Skip to content

fix: Enabled GPU based tests #649

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
dushyantbehl merged 7 commits into from
Dec 16, 2025
Merged

fix: Enabled GPU based tests #649

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
2c23201
Updated tox and pyproject
romitjain Dec 8, 2025
850430f
Updates
romitjain Dec 10, 2025
a86c868
Updated ci/cd
romitjain Dec 11, 2025
ec5c79c
Added monkey patching
romitjain Dec 11, 2025
a90a3b4
Added copy for global args
romitjain Dec 11, 2025
8abb968
Linting fixes
romitjain Dec 11, 2025
2d6ff00
Fixed errors
romitjain Dec 11, 2025
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
8 changes: 8 additions & 0 deletions pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,14 @@ fms-accel = ["fms-acceleration>=0.6.2"]
gptq-dev = ["auto_gptq>0.4.2", "optimum>=1.15.0"]
mamba = ["mamba_ssm[causal-conv1d]>=2.0.0,<3.0.0"]
scanner-dev = ["HFResourceScanner>=0.1.0"]
fms-accel-all = [
"fms-acceleration>=0.6.2",
"fms-acceleration-peft",
"fms-acceleration-foak",
"fms-acceleration-aadp",
"fms-acceleration-moe",
"fms-acceleration-odm"
]

[tool.setuptools.packages.find]
exclude = ["tests", "tests.*"]
Expand Down
33 changes: 16 additions & 17 deletions tests/acceleration/test_acceleration_framework.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -263,10 +263,10 @@ def test_framework_raises_if_used_with_missing_package():
ValueError, match="No acceleration framework package found."
):
sft_trainer.train(
MODEL_ARGS,
DATA_ARGS,
TRAIN_ARGS,
PEFT_LORA_ARGS,
copy.deepcopy(MODEL_ARGS),
copy.deepcopy(DATA_ARGS),
copy.deepcopy(TRAIN_ARGS),
copy.deepcopy(PEFT_LORA_ARGS),
quantized_lora_config=quantized_lora_config,
)

Expand Down Expand Up @@ -320,9 +320,9 @@ def test_framework_raises_due_to_invalid_arguments(
with pytest.raises(exception, match=exception_msg):
sft_trainer.train(
model_args,
DATA_ARGS,
copy.deepcopy(DATA_ARGS),
train_args,
peft_config,
copy.deepcopy(peft_config),
quantized_lora_config=quantized_lora_config,
)

Expand Down Expand Up @@ -379,7 +379,7 @@ def test_framework_initialized_properly_peft(
train_args = copy.deepcopy(TRAIN_ARGS)
train_args.output_dir = tempdir
train_args.save_strategy = "no"
train_args.fp16 = True
train_args.bf16 = True
Copy link
Collaborator Author

@romitjain romitjain Dec 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fp16 upcasting is not allowed

peft_args = copy.deepcopy(PEFT_LORA_ARGS)
peft_args.target_modules = ["q_proj", "k_proj"]

Expand All @@ -395,7 +395,7 @@ def test_framework_initialized_properly_peft(
with instantiate_model_patcher():
sft_trainer.train(
model_args,
DATA_ARGS,
copy.deepcopy(DATA_ARGS),
train_args,
peft_args,
quantized_lora_config=quantized_lora_config,
Expand Down Expand Up @@ -430,7 +430,7 @@ def test_framework_initialized_properly_foak():
train_args = copy.deepcopy(TRAIN_ARGS)
train_args.output_dir = tempdir
train_args.save_strategy = "no"
train_args.fp16 = True
train_args.bf16 = True
Copy link
Collaborator Author

@romitjain romitjain Dec 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

same reason as above

peft_args = copy.deepcopy(PEFT_LORA_ARGS)
peft_args.target_modules = ["q_proj", "k_proj"]

Expand Down Expand Up @@ -465,7 +465,7 @@ def test_framework_initialized_properly_foak():
with instantiate_model_patcher():
sft_trainer.train(
model_args,
DATA_ARGS,
copy.deepcopy(DATA_ARGS),
train_args,
peft_args,
quantized_lora_config=quantized_lora_config,
Expand Down Expand Up @@ -613,8 +613,8 @@ def test_error_raised_with_paddingfree_and_flash_attn_disabled():
model_args.use_flash_attn = False
sft_trainer.train(
model_args,
DATA_ARGS,
TRAIN_ARGS,
copy.deepcopy(DATA_ARGS),
copy.deepcopy(TRAIN_ARGS),
attention_and_distributed_packing_config=attention_and_distributed_packing_config,
)

Expand All @@ -637,8 +637,8 @@ def test_error_raised_with_multipack_and_paddingfree_disabled():
model_args = copy.deepcopy(MODEL_ARGS)
sft_trainer.train(
model_args,
DATA_ARGS,
TRAIN_ARGS,
copy.deepcopy(DATA_ARGS),
copy.deepcopy(TRAIN_ARGS),
attention_and_distributed_packing_config=attention_and_distributed_packing_config,
)

Expand All @@ -664,7 +664,7 @@ def test_error_raised_with_packing_and_paddingfree_enabled():
train_args.packing = True
sft_trainer.train(
model_args,
DATA_ARGS,
copy.deepcopy(DATA_ARGS),
train_args,
attention_and_distributed_packing_config=attention_and_distributed_packing_config,
)
Expand Down Expand Up @@ -693,7 +693,6 @@ def test_error_raised_with_fused_lora_enabled_without_quantized_argument():
train_args = copy.deepcopy(TRAIN_ARGS)
train_args.output_dir = tempdir
train_args.save_strategy = "no"
train_args.fp16 = True
peft_args = copy.deepcopy(PEFT_LORA_ARGS)
peft_args.target_modules = ["q_proj", "k_proj"]

Expand All @@ -713,7 +712,7 @@ def test_error_raised_with_fused_lora_enabled_without_quantized_argument():
with instantiate_model_patcher():
sft_trainer.train(
model_args,
DATA_ARGS,
copy.deepcopy(DATA_ARGS),
train_args,
peft_args,
quantized_lora_config=None,
Expand Down
1 change: 1 addition & 0 deletions tests/artifacts/language_models/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,3 +20,4 @@
### Constants used for model path
PREDEFINED_MODEL_PATH = os.path.join(os.path.dirname(__file__))
MAYKEYE_TINY_LLAMA_CACHED = os.path.join(PREDEFINED_MODEL_PATH, "maykeye-tinyllama-v0")
TINYMIXTRAL_MOE = "Isotonic/TinyMixtral-4x248M-MoE"
Loading