{megapaks} Rollback Nunchaku to 1.0.2

#158

https://github.com/nunchaku-tech/nunchaku/blob/v1.1.0/.github/workflows/release-build.yaml#L68

Note that PyTorch 2.8 is not affected.

Author: YanWenKun

cu126-megapak/Dockerfile

@@ -310,7 +310,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
     pip install -U uv \
     # Nunchaku (binary pair with PyTorch)
     && pip install \
-https://github.com/nunchaku-tech/nunchaku/releases/download/v1.1.0/nunchaku-1.1.0+torch2.9-cp312-cp312-linux_x86_64.whl \
+https://github.com/nunchaku-tech/nunchaku/releases/download/v1.0.2/nunchaku-1.0.2+torch2.9-cp312-cp312-linux_x86_64.whl \
     # FlashAttention (version pair with xFormers, binary pair with PyTorch & CUDA)
     && pip install \
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.5.4/flash_attn-2.8.3+cu126torch2.9-cp312-cp312-linux_x86_64.whl

cu128-megapak-pt29/Dockerfile

@@ -238,7 +238,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
     pip install -U uv \
     # Nunchaku (binary pair with PyTorch)
     && pip install \
-https://github.com/nunchaku-tech/nunchaku/releases/download/v1.1.0/nunchaku-1.1.0+torch2.9-cp312-cp312-linux_x86_64.whl \
+https://github.com/nunchaku-tech/nunchaku/releases/download/v1.0.2/nunchaku-1.0.2+torch2.9-cp312-cp312-linux_x86_64.whl \
     # FlashAttention (version pair with xFormers, binary pair with PyTorch & CUDA)
     && pip install \
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.0/flash_attn-2.8.3+cu128torch2.9-cp312-cp312-linux_x86_64.whl \

cu128-megapak/Dockerfile

@@ -310,7 +310,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
     pip install -U uv \
     # Nunchaku (binary pair with PyTorch)
     && pip install \
-https://github.com/nunchaku-tech/nunchaku/releases/download/v1.1.0/nunchaku-1.1.0+torch2.9-cp312-cp312-linux_x86_64.whl \
+https://github.com/nunchaku-tech/nunchaku/releases/download/v1.0.2/nunchaku-1.0.2+torch2.9-cp312-cp312-linux_x86_64.whl \
     # FlashAttention (version pair with xFormers, binary pair with PyTorch & CUDA)
     && pip install \
 https://github.com/mjun0812/flash-attention-prebuild-wheels/releases/download/v0.7.0/flash_attn-2.8.3+cu128torch2.9-cp312-cp312-linux_x86_64.whl \