AttributeError: 'NoneType' object has no attribute 'blocks' when running Cache-DiT example with Wan2.2 model

Question

I’m trying to use Cache-DiT to accelerate inference for the Wan2.2 model. However, when I run the example script,

python run_wan_2.2_i2v.py --steps 28 --cache

I get the following error.

Namespace(cache=True, compile=False, fuse_lora=False, quantize=False, quantize_type='fp8_w8a8_dq', steps=28, Fn=8, Bn=0, rdt=0.08, max_warmup_steps=8, max_cached_steps=-1, max_continuous_cached_steps=-1, taylorseer=False, taylorseer_order=1, height=None, width=None, parallel_type=None)
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████| 5/5 [00:05<00:00,  1.05s/it]
Loading pipeline components...:  60%|█████████████████████████████████▌                      | 3/5 [00:05<00:03,  1.55s/it]The config attributes {'clip_output': False} were passed to AutoencoderKLWan, but are not expected and will be ignored. Please verify your config.json configuration file.
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 10.46it/s]
Loading pipeline components...: 100%|████████████████████████████████████████████████████████| 5/5 [00:07<00:00,  1.46s/it]
Enable model cpu offload for low memory device.
Traceback (most recent call last):
  File "/home/ella/1141/cache-dit/examples/pipeline/run_wan_2.2_i2v.py", line 61, in <module>
    pipe.transformer_2.blocks,
AttributeError: 'NoneType' object has no attribute 'blocks'

I set the model via environment variable:

export WAN_MODEL_NAME="Wan-AI/Wan2.2-TI2V-5B-Diffusers"

Then I ran the example script directly from: examples/pipeline/run_wan_2.2_i2v.py

What I’ve Tried

Running without --cache → works fine
Switching to Wan2.2-T2V-14B → works fine

What does this error mean? Does it indicate that my model isn’t compatible with Cache-DiT, or is there something wrong with the example script?

hongxigoo · Accepted Answer · 2025-10-27 09:23:26Z

The error occurs because the current run_wan_2.2_i2v.py example in Cache-DiT assumes that the loaded pipeline contains two Transformer modules — pipe.transformer and pipe.transformer_2.

However, the Wan2.2-TI2V-5B model only includes a single transformer, so pipe.transformer_2 is None. When the script tries to run:

pipe.transformer_2.blocks,

it fails with:

AttributeError: 'NoneType' object has no attribute 'blocks'

Root Cause :

14B (T2V) models → dual-transformer architecture (text + video).
5B (TI2V) models → single-transformer architecture. The Cache-DiT example was written for models with two transformers. To make the example work with the 5B model, you can modify the script to handle both cases dynamically. Open examples/pipeline/run_wan_2.2_i2v.py and find the section where Cache-DiT is initialized (around line 61). Replace this part:

pipe.transformer_2.blocks,

with this conditional logic:

if getattr(pipe, "transformer_2", None) is not None:
    cache_dit = CacheDiT([pipe.transformer.blocks, pipe.transformer_2.blocks])
else:
    cache_dit = CacheDiT([pipe.transformer.blocks])

This ensures that Cache-DiT initializes correctly whether the model has one or two transformer modules.

Collectives™ on Stack Overflow

AttributeError: 'NoneType' object has no attribute 'blocks' when running Cache-DiT example with Wan2.2 model

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related