invoke-ai/InvokeAI · error · TypeError

Expected PidNet for PiD decoder, got {type(pid_net).__name__

Error message

Expected PidNet for PiD decoder, got {type(pid_net).__name__}.

What it means

The PiD decoder network loaded for FLUX.2 decode must be an instance of the PidNet class. Any other class means the model record under the PiD model identifier is not a PiD net — corrupted, mislabeled, or an incompatible VAE/diffusion model.

Source

Thrown at invokeai/app/invocations/flux2_pid_decode.py:216

        # fit the decode on its own, so we deliberately do NOT evict every other model here.
        context.models.offload_from_vram(self.gemma2_encoder.text_encoder)
        TorchDevice.empty_cache()

        # 4) Run PiD decode (the loader already returns a live PidNet).
        pid_info = context.models.load(self.pid_decoder.decoder)
        # The working-memory estimate scales with the OUTPUT pixel count, so it must see the PACKED latent
        # (spatial H/16), not the unpacked one - otherwise it over-reserves by 4x.
        # Read once: the estimate and the decode must agree, or the cache reserves headroom for a
        # peak that will not happen (or too little for one that will).
        pid_memory_optimization = context.config.get().pid_memory_optimization
        estimated_working_memory = estimate_pid_decode_working_memory(
            packed,
            BaseModelType.Flux2,
            pid_memory_optimization,
        )
        with pid_info.model_on_device(working_mem_bytes=estimated_working_memory) as (_, pid_net):
            if not isinstance(pid_net, PidNet):
                raise TypeError(f"Expected PidNet for PiD decoder, got {type(pid_net).__name__}.")
            device = TorchDevice.choose_torch_device()
            dtype = next(iter(pid_net.parameters())).dtype

            # The packed latent is already BN-denormalized (raw VAE-input space); the scalar transform below is
            # identity for current FLUX.2 VAEs and only bites if a VAE ever exposes real scalar constants.
            denorm_latent = packed.to(device=device, dtype=dtype) / scaling_factor + shift_factor
            context.logger.info(
                f"FLUX.2 PiD denorm_latent stats[min={denorm_latent.min().item():.3f} "
                f"max={denorm_latent.max().item():.3f} mean={denorm_latent.mean().item():.3f}] "
                f"using scale={scaling_factor:.4f} shift={shift_factor:.4f}"
            )
            caption_embs = caption_embs.to(device=device, dtype=dtype)

            context.util.signal_progress("Running PiD decoder")
            decoder = PiDDecoder(pid_net, backbone=BaseModelType.Flux2)
            x0 = decoder.decode(
                latent=denorm_latent,
                caption_embs=caption_embs,

View on GitHub (pinned to 0b6a024f2f)

Solutions

  1. Re-download the PiD decoder model through the model manager
  2. Verify the decoder model identifier points at the FLUX.2 PiD net, not another model
  3. Re-import the model with the correct model type so it loads as PidNet
  4. Update InvokeAI if model-class detection changed between versions

Example fix

// before: decoder key points at flux2 VAE
decoder=ModelIdentifierField(key='flux2-vae-001')
// after
decoder=ModelIdentifierField(key='flux2-pid-decoder-001')
Defensive patterns

Strategy: type-guard

Validate before calling

pid_info = context.models.load(pid_decoder)
if type(pid_info.model).__name__ != 'PidNet':
    raise TypeError(f'{pid_decoder.key} is not a PidNet')

Type guard

def is_pid_net(obj) -> bool:
    return isinstance(obj, PidNet)

Try / catch

try:
    result = pid_decode.invoke(context)
except TypeError as e:
    if 'PidNet' in str(e):
        redownload_pid_decoder_model()
    raise

Prevention

When it happens

Trigger: pid_info.model_on_device(working_mem_bytes=...) yields a non-PidNet object in invoke; the node's decoder field references a wrong model key or a corrupted download.

Common situations: PiD decoder files incompletely downloaded; a user-selected model key pointing at the FLUX.2 VAE or another backbone; model-manager records stale after upgrade; wrong model type chosen at import.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/cd299eaeb2a2bd5f. Report an issue: GitHub.