RuntimeError: CUDA error: no kernel image is available for execution on the device

hello,

first of all: thank you for framepack!!

but currently i don't get it run.. i got following error message:

Unloaded CLIPTextModel as complete.
Unloaded DynamicSwap_LlamaModel as complete.
Unloaded CLIPTextModel as complete.
Unloaded SiglipVisionModel as complete.
Unloaded AutoencoderKLHunyuanVideo as complete.
Unloaded DynamicSwap_HunyuanVideoTransformer3DModelPacked as complete.
Unloaded DynamicSwap_LlamaModel as complete.
Unloaded CLIPTextModel as complete.
Unloaded SiglipVisionModel as complete.
Unloaded AutoencoderKLHunyuanVideo as complete.
Unloaded DynamicSwap_HunyuanVideoTransformer3DModelPacked as complete.
Loaded CLIPTextModel to cuda:0 as complete.
Traceback (most recent call last):
File "V:\framepack_cu126_torch26\webui\demo_gradio.py", line 126, in worker
llama_vec, clip_l_pooler = encode_prompt_conds(prompt, text_encoder, text_encoder_2, tokenizer, tokenizer_2)
File "V:\framepack_cu126_torch26\system\python\lib\site-packages\torch\utils\_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "V:\framepack_cu126_torch26\webui\diffusers_helper\hunyuan.py", line 31, in encode_prompt_conds
llama_attention_length = int(llama_attention_mask.sum())
RuntimeError: CUDA error: no kernel image is available for execution on the device
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

my system specs are following:
- motherboard: msi meg x570 unify
- ram: 32gb
- gpu: rtx5070 ti 16gb expert

if you need more, just let me know.

many greetings!

Start at webui/demo_gradio.py:126 and follow encode_prompt_conds in webui/diffusers_helper/hunyuan.py:31, using the reported RTX 5070 Ti environment to reproduce the traceback. Inspect the bundled PyTorch runtime and the CUDA operation that fails; done means the compatibility cause is identified and the reported execution path no longer raises this error.

Written by the indexing model from the issue text.