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

Fix BitsAndBytes JSON Serializable #191

Merged
merged 9 commits into from
Aug 20, 2024
Merged

Fix BitsAndBytes JSON Serializable #191

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
83fc87b
Update run_sft.py
deep-diver Aug 6, 2024
a0f38e3
fix BitsAndBytes JSON serializable
deep-diver Aug 8, 2024
e911214
get_quantization_config to return dict
deep-diver Aug 15, 2024
91680dd
Merge branch 'main' into fix-json-serializable
deep-diver Aug 19, 2024
3e78440
Merge branch 'main' into fix-json-serializable
lewtun Aug 19, 2024
b9fbf14
Merge branch 'main' into fix-json-serializable
lewtun Aug 19, 2024
aa7b0a1
to_dict() for load_in_8bit too
deep-diver Aug 20, 2024
9c6631c
convert quant test to use dict subscriptions instead of dot syntax
deep-diver Aug 20, 2024
7911200
Remove torch
deep-diver Aug 20, 2024
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
4 changes: 2 additions & 2 deletions src/alignment/model_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,11 +52,11 @@ def get_quantization_config(model_args: ModelArguments) -> BitsAndBytesConfig |
bnb_4bit_quant_type=model_args.bnb_4bit_quant_type,
bnb_4bit_use_double_quant=model_args.use_bnb_nested_quant,
bnb_4bit_quant_storage=model_args.bnb_4bit_quant_storage,
)
).to_dict()
elif model_args.load_in_8bit:
quantization_config = BitsAndBytesConfig(
load_in_8bit=True,
)
).to_dict()
else:
quantization_config = None

Expand Down
12 changes: 5 additions & 7 deletions tests/test_model_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,8 +14,6 @@
# limitations under the License.
import unittest

import torch

from alignment import (
DataArguments,
ModelArguments,
Expand All @@ -31,15 +29,15 @@ class GetQuantizationConfigTest(unittest.TestCase):
def test_4bit(self):
model_args = ModelArguments(load_in_4bit=True)
quantization_config = get_quantization_config(model_args)
self.assertTrue(quantization_config.load_in_4bit)
self.assertEqual(quantization_config.bnb_4bit_compute_dtype, torch.float16)
self.assertEqual(quantization_config.bnb_4bit_quant_type, "nf4")
self.assertFalse(quantization_config.bnb_4bit_use_double_quant)
self.assertTrue(quantization_config["load_in_4bit"])
self.assertEqual(quantization_config["bnb_4bit_compute_dtype"], "float16")
self.assertEqual(quantization_config["bnb_4bit_quant_type"], "nf4")
self.assertFalse(quantization_config["bnb_4bit_use_double_quant"])

def test_8bit(self):
model_args = ModelArguments(load_in_8bit=True)
quantization_config = get_quantization_config(model_args)
self.assertTrue(quantization_config.load_in_8bit)
self.assertTrue(quantization_config["load_in_8bit"])

def test_no_quantization(self):
model_args = ModelArguments()
Expand Down
Loading