CPU->GPU Memcpy failed when finetuning with STS-B

[xavinatalia]

When Finetuning XLNet-large with STS-B using 4 3090 with cuda version 11.7, I got following message:
2024-04-26 23:17:09.192214: E tensorflow/stream_executor/cuda/cuda_blas.cc:698] failed to run cuBLAS routine cublasGemmBatchedEx: CUBLAS_STATUS_EXECUTION_FAILED
2024-04-26 23:17:09.192279: E tensorflow/stream_executor/cuda/cuda_blas.cc:2620] Internal: failed BLAS call, see log for details
2024-04-26 23:17:12.114126: E tensorflow/stream_executor/cuda/cuda_blas.cc:698] failed to run cuBLAS routine cublasGemmBatchedEx: CUBLAS_STATUS_EXECUTION_FAILED
2024-04-26 23:17:12.114166: E tensorflow/stream_executor/cuda/cuda_blas.cc:2620] Internal: failed BLAS call, see log for details
2024-04-26 23:17:12.114698: E tensorflow/stream_executor/cuda/cuda_blas.cc:698] failed to run cuBLAS routine cublasGemmBatchedEx: CUBLAS_STATUS_EXECUTION_FAILED
2024-04-26 23:17:12.114761: E tensorflow/stream_executor/cuda/cuda_blas.cc:2620] Internal: failed BLAS call, see log for details
2024-04-26 23:17:12.121236: E tensorflow/stream_executor/cuda/cuda_blas.cc:698] failed to run cuBLAS routine cublasGemmBatchedEx: CUBLAS_STATUS_EXECUTION_FAILED
2024-04-26 23:17:12.121270: E tensorflow/stream_executor/cuda/cuda_blas.cc:2620] Internal: failed BLAS call, see log for details
2024-04-26 23:17:12.659520: I tensorflow/stream_executor/stream.cc:2079] [stream=0x4854880,impl=0x46ed360] did not wait for [stream=0x4854590,impl=0x46ed390]
2024-04-26 23:17:12.659567: I tensorflow/stream_executor/stream.cc:5027] [stream=0x4854880,impl=0x46ed360] did not memcpy host-to-device; source: 0x7f13e001b300
2024-04-26 23:17:12.659624: F tensorflow/core/common_runtime/gpu/gpu_util.cc:339] CPU->GPU Memcpy failed
已放弃 (核心已转储)