Skip to content

Fix LARS/LAMB optimizer support and non-contiguous tensor handling on XPU #1902

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
jiqing-feng wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Fix LARS/LAMB optimizer support and non-contiguous tensor handling on XPU #1902

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
1baac13
enable lars on XPU
jiqing-feng Mar 19, 2026
b6bfbeb
fix optimizer
jiqing-feng Mar 19, 2026
9e70800
fix contiguous
jiqing-feng Mar 19, 2026
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
2 changes: 2 additions & 0 deletions bitsandbytes/backends/default/ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -357,9 +357,11 @@ def _(

name2optimizer_id = {
"momentum": MOMENTUM,
"lars": MOMENTUM,
"rmsprop": RMSPROP,
"adagrad": ADAGRAD,
"adam": ADAM,
"lamb": ADAM,
"lion": LION,
"ademamix": ADEMAMIX,
}
Expand Down
15 changes: 14 additions & 1 deletion bitsandbytes/backends/triton/kernels_optim.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,9 +24,11 @@

name2optimizer_id = {
"momentum": MOMENTUM,
"lars": MOMENTUM,
"rmsprop": RMSPROP,
"adagrad": ADAGRAD,
"adam": ADAM,
"lamb": ADAM,
"lion": LION,
"ademamix": ADEMAMIX,
}
Expand Down Expand Up @@ -121,7 +123,8 @@ def _optimizer_precondition_1state_32bit(

if OPTIMIZER_ID == 0: # MOMENTUM
if step == 1:
s1_vals = g_vals
# Cast to fp32 to avoid type mismatch: s1_vals is fp32 but g_vals may be fp16.
s1_vals = g_vals.to(tl.float32)
else:
s1_vals = s1_vals * beta1 + g_vals
update_norm = s1_vals * s1_vals
Expand Down Expand Up @@ -313,6 +316,10 @@ def _optimizer_update_1state_32bit_triton_kernel(
"preprocess": _optimizer_precondition_2state_32bit,
"update": _optimizer_update_2state_32bit_triton_kernel,
},
"lamb": {
"preprocess": _optimizer_precondition_2state_32bit,
"update": _optimizer_update_2state_32bit_triton_kernel,
},
"ademamix": {
"preprocess": _optimizer_precondition_2state_32bit,
"update": _optimizer_update_2state_32bit_triton_kernel,
Expand All @@ -321,6 +328,10 @@ def _optimizer_update_1state_32bit_triton_kernel(
"preprocess": _optimizer_precondition_1state_32bit,
"update": _optimizer_update_1state_32bit_triton_kernel,
},
"lars": {
"preprocess": _optimizer_precondition_1state_32bit,
"update": _optimizer_update_1state_32bit_triton_kernel,
},
"rmsprop": {
"preprocess": _optimizer_precondition_1state_32bit,
"update": _optimizer_update_1state_32bit_triton_kernel,
Expand Down Expand Up @@ -1065,9 +1076,11 @@ def _optimizer_update_2state_8bit_blockwise_triton_kernel(

name2optimizer_fn = {
"momentum": _optimizer_update_1state_8bit_blockwise_triton_kernel,
"lars": _optimizer_update_1state_8bit_blockwise_triton_kernel,
"rmsprop": _optimizer_update_1state_8bit_blockwise_triton_kernel,
"adagrad": _optimizer_update_1state_8bit_blockwise_triton_kernel,
"adam": _optimizer_update_2state_8bit_blockwise_triton_kernel,
"lamb": _optimizer_update_2state_8bit_blockwise_triton_kernel,
"lion": _optimizer_update_1state_8bit_blockwise_triton_kernel,
"ademamix": _optimizer_update_2state_8bit_blockwise_triton_kernel,
}
Expand Down
4 changes: 2 additions & 2 deletions bitsandbytes/backends/triton/ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ def quantize_blockwise(A: torch.Tensor, code: torch.Tensor, blocksize: int) -> t
torch._check_is_size(blocksize)
# torch._check(A.dtype == torch.float32, lambda: f"A must be float32 on xpu, got {A.dtype}")
with torch_accelerator_module.device(A.device):
out, absmax = kernels_8bit_quant.quantize_blockwise_triton(A, code, blocksize)
out, absmax = kernels_8bit_quant.quantize_blockwise_triton(A.contiguous(), code, blocksize)
return out, absmax.float()


Expand All @@ -30,7 +30,7 @@ def dequantize_blockwise(
# torch._check(dtype == torch.float32, lambda: f"dtype must be float32 on xpu, got {dtype}")
with torch_accelerator_module.device(A.device):
out = kernels_8bit_quant.dequant_8bit_blockwise(
A,
A.contiguous(),
absmax,
code,
blocksize,
Expand Down