Skip to content

Fix: Handle LowLevelZeroPlugin with use_fp8=True #6419

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
Truong5724 wants to merge 3 commits into
base: main
Choose a base branch
from
+20 −25
Open

Fix: Handle LowLevelZeroPlugin with use_fp8=True #6419

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
740071a
Fix: Handle TorchDynamo incompatible with torch._scaled_mm
Truong5724 Apr 15, 2026
e332f9e
Merge branch 'hpcaitech:main' into Truong-Fix-FP8Scaling
Truong5724 Apr 15, 2026
c3571c7
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot] Apr 15, 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
45 changes: 20 additions & 25 deletions colossalai/quantization/fp8.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -797,37 +797,32 @@ def forward(
ctx.w_fp8_t = w_fp8.t()
ctx.inv_scale_x = inv_scale_x
ctx.inv_scale_w = inv_scale_w
out = torch._scaled_mm(
x_fp8,
ctx.w_fp8_t,
bias=bias,
out_dtype=ctx.out_dtype,
scale_a=inv_scale_x,
scale_b=inv_scale_w,
use_fast_accum=True,
)[0]

# Dequantize and compute matrix multiplication (compatible with TorchDynamo)
x_deq = x_fp8.to(ctx.out_dtype) * inv_scale_x
w_t_deq = ctx.w_fp8_t.to(ctx.out_dtype) * inv_scale_w

out = x_deq @ w_t_deq
if bias is not None:
out = out + bias.to(ctx.out_dtype)

out = out.to(ctx.out_dtype)
return out.reshape(*ctx.x_shape[:-1], w.shape[0])

@staticmethod
def backward(ctx: Any, out_grad) -> Any:
out_grad = out_grad.reshape(-1, out_grad.shape[-1])
out_grad_fp8, out_grad_scale = cast_to_fp8(out_grad, fp8_format="e5m2")
x_grad = torch._scaled_mm(
out_grad_fp8,
ctx.w_fp8_t.contiguous().t(),
out_dtype=ctx.out_dtype,
scale_a=out_grad_scale,
scale_b=ctx.inv_scale_w,
use_fast_accum=True,
)[0]
w_grad = torch._scaled_mm(
out_grad_fp8.t().contiguous(),
ctx.x_fp8.t().contiguous().t(),
out_dtype=ctx.out_dtype,
scale_a=out_grad_scale,
scale_b=ctx.inv_scale_x,
use_fast_accum=True,
)[0]

# Dequantize (force contiguous after cast)
out_grad_deq = (out_grad_fp8.to(ctx.out_dtype) * out_grad_scale).contiguous()
w_t_deq = (ctx.w_fp8_t.to(ctx.out_dtype) * ctx.inv_scale_w).contiguous()
x_deq = (ctx.x_fp8.to(ctx.out_dtype) * ctx.inv_scale_x).contiguous()

# Compute gradients
x_grad = out_grad_deq @ w_t_deq.t()
w_grad = out_grad_deq.t() @ x_deq

bias_grad = None
if ctx.has_bias:
bias_grad = out_grad.sum(0)
Expand Down