喵哩个咪

Failed to run on colab

4

I try to compile convnet_cuda in colab (gpu) error: ``` /root/go/src/gorgonia.org/gorgonia/cuda.go:141:23: v.Pointer undefined (type Value has no field or method Pointer) /root/go/src/gorgonia.org/gorgonia/cuda.go:186:48: v.Pointer undefined (type Value has no field or...

可以较小的L值可以生成完整的视频,现在L值太小的话只会生成前几秒

Is it possible to reduce VRAM usage during VAE encode?

4

When there is only 16G VRAM, an OOM error occurs in the VAE encoding stage when using a resolution of 1024.

设备是MI50 使用模型fastllm/Qwen3-Next-80B-A3B-Instruct-int4g-fp16-mixed, 模型可以加载成功 推理时会爆Segmentation fault (core dumped) 异常发生在 ``` Linear(attenInput, weight["model.layers." + std::to_string(i) + ".mlp.shared_expert_gate.weight"], Data(), sharedGate); ``` src\devices\cuda\fastllm-cuda.cu ``` status = cublasGemmEx(fastllmCublasHandle, CUBLAS_OP_T, CUBLAS_OP_N, k, n, m, &h_alpha, (half...