}
LOG("%.2f minutes\n", total_seconds / 60.0);
}
- LOG("\n");
//LOG_DBG("%s: using tokens %d...%d\n",__func__,params.n_ctx - params.ppl_stride + start, params.n_ctx + start);
for (int j = n_ctx - params.ppl_stride - 1; j < n_ctx - 1; ++j) {
}
LOG("%.2f minutes\n", total_seconds / 60.0);
}
- LOG("\n");
for (int seq = 0; seq < n_seq_batch; seq++) {
const float * all_logits = num_batches > 1 ? logits.data() : llama_get_logits_ith(ctx, seq*n_ctx + first);