docs: fix flash-attention2 typos (#576)

docs/source/usage/attention.rst

@@ -12,6 +12,6 @@ and 50-series GPUs compared to FlashAttention-2, without precision loss.
 
 The key change from `Basic Usage <./basic_usage>`_ is use ``transformer.set_attention_impl("nunchaku-fp16")`` to enable FP16 attention.
 While FlashAttention-2 is the default, FP16 attention offers better performance on modern NVIDIA GPUs.
-Switch back with ``transformer.set_attention_impl("flash-attention2")``.
+Switch back with ``transformer.set_attention_impl("flashattn2")``.
 
 For more details, see :meth:`~nunchaku.models.transformers.transformer_flux.NunchakuFluxTransformer2dModel.set_attention_impl`.

nunchaku/models/transformers/transformer_flux.py

@@ -654,7 +654,7 @@ class NunchakuFluxTransformer2dModel(FluxTransformer2DModel, NunchakuModelLoader
         impl : str
             Attention implementation to use. Supported values:
 
-            - ``"flash-attention2"`` (default): Standard FlashAttention-2.
+            - ``"flashattn2"`` (default): Standard FlashAttention-2.
             - ``"nunchaku-fp16"``: Uses FP16 attention accumulation, up to 1.2× faster than FlashAttention-2 on NVIDIA 30-, 40-, and 50-series GPUs.
         """
         block = self.transformer_blocks[0]