Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Support Seq2SeqLM model class (to facilitate the CodeT5+ models) #104

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Support Seq2SeqLM model class (to facilitate the CodeT5+ models) #104

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c95b1d8
Added support in the Seq2SeqLM model class (to facilitate the CodeT5+…
Jun 26, 2023
641a397
Better logging message
Jun 26, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 17 additions & 4 deletions main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,8 @@
import torch
import transformers
from accelerate import Accelerator
from transformers import AutoModelForCausalLM, AutoTokenizer, HfArgumentParser
from transformers import (AutoModelForCausalLM, AutoModelForSeq2SeqLM,
AutoTokenizer, HfArgumentParser)

from lm_eval.arguments import EvalArguments
from lm_eval.evaluator import Evaluator
Expand All @@ -29,6 +30,9 @@ def __iter__(self):
yield choice


MODEL_CLASSES = {"CausalLM": AutoModelForCausalLM, "Seq2SeqLM": AutoModelForSeq2SeqLM}


def parse_args():
parser = HfArgumentParser(EvalArguments)

Expand All @@ -37,6 +41,12 @@ def parse_args():
default="codeparrot/codeparrot-small",
help="Model to evaluate, provide a repo name in Hugging Face hub or a local path",
)
parser.add_argument(
"--model_class",
default="CausalLM",
choices=["CausalLM", "Seq2SeqLM"],
help="The model will be loaded using transformer's AutoModelFor<model_class>.",
)
parser.add_argument(
"--revision",
default=None,
Expand Down Expand Up @@ -187,11 +197,14 @@ def main():
raise ValueError(
f"Non valid precision {args.precision}, choose from: fp16, fp32, bf16"
)
model_class = MODEL_CLASSES[args.model_class]
if accelerator.is_main_process:
print(f"The model will be loaded using the class: AutoModelFor{model_class.__name__}")
if args.load_in_8bit:
print("Loading model in 8bit")
current_device = accelerator.process_index
# the model needs to fit in one GPU
model = AutoModelForCausalLM.from_pretrained(
model = model_class.from_pretrained(
args.model,
revision=args.revision,
load_in_8bit=args.load_in_8bit,
Expand All @@ -203,7 +216,7 @@ def main():
print("Loading model in 4bit")
current_device = accelerator.process_index
# the model needs to fit in one GPU
model = AutoModelForCausalLM.from_pretrained(
model = model_class.from_pretrained(
args.model,
revision=args.revision,
load_in_4bit=args.load_in_4bit,
Expand All @@ -213,7 +226,7 @@ def main():
)
else:
print(f"Loading model in {args.precision}")
model = AutoModelForCausalLM.from_pretrained(
model = model_class.from_pretrained(
args.model,
revision=args.revision,
torch_dtype=dict_precisions[args.precision],
Expand Down