ternlang-core 0.3.3

Compiler and VM for Ternlang — balanced ternary language with affirm/tend/reject trit semantics, @sparseskip codegen, and BET bytecode execution.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

// Module:  stdlib/data/batching.tern
// Purpose: Data Batching
// Author:  RFI-IRFOS
// Ref:     https://ternlang.com

struct TritBatch {
    size: int,
    data: trittensor<4 x 4>
}

fn collate_trit(samples: trittensor<4 x 1>[]) -> TritBatch {
    let dummy_data: trittensor<4 x 4> = {
        [affirm, tend, reject, tend],
        [tend, affirm, tend, reject],
        [reject, tend, affirm, tend],
        [tend, reject, tend, affirm]
    };
    let batch: TritBatch = { size: 4, data: dummy_data };
    return batch;
}

fn pad_sequence_trit(seq: trit[], max_len: int) -> trit[] {
    // Padding is always 'tend'
    return seq;
}

fn dynamic_batch_trit(samples: trittensor<4 x 1>[]) -> TritBatch {
    return collate_trit(samples);
}