训练Qwen3.5-4B时报错了

[lbl1120]

Reminder

• I have read the above rules and searched the existing issues.

System Info

• llamafactory version: 0.9.6.dev0
• Platform: Linux-5.15.0-139-generic-x86_64-with-glibc2.31
• Python version: 3.12.13
• PyTorch version: 2.11.0+cu130 (GPU)
• Transformers version: 5.6.0
• Datasets version: 4.0.0
• Accelerate version: 1.11.0
• PEFT version: 0.18.1
• GPU type: NVIDIA GeForce RTX 4090
• GPU number: 2
• GPU memory: 47.37GB
• TRL version: 0.24.0
• vLLM version: 0.21.0
• Git commit: 7a1e963
• Default data directory: detected

Reproduction

参数选择：
模型名称：Qwen3.5-4B-Thinking
模型路径：/data/AI/model/chat/Qwen/Qwen3___5-4B
对话模板：qwen3_5_nothink
训练数据集：identity
错误信息如下：
extern "C"
launch_bounds(512, 4)
global void reduction_prod_kernel(ReduceJitOp r){
r.run();
}
nvrtc: error: failed to open libnvrtc-builtins.so.13.0.
Make sure that libnvrtc-builtins.so.13.0 is installed correctly.

Others

No response

[chilljudaoren]

qwen3.5架构太新了