A system that prompts LLMs to generate and autotune hardware-primitive-level GEMM and convolution kernels, outperforming OpenBLAS and cuBLAS on several platforms.
DeepSeek-V3 Technical Report
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
QiMeng-TensorOp: Automatically Generating High-Performance Tensor Operators with Hardware Primitives
A system that prompts LLMs to generate and autotune hardware-primitive-level GEMM and convolution kernels, outperforming OpenBLAS and cuBLAS on several platforms.