Struct TensorShardSpec 

#[non_exhaustive]
pub struct TensorShardSpec {
    pub full_shape: Vec<usize>,
    pub shard_dim: usize,
    pub shard_sizes: Vec<usize>,
}

Describes how a single tensor is sharded across ranks.

Marked #[non_exhaustive] so future fields (e.g., dtype, ordering, per-rank ranges for non-uniform shards) can be added without a major version bump. Construct via in-crate helpers (e.g. flat_shard_metadata) and access fields by name.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

full_shape: Vec<usize>

Full (unsharded) shape of the tensor.

shard_dim: usize

Which dimension is split across ranks.

shard_sizes: Vec<usize>

Size along shard_dim for each rank. The sum must equal full_shape[shard_dim].

Trait Implementations

impl Clone for TensorShardSpec

fn clone(&self) -> TensorShardSpec

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TensorShardSpec

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TensorShardSpec

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TensorShardSpec

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.