diff --git a/studio/backend/core/inference/diffusion_auto_policy.py b/studio/backend/core/inference/diffusion_auto_policy.py index 77df14464c..254ccbafc5 100644 --- a/studio/backend/core/inference/diffusion_auto_policy.py +++ b/studio/backend/core/inference/diffusion_auto_policy.py @@ -189,6 +189,7 @@ def resolve_dense_quant_candidate( prequant_available = False try: from .diffusion_prequant import local_prequant_path_ready, resolve_prequant_source + src = resolve_prequant_source(fam, scheme, path_override = prequant_path) # A request-supplied local path override is only usable if the loader will accept it # (allowlisted AND present); otherwise load_prequantized_transformer refuses it and diff --git a/studio/backend/core/inference/diffusion_cache.py b/studio/backend/core/inference/diffusion_cache.py index 05c781a30f..ad5eb50127 100644 --- a/studio/backend/core/inference/diffusion_cache.py +++ b/studio/backend/core/inference/diffusion_cache.py @@ -125,7 +125,9 @@ def apply_step_cache( # reuse the CacheMixin FluxTransformer2DModel yet their __call__ opens no cache_context, # so engaging FBCache there would crash every default generation -- run uncached instead. if not _pipeline_opens_cache_context(pipe): - _warn(logger, mode, RuntimeError("pipeline __call__ opens no cache_context; running uncached")) + _warn( + logger, mode, RuntimeError("pipeline __call__ opens no cache_context; running uncached") + ) return None try: try: