From cb6e5043f63533292fecc47dc282ae811e5d425d Mon Sep 17 00:00:00 2001 From: Mamy Ratsimbazafy Date: Mon, 10 Nov 2025 10:37:08 +0100 Subject: [PATCH] Fix SpinQuant -> Quip link Signed-off-by: Mamy Ratsimbazafy --- examples/transform/README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/examples/transform/README.md b/examples/transform/README.md index fcf8aa73c6..89c70202a5 100644 --- a/examples/transform/README.md +++ b/examples/transform/README.md @@ -41,7 +41,7 @@ recipe = [ ] ``` -Note that `QuIPModifier` can be customized. For a full list of the available arguments, see the [docstring](/src/llmcompressor/modifiers/transform/spinquant/base.py) or documentation. +Note that `QuIPModifier` can be customized. For a full list of the available arguments, see the [docstring](/src/llmcompressor/modifiers/transform/quip/base.py) or documentation. * `rotations` determines which of the input rotation (v) or output rotations (u) should be used. * `transform_block_size` determines the size of the hadamard. Smaller hadamards require less cost at runtime. @@ -79,4 +79,4 @@ python3 benchmarks/benchmark_latency.py --model path/to/Llama-3.2-1B-Instruct-qu #### Quantized Model Latency (sec) #### | Base W4A16 | Hadacore | GEMM | | - | - | - | -| 0.4402 | 0.4489 | 1.2917 | \ No newline at end of file +| 0.4402 | 0.4489 | 1.2917 |