vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fixed incorrect index when going out of context

Browse source
This commit is contained in:
JohannesGaessler 2023-06-13 16:15:11 +02:00
parent 20e76a0764
commit e8528d4d2f
1 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

3
ggml-cuda.cu
View file

@ -1920,7 +1920,8 @@ static void ggml_cuda_op(const ggml_tensor * src0, const ggml_tensor * src1, ggm
cudaMemcpyDeviceToDevice, cudaStream_memcpy_src1));
}
} else if (src1_on_device && !src1_is_contiguous) {
CUDA_CHECK(ggml_cuda_tensor_2d(src1_ddf_i, src1, i03, i02, i01_low, i01_high, cudaStream_main));
GGML_ASSERT(!split);
CUDA_CHECK(ggml_cuda_tensor_2d(src1_ddf_i, src1, i03, i02, 0, ne11, cudaStream_main));
} else {
GGML_ASSERT(false);
}