Closed mtsokol closed 4 months ago
Hi @willow-ahrens,
I think reduction functions could adhere to promotion rules instead of always returning float data type:
float
julia> using Finch julia> maximum(Tensor(ones(Int8, 4,4))) 1.0 julia> minimum(Tensor(ones(Int8, 4,4))) 1.0
Hi @willow-ahrens,
I think reduction functions could adhere to promotion rules instead of always returning
float
data type: