remove torch.save from remnant code.

2025-11-26 11:50:56 +05:30
4 changed files with 3 additions and 10 deletions
@@ -29,7 +29,7 @@ Cache methods speedup diffusion transformers by storing and reusing intermediate

 [[autodoc]] apply_faster_cache

-## FirstBlockCacheConfig
+### FirstBlockCacheConfig

 [[autodoc]] FirstBlockCacheConfig

@@ -66,8 +66,4 @@ config = FasterCacheConfig(
    tensor_format="BFCHW",
 )
 pipeline.transformer.enable_cache(config)
-```
-
-## FirstBlockCache
-
-[FirstBlock Cache](https://huggingface.co/docs/diffusers/main/en/api/cache#diffusers.FirstBlockCacheConfig) builds on the ideas of [TeaCache](https://huggingface.co/papers/2411.19108). It is much simpler to implement generically for a wide range of models and has been integrated first for experimental purposes.
+```
@@ -41,11 +41,9 @@ class CacheMixin:
        Enable caching techniques on the model.

        Args:
-            config (`Union[PyramidAttentionBroadcastConfig, FasterCacheConfig, FirstBlockCacheConfig]`):
+            config (`Union[PyramidAttentionBroadcastConfig]`):
                The configuration for applying the caching technique. Currently supported caching techniques are:
                    - [`~hooks.PyramidAttentionBroadcastConfig`]
-                    - [`~hooks.FasterCacheConfig`]
-                    - [`~hooks.FirstBlockCacheConfig`]

        Example:

@@ -861,7 +861,6 @@ class Flux2Pipeline(DiffusionPipeline, Flux2LoraLoaderMixin):
        if output_type == "latent":
            image = latents
        else:
-            torch.save({"pred": latents}, "pred_d.pt")
            latents = self._unpack_latents_with_ids(latents, latent_ids)

            latents_bn_mean = self.vae.bn.running_mean.view(1, -1, 1, 1).to(latents.device, latents.dtype)