Struct TrainingTaskSpec 

pub struct TrainingTaskSpec {

    pub task_id: String,
    pub sponsor_did: String,
    pub sponsor_address: Address,
    pub architecture: ArchitectureSpec,
    pub tier: TrainingTier,
    pub aggregation: AggregationRule,
    pub trainer_count: u32,
    pub quorum: u32,
    pub inner_steps: u32,
    pub max_rounds: u32,
    pub grace_window_ms: u64,
    pub reward_pool: u128,
    pub dataset_ref: String,
    pub dataset_hash: Hash,
    pub min_throughput: Option<u64>,
    pub created_at: Timestamp,
    pub metadata: HashMap<String, Value>,
}

The on-chain description of a training run. Posted by the sponsor, referenced by trainers, and committed verbatim into the final TrainingReceipt.

Fields

task_id: String

Unique task ID (deterministic hash of canonicalized task fields).

sponsor_did: String

Sponsor DID (did:tenzro:human:... or did:tenzro:machine:...).

sponsor_address: Address

Sponsor’s address for any unspent escrow refund.

architecture: ArchitectureSpec

Architecture being trained.

tier: TrainingTier

Trust tier required for trainer enrollment.

aggregation: AggregationRule

Aggregation rule the syncer must apply.

trainer_count: u32

Number of trainers (M). Total slots.

quorum: u32

Quorum (K). Outer gradient is accepted once K of M have submitted for the same fragment.

inner_steps: u32

Inner SGD steps between outer gradient submissions (H).

max_rounds: u32

Total outer rounds the run will execute.

grace_window_ms: u64

Adaptive grace window τ in milliseconds. Stragglers are absorbed up to this delay; their submissions count if they land within τ.

reward_pool: u128

Total reward pool escrowed by sponsor (TNZO, in attoTNZO).

dataset_ref: String

Reference to the dataset. Format depends on tier:

• Open / Verified (public): ipfs://..., ar://..., https://...
• Verified (encrypted-at-rest): enc:... (key sealed to TEE)
• Confidential (TEE-resident): tee://... (data never leaves owner)

dataset_hash: Hash

Hash of the dataset’s manifest (root of shard hash tree). Bound into the receipt for provenance.

min_throughput: Option<u64>

Optional minimum throughput requirement (samples/sec) for trainer enrollment. Syncer enforces at enrollment.

created_at: Timestamp

Posting timestamp.

metadata: HashMap<String, Value>

Free-form metadata (eval suite, target perplexity, etc.).

Trait Implementations

impl Clone for TrainingTaskSpec

fn clone(&self) -> TrainingTaskSpec

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TrainingTaskSpec

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TrainingTaskSpec

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for TrainingTaskSpec

fn eq(&self, other: &TrainingTaskSpec) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for TrainingTaskSpec

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

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for TrainingTaskSpec