两张图的时候，报tensors size不对，是对图片有什么要求吗

[liujian19911023]

[01-16 08:17:25] Sampling params:
width: 1024
height: 1024
num_frames: 1
prompt: 两个人一起做个嘘的手势
neg_prompt: Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards
seed: 1024
infer_steps: 50
num_outputs_per_prompt: 1
guidance_scale: 1.5
embedded_guidance_scale: 6.0
n_tokens: None
flow_shift: None
image_path: ['outputs/uploads/9378d2a2-b159-4da3-9e73-268ab4cab660_0_input_0.png', 'outputs/uploads/9378d2a2-b159-4da3-9e73-268ab4cab660_1_input_1.png']
save_output: True
output_file_path: outputs/9378d2a2-b159-4da3-9e73-268ab4cab660.jpg

[01-16 08:17:25] Running pipeline stages: ['GlmImageBeforeDenoisingStage', 'denoising_stage', 'decoding_stage']
[01-16 08:17:25] [GlmImageBeforeDenoisingStage] started...
[01-16 08:18:12] generate_prior_tokens time: 47.13891243934631
[01-16 08:18:14] [GlmImageBeforeDenoisingStage] Error during execution after 49159.4292 ms: Sizes of tensors must match except in dimension 2. Expected size 7040 but got size 1643 for tensor number 1 in the list.
Traceback (most recent call last):
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/pipelines_core/stages/base.py", line 200, in call
result = self.forward(batch, server_args)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 120, in decorate_context
return func(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/models/model_stages/glm_image.py", line 749, in forward
_ = self.transformer(
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/models/dits/glm_image.py", line 795, in forward
hidden_states, encoder_hidden_states = block(
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/models/dits/glm_image.py", line 503, in forward
attn_hidden_states, attn_encoder_hidden_states = self.attn1(
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
return forward_call(*args, **kwargs)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/models/dits/glm_image.py", line 395, in forward
kv_cache.store(key, value)
File "/home/robinx/miniconda3/envs/image/lib/python3.10/site-packages/sglang/multimodal_gen/runtime/models/dits/glm_image.py", line 51, in store
self.k_cache = torch.cat([self.k_cache, k], dim=2)
RuntimeError: Sizes of tensors must match except in dimension 2. Expected size 7040 but got size 1643 for tensor number 1 in the list.