ollama/llama/llama-hparams.cpp

/**
 * llama.cpp - commit 46e3556e01b824e52395fb050b29804b6cff2a7c - do not edit this file
 *
 * MIT License
 *
 * Copyright (c) 2023-2024 The ggml authors
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include "llama-hparams.h"

#include "ggml.h"

#include <algorithm>

uint32_t llama_hparams::n_head(uint32_t il) const {
    if (il < n_layer) {
        return n_head_arr[il];
    }

    GGML_ABORT("fatal error");
}

uint32_t llama_hparams::n_head_kv(uint32_t il) const {
    if (il < n_layer) {
        return n_head_kv_arr[il];
    }

    GGML_ABORT("fatal error");
}

uint32_t llama_hparams::n_ff(uint32_t il) const {
    if (il < n_layer) {
        return n_ff_arr[il];
    }

    GGML_ABORT("fatal error");
}

uint32_t llama_hparams::n_gqa(uint32_t il) const {
    const uint32_t n_head    = this->n_head(il);
    const uint32_t n_head_kv = this->n_head_kv(il);

    if (n_head_kv == 0) {
        return 0;
    }

    return n_head/n_head_kv;
}

uint32_t llama_hparams::n_embd_k_gqa(uint32_t il) const {
    const uint32_t n_head_kv = this->n_head_kv(il);

    return n_embd_head_k * n_head_kv;
}

uint32_t llama_hparams::n_embd_v_gqa(uint32_t il) const {
    const uint32_t n_head_kv = this->n_head_kv(il);

    return n_embd_head_v * n_head_kv;
}

uint32_t llama_hparams::n_embd_k_s() const {
    if (wkv_head_size != 0) {
        // for RWKV models
        return 2 * n_embd;
    }

    // TODO: maybe support other convolution strides than 1
    // NOTE: since the first column of the conv_state is shifted out each time, it's not actually needed
    return (ssm_d_conv > 0 ? ssm_d_conv - 1 : 0) * ssm_d_inner;
}

uint32_t llama_hparams::n_embd_v_s() const {
    if (wkv_head_size != 0) {
        // corresponds to RWKV's wkv_states size
        return n_embd * wkv_head_size;
    }

    // corresponds to Mamba's ssm_states size
    return ssm_d_state * ssm_d_inner;
}

bool llama_hparams::n_bskcn(uint32_t n, uint32_t il) const {
    if (il < n_layer) {
        return n_bskcn_arr[n][il] > 0;
    }

    GGML_ABORT("fatal error");
}

bool llama_hparams::cross_attention_layers(uint32_t il) const {
    return std::find(cross_attn_layers.begin(), cross_attn_layers.end(), il) != cross_attn_layers.end();
}