File size: 2,838 Bytes
225c8e1 fda667f 225c8e1 fda667f dd6a508 cd95ea8 fda667f 225c8e1 cd95ea8 225c8e1 23577b9 225c8e1 fda667f 225c8e1 |
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 70 71 72 73 |
import warnings
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from typing import Any, Dict
class EndpointHandler:
def __init__(self, path='', torch_dtype=torch.bfloat16, trust_remote_code=True):
self.model = AutoModelForCausalLM.from_pretrained(
path,
torch_dtype=torch_dtype,
trust_remote_code=trust_remote_code
)
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neox-20b")
if tokenizer.pad_token_id is None:
warnings.warn(
"pad_token_id is not set for the tokenizer. Using eos_token_id as pad_token_id."
)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "right" # "left"
self.tokenizer = tokenizer
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.model.eval()
self.model.to(device=self.device, dtype=torch_dtype)
self.generate_kwargs = {
"temperature": 0.01,
"top_p": 0.92,
"top_k": 0,
"max_new_tokens": 512,
"use_cache": True,
"do_sample": True,
"eos_token_id": self.tokenizer.eos_token_id,
"pad_token_id": self.tokenizer.pad_token_id,
"repetition_penalty": 1.0
}
def format_instruction(self, instruction):
INSTRUCTION_KEY = "### Instruction:"
RESPONSE_KEY = "### Response:"
END_KEY = "### End"
INTRO_BLURB = "Below is an instruction that describes a task. Write a response that appropriately completes the request."
PROMPT_FOR_GENERATION_FORMAT = """{intro}
{instruction_key}
{instruction}
{response_key}
""".format(
intro=INTRO_BLURB,
instruction_key=INSTRUCTION_KEY,
instruction="{instruction}",
response_key=RESPONSE_KEY,
)
return PROMPT_FOR_GENERATION_FORMAT.format(instruction=instruction)
def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
# process input
inputs = data.pop("inputs", data)
parameters = data.pop("parameters", None)
# preprocess
s = PROMPT_FOR_GENERATION_FORMAT.format(instruction=inputs)
input_ids = self.tokenizer(s, return_tensors="pt").input_ids.to(self.device)
gkw = {**self.generate_kwargs, **parameters}
# pass inputs with all kwargs in data
with torch.no_grad():
output_ids = self.model.generate(input_ids, **gkw)
# Slice the output_ids tensor to get only new tokens
new_tokens = output_ids[0, len(input_ids[0]) :]
output_text = self.tokenizer.decode(new_tokens, skip_special_tokens=True)
return [{"generated_text": output_text}] |