Skip to content

Support model export for int4 wo #653

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

Open
meenchen wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Support model export for int4 wo #653

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
1 change: 1 addition & 0 deletions modelopt/torch/export/model_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
QUANTIZATION_INT8_SQ = "int8_sq"
QUANTIZATION_INT8_WO = "int8_wo"
QUANTIZATION_INT4_AWQ = "int4_awq"
QUANTIZATION_INT4_WO = "int4_wo"
QUANTIZATION_W4A8_AWQ = "w4a8_awq"
QUANTIZATION_NVFP4 = "nvfp4"
QUANTIZATION_W4A8_NVFP4_FP8 = "w4a8_nvfp4_fp8"
Expand Down
15 changes: 13 additions & 2 deletions modelopt/torch/export/quant_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,7 @@
QUANTIZATION_FP8_PB_WO,
QUANTIZATION_FP8_PC_PT,
QUANTIZATION_INT4_AWQ,
QUANTIZATION_INT4_WO,
QUANTIZATION_INT8_SQ,
QUANTIZATION_INT8_WO,
QUANTIZATION_MXFP4,
Expand Down Expand Up @@ -463,7 +464,10 @@ def _get_quantization_from_layer(layer, quantizer_attr_names: QuantizerAttrNames
assert len(weight_quantizer.block_sizes) > 0 and weight_quantizer.block_sizes[-1] > 0, (
"Invalid block_sizes for INT4 quantizer"
)
return QUANTIZATION_INT4_AWQ
if hasattr(weight_quantizer, "pre_quant_scale") and weight_quantizer.pre_quant_scale:
return QUANTIZATION_INT4_AWQ
else:
return QUANTIZATION_INT4_WO

if weight_quantizer.num_bits == 8:
if input_quantizer is not None and input_quantizer.is_enabled:
Expand Down Expand Up @@ -634,6 +638,13 @@ def process_layer_quant_config(layer_config_dict):
"has_zero_point": False,
"pre_quant_scale": True,
}
elif v == "int4_wo":
layer_config = {
"quant_algo": "W4A16",
"group_size": block_size_value,
"has_zero_point": False,
"pre_quant_scale": True,
}
elif v == "w4a8_awq":
layer_config = {
"quant_algo": "W4A8_AWQ",
Expand Down Expand Up @@ -810,7 +821,7 @@ def to_quantized_weight(
)
return (weight / weights_scaling_factor[:, None]).to(torch.float8_e4m3fn)

if quantization in [QUANTIZATION_INT4_AWQ, QUANTIZATION_W4A8_AWQ]:
if quantization in [QUANTIZATION_INT4_AWQ, QUANTIZATION_W4A8_AWQ, QUANTIZATION_INT4_WO]:
return pack_int4_in_uint8(weight, weights_scaling_factor)

if quantization in [QUANTIZATION_NVFP4, QUANTIZATION_NVFP4_AWQ, QUANTIZATION_W4A8_NVFP4_FP8]:
Expand Down
Loading