Build SmoothQuant release pipeline #3010
Open
+83
−9
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Adds SMOOTHQUANT-W8A8 quantization method to the TorchAO model release pipeline. - Adjusted defaults: Increased calibration samples from 10 to 128 to ensure consistency, reduced max sequence length (SeqLen) from 2048 to 1024 - Updated HF CLI command: `huggingface-cli login` to `hf auth login` Test plan: ```bash python quantize_and_upload.py --model_id Qwen/Qwen3-8B --quant SMOOTHQUANT-W8A8 --push_to_hub --task bbh ```
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3010
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
namgyu-youn
changed the title
jerryzh168
reviewed
Sep 18, 2025
|
|
||
| ### AWQ-INT4 | ||
| [AWQ](https://arxiv.org/abs/2306.00978) is a technique to improve accuracy for weight only quantization. It improves accuracy by preserving "salient" weight channels that has high impact on the accuracy of output, through multiplying the weight channel by a scale, and do the reverse for the correspnoding activation, since activation is not quantized, there is no additional loss from activation, while the quantization loss from weight can be reduced. | ||
| ### SMOOTHQUANT-W8A8 & AWQ-INT4 |
There was a problem hiding this comment.
can you add a separate section for smoothquant?
There was a problem hiding this comment.
Yes, separating them and linking them seems better.
namgyu-youn
changed the title
jerryzh168
reviewed
Sep 22, 2025
| "model.embed_tokens": _int8_int4_embedding_config, | ||
| } | ||
| ), | ||
| "SMOOTHQUANT-W8A8": Int8DynamicActivationInt8WeightConfig(), |
There was a problem hiding this comment.
to standardize on naming, this should be: SmoothQuant-INT8-INT8 I think
There was a problem hiding this comment.
Thanks for correcting it; I missed the standard name in this script.
jerryzh168
reviewed
Sep 22, 2025
|
|
||
| Note: for initial release, please include `--populate_model_card_template` to populate model card template. | ||
|
|
||
| ### SMOOTHQUANT-W8A8 |
There was a problem hiding this comment.
can you add the command to generate smoothquant checkpoints as well? similar to AWQ-INT4
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
Adds SMOOTHQUANT-W8A8 quantization method to the TorchAO model release pipeline.
max_sequence_length(SeqLen) from 2048 to 1024huggingface-cli logintohf auth loginTest plan: