Instructions to use abenzerps/Qwen-Image-2.1-Turbo-Quantized with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use abenzerps/Qwen-Image-2.1-Turbo-Quantized with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download abenzerps/Qwen-Image-2.1-Turbo-Quantized --local-dir Qwen-Image-2.1-Turbo-Quantized
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
A different Error with KSampler: "Expected weight to be of same shape as normalized_shape, but got weight of shape [136] and normalized_shape = [128]"
Using leejet GGUF, loading seems to be fine. However, I am getting this error (using the official i2i workflow with just the loading of Unet Load GGUF change:
Expected weight to be of same shape as normalized_shape, but got weight of shape [136] and normalized_shape = [128]
Sorry, this is the first ever bug report I am doing; totally a noob here. Hopefully I have attached enough info.
Appreciate your effort to make this happen.
=== Error Log ===
Node threw an error during execution.
ComfyUI Error Report
Error Details
- Node ID: 459:458
- Node Type: KSampler
- Exception Type: RuntimeError
- Exception Message: RuntimeError: Expected weight to be of same shape as normalized_shape, but got weight of shape [136] and normalized_shape = [128]
Stack Trace
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\execution.py", line 547, in execute
output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\execution.py", line 352, in get_output_data
return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\execution.py", line 326, in _async_map_node_over_list
await process_inputs(input_dict, i)
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\execution.py", line 314, in process_inputs
result = f(**inputs)
^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\nodes.py", line 1624, in sample
return common_ksampler(model, seed, steps, cfg, sampler_name, scheduler, positive, negative, latent_image, denoise=denoise)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\nodes.py", line 1588, in common_ksampler
samples = comfy.sample.sample(model, noise, steps, cfg, sampler_name, scheduler, positive, negative, latent_image,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\sample.py", line 82, in sample
samples = sampler.sample(noise, positive, negative, cfg=cfg, latent_image=latent_image, start_step=start_step, last_step=last_step, force_full_denoise=force_full_denoise, denoise_mask=noise_mask, sigmas=sigmas, callback=callback, disable_pbar=disable_pbar, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1463, in sample
return sample(self.model, noise, positive, negative, cfg, self.device, sampler, sigmas, self.model_options, latent_image=latent_image, denoise_mask=denoise_mask, callback=callback, disable_pbar=disable_pbar, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1353, in sample
return cfg_guider.sample(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1335, in sample
output = executor.execute(noise, latent_image, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1262, in outer_sample
output = self.inner_sample(noise, latent_image, device, sampler, sigmas, denoise_mask, callback, disable_pbar, seed, latent_shapes=latent_shapes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1237, in inner_sample
samples = executor.execute(self, sigmas, extra_args, callback, noise, latent_image, denoise_mask, disable_pbar)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1005, in sample
samples = self.sampler_function(model_k, noise, sigmas, extra_args=extra_args, callback=k_callback, disable=disable_pbar, **self.extra_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\utils\_contextlib.py", line 124, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\k_diffusion\sampling.py", line 205, in sample_euler
denoised = model(x, sigma_hat * s_in, **extra_args)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 640, in __call__
out = self.inner_model(x, sigma, model_options=model_options, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1208, in __call__
return self.outer_predict_noise(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1215, in outer_predict_noise
).execute(x, timestep, model_options, seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 1218, in predict_noise
return sampling_function(self.inner_model, x, timestep, self.conds.get("negative", None), self.conds.get("positive", None), self.cfg, model_options=model_options, seed=seed)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 620, in sampling_function
out = calc_cond_batch(model, conds, x, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 211, in calc_cond_batch
return _calc_cond_batch_outer(model, conds, x_in, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 219, in _calc_cond_batch_outer
return executor.execute(model, conds, x_in, timestep, model_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\samplers.py", line 335, in _calc_cond_batch
output = model.apply_model(input_x, timestep_, **c).chunk(batch_chunks)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\model_base.py", line 210, in apply_model
return comfy.patcher_extension.WrapperExecutor.new_class_executor(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\model_base.py", line 254, in _apply_model
model_output = self.diffusion_model(xc, t, context=context, control=control, transformer_options=transformer_options, **extra_conds)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1776, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1787, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\ldm\qwen_image21\model.py", line 301, in forward
return comfy.patcher_extension.WrapperExecutor.new_class_executor(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\patcher_extension.py", line 113, in execute
return self.original(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\ldm\qwen_image21\model.py", line 352, in _forward
prefix_states = block(prefix_states, mod, prefix_pe, prefix_attn, prefix_states.shape[1], transformer_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1776, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1787, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\ldm\qwen_image21\model.py", line 145, in forward
x = _gated_residual(x, self.attn(_modulated_norm(self.img_norm1, x, scale1, prefix_len, zero), pe, attn_fn, prefix_len, transformer_options), gate1, prefix_len)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1776, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\modules\module.py", line 1787, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AppData\Local\Comfy-Desktop\ComfyUI-Installs\ComfyUI\ComfyUI\comfy\ldm\qwen_image21\model.py", line 104, in forward
q, k = comfy.quant_ops.ck.rms_rope(q, k, pe, q_scale, k_scale, self.norm_q.eps)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\comfy_kitchen\__init__.py", line 734, in rms_rope
return torch.ops.comfy_kitchen.rms_rope(q, k, freqs_cis, q_scale, k_scale, epsilon)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_ops.py", line 1209, in __call__
return self._op(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_tensor.py", line 1703, in __torch_function__
ret = func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_ops.py", line 1209, in __call__
return self._op(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_library\custom_ops.py", line 347, in backend_impl
result = self._backend_fns[device_type](*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_compile.py", line 54, in inner
return disable_fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_dynamo\eval_frame.py", line 1181, in _fn
return fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\_library\custom_ops.py", line 382, in wrapped_fn
return fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\comfy_kitchen\backends\eager\rope.py", line 240, in _op_rms_rope
return impl(**kwargs)
^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\comfy_kitchen\backends\eager\rope.py", line 124, in rms_rope
_rms_rope1(q, freqs_cis, q_scale, epsilon, split_half=False),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\comfy_kitchen\backends\eager\rope.py", line 85, in _rms_rope1
x_norm = torch.nn.functional.rms_norm(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\minch\AI_Tools\ComfyUI\.venv\Lib\site-packages\torch\nn\functional.py", line 2954, in rms_norm
return torch.rms_norm(input, normalized_shape, weight, eps)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
System Information
- ComfyUI Version: 0.37.0
- Arguments: ComfyUI\main.py --feature-flag show_signin_button=true --feature-flag enable_telemetry=true --base-directory C:\Users\minch\AI_Tools\ComfyUI --user-directory C:\Users\minch\AI_Tools\ComfyUI\user --database-url sqlite:///C:\Users\minch\AI_Tools\ComfyUI\user\comfyui.db --port 8000 --enable-manager --use-sage-attention --extra-model-paths-config C:\Users\minch\AppData\Roaming\Comfy Desktop\instance-model-paths\inst-1786111269309.yaml --input-directory C:\Users\minch\AI_Tools\ComfyUI\input --output-directory C:\Users\minch\AI_Tools\ComfyUI\output
- OS: win32
- Python Version: 3.12.11 (main, Aug 18 2025, 19:17:54) [MSC v.1944 64 bit (AMD64)]
- Embedded Python: false
- PyTorch Version: 2.10.0+cu130
Devices
- Name: cuda:0 NVIDIA GeForce RTX 5070 Ti : cudaMallocAsync
- Type: cuda
- VRAM Total: 17094475776
- VRAM Free: 6761931180
- Torch VRAM Total: 7650410496
- Torch VRAM Free: 39510444
BTW, another author's GGUF (Abiray/Qwen-Image-2.1-GGUF) is working fine. I am hoping to be able to try your version first, though.
Same exact problem. also noob here.
let me check
same issue here. RTX 3090 ti
If it helps, I was getting the error with the qwen-image-2.1-Q8_0.gguf file. When I switched to the qwen-image-2.1-Q4_K_M.gguf file, it worked.
confirmed. q4_k_m works fine
Thanks for testing and confirming, the issue is specifically with the Q8_0 file.
During conversion, the 1D norm weights (norm_q/norm_k) got quantized into Q8 blocks, turning the 128 elements into 136 bytes
4Γ34), which triggers the shape mismatch error in ComfyUI. The other quants (Q4_K_M, Q5_K_M, and Q6_K) kept those in BF16, which is why they work completely fine.
For now, please stick with Q4_K_M, Q5_K_M, or Q6_K. Iβll fix and re-upload the Q8_0 file with proper norm weights as soon as possible.
q4km can use,q8 error
Wrote by Claude Opus (Im lazy af)
Why it happens (abenzerps already explained it above, adding the ComfyUI side): Q8_0 stores the tiny 1D norm_q / norm_k weights as Q8 blocks β 128 values become 136 bytes. ComfyUI-GGUF normally dequantizes weights on the fly inside its patched Linear/Norm layers, but the new Qwen-Image 2.1 model code reads self.norm_q.weight directly and hands it to rms_rope, so the raw 136-byte block reaches torch.rms_norm β shape mismatch. The loader already has a rule "1D tensors shouldn't be quantized" that converts 1D BF16 tensors to fp32 at load time; it just needs to apply to any quantized type.
example to fix in ComfyUI:
Open ComfyUI/custom_nodes/ComfyUI-GGUF/loader.py
Find this line (around line 142):
if len(shape) <= 1 and tensor.tensor_type == gguf.GGMLQuantizationType.BF16:
Replace it with:
if len(shape) <= 1 and tensor.tensor_type not in {gguf.GGMLQuantizationType.F32, gguf.GGMLQuantizationType.F16}:
Fully restart ComfyUI (not just refresh the browser β the loader is only read at startup). If the model was already loaded, also click Unload models or restart, so the cached bad weights are dropped.
RuntimeError: Given normalized_shape=[8192], expected input with shape [*8192], but got input of size[1, 35, 4096]
used both q4km and q6k
rtx 3060 12gb
RuntimeError: Given normalized_shape=[8192], expected input with shape [*8192], but got input of size[1, 35, 4096]
used both q4km and q6k
rtx 3060 12gb
let me check
RuntimeError: Given normalized_shape=[8192], expected input with shape [*8192], but got input of size[1, 35, 4096]
used both q4km and q6k
rtx 3060 12gb
that happens because your ComfyUI-GGUF node is on an older commit that doesn't unpack 1D BF16 weights at load time.
The text norm weight (txt_in.text_norm.weight) has 4,096 values stored as BF16 (2 bytes each = 8,192 raw bytes). During inference, PyTorch runs weight + 1.0 and accidentally takes the raw 8,192-byte storage instead of 4,096 floats, which mismatches your text encoder output [1, 35, 4096].
How to fix it:
- Update the node: Run
git pullinsideComfyUI/custom_nodes/ComfyUI-GGUF(or update it via ComfyUI Manager).
Or edit manually inComfyUI/custom_nodes/ComfyUI-GGUF/loader.pyaround line 142 to make sure 1D tensors get dequantized:if len(shape) <= 1 and tensor.tensor_type not in {gguf.GGMLQuantizationType.F32, gguf.GGMLQuantizationType.F16}: state_dict[sd_key] = dequantize_tensor(state_dict[sd_key], dtype=torch.float32) - Fully restart ComfyUI: Close the terminal/console completely and start ComfyUI again (just refreshing the browser won't reload Python code or clear cached weights).
After restarting, both Q4/Q6 and Q8 will work without errors.
After applying the solution, please let me know if there are any issues..