Skip to content

Conversation

@yiliu30
Copy link
Contributor

@yiliu30 yiliu30 commented Nov 5, 2025

No description provided.

yiliu30 and others added 30 commits October 22, 2025 22:32
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yi <[email protected]>
Signed-off-by: root <[email protected]>
Signed-off-by: root <[email protected]>
fix
Signed-off-by: root <[email protected]>
@yiliu30 yiliu30 requested a review from wenhuach21 November 5, 2025 05:08
Signed-off-by: yiliu30 <[email protected]>
Signed-off-by: yiliu30 <[email protected]>
block, self.device_map, input_ids, self.low_gpu_mem_usage, self.batch_size, device
)
if normalize_inputs:
input_ids, input_others = normalize_input(inputs)
Copy link
Contributor

Choose a reason for hiding this comment

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

why not moving these change to llmc side


# function decorator to dump the func time
def time_logger(func: Callable) -> Callable:
"""Decorator to log the execution time of a function.
Copy link
Contributor

Choose a reason for hiding this comment

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

move to utils file

self.inner_supported_types = tuple(x for x in INNER_SUPPORTED_LAYER_TYPES if x != "FP8Linear")
self.batch_dim = None
# TODO: check with heng/weiwei
self.batch_dim = 0
Copy link
Contributor

Choose a reason for hiding this comment

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

if this is required, hidden in kwargs, and add comments

q_inputs = q_inputs.pop(input_id_str[0], None)
return inputs, q_inputs

def configure_layer_config(self, enable_gguf_official_mixed: None | bool = False):
Copy link
Contributor

Choose a reason for hiding this comment

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

better set the enable_gguf_official_mixed to True by default

q_input: Union[torch.Tensor, dict, None] = None,
normalize_inputs: bool = False,
device: Union[str, torch.device] = "cpu",
auto_offload=True
Copy link
Contributor

Choose a reason for hiding this comment

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

@n1ck-guo is working on splitting this function. We should avoid fusing too many operations that are not actually used by AutoRound itself in this function.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants