SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

docs: fix typo in 'quantization-aware training' (#39904)

Browse Source
This commit is contained in:
Minseo Kim
2025-08-06 23:52:43 +09:00
committed by GitHub
parent c7844c7a8e
commit dff6185d61
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
docs/source/en/quantization/fp_quant.md
View File

@@ -16,7 +16,7 @@ rendered properly in your Markdown viewer.
# FP-Quant
[FP-Quant](https://github.com/IST-DASLab/FP-Quant) is a family of quantization algorithms tailored for the Blackwell generation of Nvidia GPUs. The goal is to allow for efficient post-training quantization (PTQ) and quantization-aware trainin (QAT) of LLMs in the [MXFP4 and NVFP4 data-types](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
[FP-Quant](https://github.com/IST-DASLab/FP-Quant) is a family of quantization algorithms tailored for the Blackwell generation of Nvidia GPUs. The goal is to allow for efficient post-training quantization (PTQ) and quantization-aware training (QAT) of LLMs in the [MXFP4 and NVFP4 data-types](https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-final-pdf).
Currently, only PTQ with MXFP4 is supported. Models can either be quantized on the fly with `quantization_config=FPQuantConfig()`:
@@ -63,4 +63,4 @@ model.forward = torch.compile(model.forward, mode="max-autotune", fullgraph=True
FP-Quant currently performs best for very large batch size processing.
See [QuTLASS README](https://github.com/IST-DASLab/qutlass/blob/main/README.md) for speedups.
See [QuTLASS README](https://github.com/IST-DASLab/qutlass/blob/main/README.md) for speedups.