github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-10 21:40:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

[misc] update pre-commit and run all files (#4752)

Browse Source 
* [misc] update pre-commit

* [misc] run pre-commit

* [misc] remove useless configuration files

* [misc] ignore cuda for clang-format
This commit is contained in:
Hongxin Liu
2023-09-19 14:20:26 +08:00
committed by GitHub
parent 3c6b831c26
commit 079bf3cb26
1268 changed files with 50037 additions and 38444 deletions
Download Patch File Download Diff File Expand all files Collapse all files

59
tests/kit/model_zoo/transformers/opt.py
View File

@@ -20,8 +20,8 @@ def data_gen_for_causal_lm():
# LM data gen
# the `labels` of LM is the token of the output, cause no padding, use `input_ids` as `labels`
data = data_gen()
labels = data['input_ids'].clone()
data['labels'] = labels
labels = data["input_ids"].clone()
data["labels"] = labels
return data
@@ -29,8 +29,8 @@ def data_gen_for_sequence_classification():
# LM data gen
# the `labels` of LM is the token of the output, cause no padding, use `input_ids` as `labels`
data = data_gen()
labels = data['input_ids'].clone()
data['labels'] = torch.tensor([1])
data["input_ids"].clone()
data["labels"] = torch.tensor([1])
return data
@@ -38,14 +38,15 @@ def data_gen_for_question_answering():
# LM data gen
# the `labels` of LM is the token of the output, cause no padding, use `input_ids` as `labels`
data = data_gen()
data['start_positions'] = torch.tensor([0])
data['end_positions'] = torch.tensor([1])
data["start_positions"] = torch.tensor([0])
data["end_positions"] = torch.tensor([1])
return data
output_transform_fn = lambda x: x
loss_fn_for_opt_model = lambda x: torch.nn.functional.mse_loss(x.last_hidden_state, torch.ones_like(x.last_hidden_state)
)
loss_fn_for_opt_model = lambda x: torch.nn.functional.mse_loss(
x.last_hidden_state, torch.ones_like(x.last_hidden_state)
)
loss_fn_for_lm = lambda x: x.loss
config = transformers.OPTConfig(
hidden_size=128,
@@ -57,24 +58,30 @@ config = transformers.OPTConfig(
# register the following models
# transformers.OPTModel,
# transformers.OPTForCausalLM,
model_zoo.register(name='transformers_opt',
model_fn=lambda: transformers.OPTModel(config),
data_gen_fn=data_gen,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_opt_model,
model_attribute=ModelAttribute(has_control_flow=True))
model_zoo.register(name='transformers_opt_for_causal_lm',
model_fn=lambda: transformers.OPTForCausalLM(config),
data_gen_fn=data_gen_for_causal_lm,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_lm,
model_attribute=ModelAttribute(has_control_flow=True))
model_zoo.register(name='transformers_opt_for_question_answering',
model_fn=lambda: transformers.OPTForQuestionAnswering(config),
data_gen_fn=data_gen_for_question_answering,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_lm,
model_attribute=ModelAttribute(has_control_flow=True))
model_zoo.register(
name="transformers_opt",
model_fn=lambda: transformers.OPTModel(config),
data_gen_fn=data_gen,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_opt_model,
model_attribute=ModelAttribute(has_control_flow=True),
)
model_zoo.register(
name="transformers_opt_for_causal_lm",
model_fn=lambda: transformers.OPTForCausalLM(config),
data_gen_fn=data_gen_for_causal_lm,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_lm,
model_attribute=ModelAttribute(has_control_flow=True),
)
model_zoo.register(
name="transformers_opt_for_question_answering",
model_fn=lambda: transformers.OPTForQuestionAnswering(config),
data_gen_fn=data_gen_for_question_answering,
output_transform_fn=output_transform_fn,
loss_fn=loss_fn_for_lm,
model_attribute=ModelAttribute(has_control_flow=True),
)
# TODO The loss and gradient check in the test are failing, to be fixed.
# model_zoo.register(name='transformers_opt_for_sequence_classification',