Skip to content

【inference】support load or save Llama2-7b in three patterns #8712

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
wawltor merged 5 commits into from
Jul 5, 2024
Merged

【inference】support load or save Llama2-7b in three patterns #8712

Changes from 3 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
e0a7f79
fix
lizexu123 Jul 4, 2024
d557f18
fix
lizexu123 Jul 4, 2024
23a9f3f
删除注释
lizexu123 Jul 4, 2024
d11f394
恢复+删除
lizexu123 Jul 4, 2024
de3ae3a
删除+恢复
lizexu123 Jul 4, 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
12 changes: 9 additions & 3 deletions llm/predict/predictor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -347,7 +347,11 @@ def __init__(self, config: PredictorArgument, tokenizer: PretrainedTokenizer = N
super().__init__(config, tokenizer)

params_path = os.path.join(self.config.model_name_or_path, self.config.model_prefix + ".pdiparams")
model_path = os.path.join(self.config.model_name_or_path, self.config.model_prefix + ".pdmodel")

if paddle.framework.use_pir_api():
model_path = os.path.join(self.config.model_name_or_path, self.config.model_prefix + ".json")
else:
model_path = os.path.join(self.config.model_name_or_path, self.config.model_prefix + ".pdmodel")
inference_config = paddle.inference.Config(model_path, params_path)

if self.config.device == "gpu":
Expand All @@ -357,12 +361,15 @@ def __init__(self, config: PredictorArgument, tokenizer: PretrainedTokenizer = N
# set CPU configs accordingly,
# such as enable_mkldnn, set_cpu_math_library_num_threads
inference_config.disable_gpu()
inference_config.disable_glog_info()
# inference_config.disable_glog_info()
Copy link
Copy Markdown
Collaborator

@yuanlehome yuanlehome Jul 4, 2024

Choose a reason for hiding this comment

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

恢复

Copy link
Copy Markdown
Contributor Author

@lizexu123 lizexu123 Jul 4, 2024

Choose a reason for hiding this comment

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

哦哦对

inference_config.enable_new_executor()

Copy link
Copy Markdown
Collaborator

@yuanlehome yuanlehome Jul 4, 2024

Choose a reason for hiding this comment

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

恢复

if in_pir_executor_mode():
inference_config.enable_new_ir()
if in_cinn_mode():
inference_config.enable_cinn()
# if use optimized_model to inference
# inference_config.use_optimized_model(True)
Copy link
Copy Markdown
Collaborator

@yuanlehome yuanlehome Jul 4, 2024

Choose a reason for hiding this comment

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

删掉


with static_mode_guard():
self.predictor = paddle.inference.create_predictor(inference_config)
Expand All @@ -372,7 +379,6 @@ def __init__(self, config: PredictorArgument, tokenizer: PretrainedTokenizer = N
def _preprocess(self, input_text: str | list[str]):
inputs = super()._preprocess(input_text)
inputs["max_new_tokens"] = np.array(self.config.max_length, dtype="int64")

Copy link
Copy Markdown
Collaborator

@yuanlehome yuanlehome Jul 4, 2024

Choose a reason for hiding this comment

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

恢复

inputs["top_p"] = np.array(self.config.top_p, dtype="float32")
inputs["temperature"] = np.array(self.config.temperature, dtype="float32")
inputs["top_k"] = np.array(self.config.top_k, dtype="int64")
Expand Down