Function evaluate

lora/lora.py:225–240

(model, dataset, loss, tokenizer, batch_size, num_batches)

def evaluate(model, dataset, loss, tokenizer, batch_size, num_batches):
    all_losses = []
    ntokens = 0

    # num_batches can be -1 to indicate the entire set
    index_iterator = iter(range(num_batches)) if num_batches != -1 else iter(int, 1)

    for it, batch in zip(
        index_iterator,
        iterate_batches(dataset, tokenizer, batch_size),
    ):
        losses, toks = loss(model, *batch)
        all_losses.append((losses * toks).item())
        ntokens += toks.item()

    return np.sum(all_losses) / ntokens


def train(model, train_set, val_set, optimizer, loss, tokenizer, args):