add some new features for layer-wise quant #1899
Conversation
Signed-off-by: n1ck-guo <[email protected]>
⛈️ Required checks status: Has failure 🔴
Groups summary🟢 Code Scan Tests workflow
These checks are required after the changes to 🔴 Model Tests 3x workflow
These checks are required after the changes to 🔴 Unit Tests 3x-PyTorch workflow
These checks are required after the changes to Thank you for your contribution! 💜
|
for more information, see https://pre-commit.ci
| logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(filename)s L%(lineno)d: %(message)s") | ||
| logger = logging.getLogger("layer_wise_tools") |
There was a problem hiding this comment.
| logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(filename)s L%(lineno)d: %(message)s") | |
| logger = logging.getLogger("layer_wise_tools") | |
| from neural_compressor.torch.utils import logger |
| @@ -18,19 +18,24 @@ | |||
|
|
|||
| import gc | |||
| import json | |||
| import logging | |||
There was a problem hiding this comment.
| import logging |
| logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(filename)s L%(lineno)d: %(message)s") | ||
| logger = logging.getLogger("layer_wise_tools") | ||
|
|
||
| LWQ_WORKSPACE = os.path.join("layer_wise_tmp") |
There was a problem hiding this comment.
| LWQ_WORKSPACE = os.path.join("layer_wise_tmp") | |
| from neural_compressor.common import options | |
| LWQ_WORKSPACE = os.path.join(options.workspace, "lwq_tmpdir") |
| @@ -121,7 +126,7 @@ def dowload_hf_model(repo_id, cache_dir=None, repo_type=None, revision=None): | |||
| return file_path | |||
|
|
|||
|
|
|||
| def load_empty_model(pretrained_model_name_or_path, cls=AutoModelForCausalLM, **kwargs): | |||
| def load_empty_model(pretrained_model_name_or_path, cls=AutoModelForCausalLM, save_path=None, **kwargs): | |||
There was a problem hiding this comment.
save_path defaults to LWQ_WORKSPACE
| if save_path is None: | ||
| save_path = LWQ_WORKSPACE |
There was a problem hiding this comment.
remove if use default value
| pretrained_model_name_or_path, cls=AutoModelForCausalLM, device="cpu", clean_weight=True, saved_path=None, **kwargs | ||
| ): | ||
| if saved_path is None: | ||
| saved_path = LWQ_WORKSPACE |
There was a problem hiding this comment.
| pretrained_model_name_or_path, cls=AutoModelForCausalLM, device="cpu", clean_weight=True, saved_path=None, **kwargs | |
| ): | |
| if saved_path is None: | |
| saved_path = LWQ_WORKSPACE | |
| pretrained_model_name_or_path, cls=AutoModelForCausalLM, device="cpu", clean_weight=True, saved_path=LWQ_WORKSPACE, **kwargs | |
| ): |
| m.forward = partial(_forward, m, n) | ||
|
|
||
| try: | ||
| model.forward( |
There was a problem hiding this comment.
Is it applicable to all or most models?
|
marked draft and will migrate to #1883 |
Type of Change
feature
Description
add some new features for layer-wise quant, include get_weight, get_bias, update, and save/load. Make it more easy to use, like a normal model.
Expected Behavior & Potential Risk
None
How has this PR been tested?
how to reproduce the test (including hardware information)
Dependency Change?
None