You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
when using:
TheBloke/Mixtral-8x7B-Instruct-v0.1-AWQ
or the more recent and recommended:
casperhansen/mixtral-instruct-awq
with the training script provided in the examples, I get a runtime exception:
Exception has occurred: RuntimeError
CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling cublasGemmEx( handle, opa, opb, m, n, k, &falpha, a, CUDA_R_16F, lda, b, CUDA_R_16F, ldb, &fbeta, c, CUDA_R_16F, ldc, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
File "/home/christian.pala/models-zoo/models_zoo/main.py", line 5, in main cli.main(prog="models-zoo") File "/home/christian.pala/models-zoo/models_zoo/main.py", line 9, in main()RuntimeError: CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling cublasGemmEx( handle, opa, opb, m, n, k, &falpha, a, CUDA_R_16F, lda, b, CUDA_R_16F, ldb, &fbeta, c, CUDA_R_16F, ldc, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
Is adapter training not supported for mixtral?
The text was updated successfully, but these errors were encountered:
Hey folks,
when using:
TheBloke/Mixtral-8x7B-Instruct-v0.1-AWQ
or the more recent and recommended:
casperhansen/mixtral-instruct-awq
with the training script provided in the examples, I get a runtime exception:
Exception has occurred: RuntimeError
CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling
cublasGemmEx( handle, opa, opb, m, n, k, &falpha, a, CUDA_R_16F, lda, b, CUDA_R_16F, ldb, &fbeta, c, CUDA_R_16F, ldc, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
File "/home/christian.pala/models-zoo/models_zoo/main.py", line 5, in main cli.main(prog="models-zoo") File "/home/christian.pala/models-zoo/models_zoo/main.py", line 9, in main()RuntimeError: CUDA error: CUBLAS_STATUS_NOT_SUPPORTED when calling
cublasGemmEx( handle, opa, opb, m, n, k, &falpha, a, CUDA_R_16F, lda, b, CUDA_R_16F, ldb, &fbeta, c, CUDA_R_16F, ldc, CUDA_R_32F, CUBLAS_GEMM_DEFAULT_TENSOR_OP)
Is adapter training not supported for mixtral?
The text was updated successfully, but these errors were encountered: