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

int4/int8 vicuna download support #1609

Merged
merged 3 commits into from
Jun 29, 2023
Merged

int4/int8 vicuna download support #1609

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c00b0da
set task_topology_max_group to cpu_count
dan-garvey Jun 26, 2023
5c06a20
add download for int4/int8 mlir
dan-garvey Jun 29, 2023
6cbfc30
Merge branch 'main' into cpu_count_task_flag
dan-garvey Jun 29, 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
7 changes: 2 additions & 5 deletions apps/language_models/src/pipelines/vicuna_pipeline.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,9 +38,6 @@ def __init__(
super().__init__(model_name, hf_model_path, max_num_tokens)
self.max_sequence_length = 256
self.device = device
if precision in ["int4", "int8"]:
print("int4 and int8 are not supported yet, using fp32")
precision = "fp32"
self.precision = precision
self.first_vicuna_vmfb_path = first_vicuna_vmfb_path
self.second_vicuna_vmfb_path = second_vicuna_vmfb_path
Expand Down Expand Up @@ -103,7 +100,7 @@ def compile_first_vicuna(self):
else:
mlir_generated = False
if self.load_mlir_from_shark_tank:
if self.precision in ["fp32", "fp16"]:
if self.precision in ["fp32", "fp16", "int8", "int4"]:
# download MLIR from shark_tank for fp32/fp16
download_public_file(
f"gs://shark_tank/vicuna/unsharded/mlir/{self.first_vicuna_mlir_path.name}",
Expand Down Expand Up @@ -245,7 +242,7 @@ def compile_second_vicuna(self):
else:
mlir_generated = False
if self.load_mlir_from_shark_tank:
if self.precision in ["fp32", "fp16"]:
if self.precision in ["fp32", "fp16", "int8", "int4"]:
# download MLIR from shark_tank for fp32/fp16
download_public_file(
f"gs://shark_tank/vicuna/unsharded/mlir/{self.second_vicuna_mlir_path.name}",
Expand Down
Loading