Function llama_sampling_sample

common/sampling.cpp:102–216

}

llama_token llama_sampling_sample(
                  struct llama_sampling_context * ctx_sampling,
                  struct llama_context * ctx_main,
                  struct llama_context * ctx_cfg,
                  const int idx) {
    const llama_sampling_params & params = ctx_sampling->params;

    const int n_vocab = llama_n_vocab(llama_get_model(ctx_main));

    const float   temp            = params.temp;
    const int32_t top_k           = params.top_k <= 0 ? n_vocab : params.top_k;
    const float   top_p           = params.top_p;
    const float   min_p           = params.min_p;
    const float   tfs_z           = params.tfs_z;
    const float   typical_p       = params.typical_p;
    const int32_t penalty_last_n  = params.penalty_last_n < 0 ? params.n_prev : params.penalty_last_n;
    const float   penalty_repeat  = params.penalty_repeat;
    const float   penalty_freq    = params.penalty_freq;
    const float   penalty_present = params.penalty_present;
    const int     mirostat        = params.mirostat;
    const float   mirostat_tau    = params.mirostat_tau;
    const float   mirostat_eta    = params.mirostat_eta;
    const bool    penalize_nl     = params.penalize_nl;

    auto & prev = ctx_sampling->prev;
    auto & cur  = ctx_sampling->cur;

    llama_token id = 0;

    float * logits = llama_get_logits_ith(ctx_main, idx);

    // apply params.logit_bias map
    for (auto it = params.logit_bias.begin(); it != params.logit_bias.end(); it++) {
        logits[it->first] += it->second;
    }

    cur.clear();

    for (llama_token token_id = 0; token_id < n_vocab; token_id++) {
        cur.emplace_back(llama_token_data{token_id, logits[token_id], 0.0f});
    }

    llama_token_data_array cur_p = { cur.data(), cur.size(), false };

    if (ctx_cfg) {
        llama_sample_classifier_free_guidance(ctx_main, &cur_p, ctx_cfg, params.cfg_scale);
    }

    // apply penalties
    if (!prev.empty()) {
        const float nl_logit = logits[llama_token_nl(llama_get_model(ctx_main))];

        llama_sample_repetition_penalties(ctx_main, &cur_p,
                prev.data() + prev.size() - penalty_last_n,
                penalty_last_n, penalty_repeat, penalty_freq, penalty_present);

        if (!penalize_nl) {
            for (size_t idx = 0; idx < cur_p.size; idx++) {