init__(self, path="./", use_bnb=True):
bnb_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4",
bnb_4bit_compute_dtype=torch.bfloat16,
)
self.model = AutoPeftModelForCausalLM.from_pretrained(
path, load_in_8bit=False, quantization_config=bnb_config
)
self.tokenizer = AutoTokenizer.from_pretrained(path)
def __call__(self, data: Any) -> List[List[Dict[str, str]]]:
inputs = data.get("inputs", data)
prompt = f"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction: \n{inputs}\n\n### Response: \n"
parameters = data.get("parameters", {})
inputs = self.tokenizer(
prompt, return_tensors="pt", return_token_type_ids=False
).to(self.model.device)
outputs = self.model.generate(**inputs, **parameters)
return {
"generated_text": parse_output(
self.tokenizer.decode(outputs[0].tolist(), skip_special_tokens=True)
)
}