Datasets:

arm-bench
/

arm-bench-trace

Languages:

Size:

Tags:

License:

Dataset card Files Files and versions

arm-bench-trace / definitions /conv2d

25.5 kB

3 contributors

History: 4 commits

ArtificialRay

Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision

009b323 about 1 month ago

conv2d_fp32_kh1_kw1_sh1_sw1_dh1_dw1_p0.json

2.11 kB
add baseline solution source at all neural network kernel definitions about 1 month ago
conv2d_fp32_kh1_kw1_sh2_sw2_dh1_dw1_p0.json

2.11 kB
add baseline solution source at all neural network kernel definitions about 1 month ago
conv2d_fp32_kh3_kw3_sh1_sw1_dh1_dw1_p1.json

2.11 kB
add baseline solution source at all neural network kernel definitions about 1 month ago
conv2d_fp32_kh3_kw3_sh2_sw2_dh1_dw1_p1.json

2.11 kB
add baseline solution source at all neural network kernel definitions about 1 month ago
conv2d_fp32_kh7_kw7_sh2_sw2_dh1_dw1_p3.json

2.11 kB
add baseline solution source at all neural network kernel definitions about 1 month ago
conv2d_w8a8ch_kh1_kw1_sh1_sw1_dh1_dw1_p0.json

2.99 kB
Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision about 1 month ago
conv2d_w8a8ch_kh1_kw1_sh2_sw2_dh1_dw1_p0.json

2.99 kB
Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision about 1 month ago
conv2d_w8a8ch_kh3_kw3_sh1_sw1_dh1_dw1_p1.json

2.99 kB
Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision about 1 month ago
conv2d_w8a8ch_kh3_kw3_sh2_sw2_dh1_dw1_p1.json

2.99 kB
Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision about 1 month ago
conv2d_w8a8ch_kh7_kw7_sh2_sw2_dh1_dw1_p3.json

2.99 kB
Update all w8a8ch definitions' correctness reference, aligns with what llama.cpp low-bit kernel correctness reference style: dequant input, then calculate in full precision about 1 month ago