Fridah/static fp4 export

[Fridah-nv]

What does this PR do?

Type of change: ?

Overview: ?

Usage

# Add a code snippet demonstrating how to use this

Testing

Before your PR is "Ready for review"

• Make sure you read and follow Contributor guidelines and your commits are signed.
• Is this change backward compatible?: Yes/No
• Did you write any new necessary tests?: Yes/No
• Did you add or update any necessary documentation?: Yes/No
• Did you update Changelog?: Yes/No

Additional Information

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

• 🔍 Trigger a full review

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch fridah/static-fp4-export

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[realAsma]

Activation is quantized, can we make it more evident?

Suggested change

	NVFP4_WEIGHT_MSE_FP8_SWEEP_CFG = {
	NVFP4_W4A4_WEIGHT_MSE_FP8_SWEEP_CFG = {

[realAsma]

nit:
This is the way I think about FP8 quantization of scale. I think it is more intuitive as following

Suggested change

	weight_scaling_factor_2 = global_amax / (6.0 * 448.0)
	per_block_scale = per_block_amax / (6.0 * weight_scaling_factor_2.to(per_block_amax.device))
	per_block_scale[per_block_scale == 0] = 1.0
	# Reshape per_block_scale to match weight's block structure: (rows, num_blocks_per_row)
	num_blocks_per_row = weight.shape[-1] // block_size
	expected_shape = (*weight.shape[:-1], num_blocks_per_row)
	per_block_scale = per_block_scale.view(expected_shape)
	return per_block_scale.to(torch.float8_e4m3fn)
	per_block_scale_max = global_amax.float() / 6.0 # importance do the scale in float
	per_block_scale = per_block_amax.float() / 6.0
	per_block_scale[per_block_scale == 0] = 1.0
	# Reshape per_block_scale to match weight's block structure: (rows, num_blocks_per_row)
	num_blocks_per_row = weight.shape[-1] // block_size
	expected_shape = (*weight.shape[:-1], num_blocks_per_row)
	per_block_scale = per_block_scale.view(expected_shape)
	per_block_scale_fp8 = (per_block_scale * 448.0 / per_block_scale_max).to(torch.float8_e4m3fn)
	return per_block_scale_fp8

[realAsma]

Can we use this

Suggested change

	global_amax = weight_quantizer._global_amax.float()
	global_amax = weight_quantizer.global_amax.float()

[realAsma]

Suggested change

	assert (
	hasattr(weight_quantizer, "_global_amax") and weight_quantizer._global_amax is not None
	)
	assert weight_quantizer.global_amax is not None