Skip to content

FlexGen reference #730

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

Merged
tjruwase merged 4 commits into from
Sep 13, 2023
Merged

FlexGen reference #730

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
cee3857
FlexGen reference
tjruwase Sep 12, 2023
c4f713b
Merge branch 'master' into olruwase/zero_inference
tjruwase Sep 12, 2023
2d9a6e6
Fix DS version and opt issue
tjruwase Sep 13, 2023
fd658e1
Fix script
tjruwase Sep 13, 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
1 change: 1 addition & 0 deletions inference/huggingface/zero_inference/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -172,4 +172,5 @@ In running example above, only two fully connected layers (fc1 and fc2) and the
## References

- DeepSpeed [ZeRO-Inference](https://www.deepspeed.ai/2022/09/09/zero-inference.html)
- Sheng, Ying et al. [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://arxiv.org/abs/2303.06865)
- Shen, Sheng, et al. "Q-bert: Hessian based ultra low precision quantization of bert." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 34. No. 05. 2020.
20 changes: 14 additions & 6 deletions inference/huggingface/zero_inference/run_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@
from packaging import version


assert version.parse(deepspeed.__version__) >= version.parse("0.10.2"), "ZeRO-Inference with weight quantization and kv cache offloading is available only in DeepSpeed 0.10.3+, please upgrade DeepSpeed"
assert version.parse(deepspeed.__version__) >= version.parse("0.10.3"), "ZeRO-Inference with weight quantization and kv cache offloading is available only in DeepSpeed 0.10.3+, please upgrade DeepSpeed"

def get_model_config(model_name):
if "175b" in model_name:
Expand Down Expand Up @@ -161,11 +161,19 @@ def run_generation(
return_token_type_ids = True
padding_side = "left" if config.model_type in ["opt"] else "right"

tokenizer = AutoTokenizer.from_pretrained(
model_name,
return_token_type_ids=return_token_type_ids,
padding_side=padding_side
)
if config.model_type == "opt":
tokenizer = AutoTokenizer.from_pretrained(
model_name.replace("175b", "66b"),
return_token_type_ids=return_token_type_ids,
padding_side=padding_side
)
else:
tokenizer = AutoTokenizer.from_pretrained(
model_name,
return_token_type_ids=return_token_type_ids,
padding_side=padding_side
)


tokenizer.pad_token = tokenizer.eos_token

Expand Down