flux.hpp

#ifndef __FLUX_HPP__
#define __FLUX_HPP__

#include <vector>

#include "ggml_extend.hpp"
#include "model.h"

#define FLUX_GRAPH_SIZE 10240

namespace Flux {

    struct MLPEmbedder : public UnaryBlock {
    public:
        MLPEmbedder(int64_t in_dim, int64_t hidden_dim) {
            blocks["in_layer"]  = std::shared_ptr<GGMLBlock>(new Linear(in_dim, hidden_dim, true));
            blocks["out_layer"] = std::shared_ptr<GGMLBlock>(new Linear(hidden_dim, hidden_dim, true));
        }

        struct ggml_tensor* forward(struct ggml_context* ctx, struct ggml_tensor* x) {
            // x: [..., in_dim]
            // return: [..., hidden_dim]
            auto in_layer  = std::dynamic_pointer_cast<Linear>(blocks["in_layer"]);
            auto out_layer = std::dynamic_pointer_cast<Linear>(blocks["out_layer"]);

            x = in_layer->forward(ctx, x);
            x = ggml_silu_inplace(ctx, x);
            x = out_layer->forward(ctx, x);
            return x;
        }
    };

    class RMSNorm : public UnaryBlock {
    protected:
        int64_t hidden_size;
        float eps;

        void init_params(struct ggml_context* ctx, ggml_type wtype) {
            params["scale"] = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, hidden_size);
        }

    public:
        RMSNorm(int64_t hidden_size,
                float eps = 1e-06f)
            : hidden_size(hidden_size),
              eps(eps) {}

        struct ggml_tensor* forward(struct ggml_context* ctx, struct ggml_tensor* x) {
            struct ggml_tensor* w = params["scale"];
            x                     = ggml_rms_norm(ctx, x, eps);
            x                     = ggml_mul(ctx, x, w);
            return x;
        }
    };

    struct QKNorm : public GGMLBlock {
    public:
        QKNorm(int64_t dim) {
            blocks["query_norm"] = std::shared_ptr<GGMLBlock>(new RMSNorm(dim));
            blocks["key_norm"]   = std::shared_ptr<GGMLBlock>(new RMSNorm(dim));
        }

        struct ggml_tensor* query_norm(struct ggml_context* ctx, struct ggml_tensor* x) {
            // x: [..., dim]
            // return: [..., dim]
            auto norm = std::dynamic_pointer_cast<RMSNorm>(blocks["query_norm"]);

            x = norm->forward(ctx, x);
            return x;
        }

        struct ggml_tensor* key_norm(struct ggml_context* ctx, struct ggml_tensor* x) {
            // x: [..., dim]
            // return: [..., dim]
            auto norm = std::dynamic_pointer_cast<RMSNorm>(blocks["key_norm"]);

            x = norm->forward(ctx, x);
            return x;
        }
    };

    __STATIC_INLINE__ struct ggml_tensor* apply_rope(struct ggml_context* ctx,
                                                     struct ggml_tensor* x,
                                                     struct ggml_tensor* pe) {
        // x: [N, L, n_head, d_head]
        // pe: [L, d_head/2, 2, 2]
        int64_t d_head = x->ne[0];
        int64_t n_head = x->ne[1];
        int64_t L      = x->ne[2];
        int64_t N      = x->ne[3];
        x              = ggml_cont(ctx, ggml_permute(ctx, x, 0, 2, 1, 3));       // [N, n_head, L, d_head]
        x              = ggml_reshape_4d(ctx, x, 2, d_head / 2, L, n_head * N);  // [N * n_head, L, d_head/2, 2]
        x              = ggml_cont(ctx, ggml_permute(ctx, x, 3, 0, 1, 2));       // [2, N * n_head, L, d_head/2]

        int64_t offset = x->nb[2] * x->ne[2];
        auto x_0       = ggml_view_3d(ctx, x, x->ne[0], x->ne[1], x->ne[2], x->nb[1], x->nb[2], offset * 0);  // [N * n_head, L, d_head/2]
        auto x_1       = ggml_view_3d(ctx, x, x->ne[0], x->ne[1], x->ne[2], x->nb[1], x->nb[2], offset * 1);  // [N * n_head, L, d_head/2]
        x_0            = ggml_reshape_4d(ctx, x_0, 1, x_0->ne[0], x_0->ne[1], x_0->ne[2]);                    // [N * n_head, L, d_head/2, 1]
        x_1            = ggml_reshape_4d(ctx, x_1, 1, x_1->ne[0], x_1->ne[1], x_1->ne[2]);                    // [N * n_head, L, d_head/2, 1]
        auto temp_x    = ggml_new_tensor_4d(ctx, x_0->type, 2, x_0->ne[1], x_0->ne[2], x_0->ne[3]);
        x_0            = ggml_repeat(ctx, x_0, temp_x);  // [N * n_head, L, d_head/2, 2]
        x_1            = ggml_repeat(ctx, x_1, temp_x);  // [N * n_head, L, d_head/2, 2]

        pe        = ggml_cont(ctx, ggml_permute(ctx, pe, 3, 0, 1, 2));  // [2, L, d_head/2, 2]
        offset    = pe->nb[2] * pe->ne[2];
        auto pe_0 = ggml_view_3d(ctx, pe, pe->ne[0], pe->ne[1], pe->ne[2], pe->nb[1], pe->nb[2], offset * 0);  // [L, d_head/2, 2]
        auto pe_1 = ggml_view_3d(ctx, pe, pe->ne[0], pe->ne[1], pe->ne[2], pe->nb[1], pe->nb[2], offset * 1);  // [L, d_head/2, 2]

        auto x_out = ggml_add_inplace(ctx, ggml_mul(ctx, x_0, pe_0), ggml_mul(ctx, x_1, pe_1));  // [N * n_head, L, d_head/2, 2]
        x_out      = ggml_reshape_3d(ctx, x_out, d_head, L, n_head * N);                         // [N*n_head, L, d_head]
        return x_out;
    }

    __STATIC_INLINE__ struct ggml_tensor* attention(struct ggml_context* ctx,
                                                    struct ggml_tensor* q,
                                                    struct ggml_tensor* k,
                                                    struct ggml_tensor* v,
                                                    struct ggml_tensor* pe) {
        // q,k,v: [N, L, n_head, d_head]
        // pe: [L, d_head/2, 2, 2]
        // return: [N, L, n_head*d_head]
        q = apply_rope(ctx, q, pe);  // [N*n_head, L, d_head]
        k = apply_rope(ctx, k, pe);  // [N*n_head, L, d_head]

        auto x = ggml_nn_attention_ext(ctx, q, k, v, v->ne[1], NULL, false, true);  // [N, L, n_head*d_head]
        return x;
    }

    struct SelfAttention : public GGMLBlock {
    public:
        int64_t num_heads;

    public:
        SelfAttention(int64_t dim,
                      int64_t num_heads = 8,
                      bool qkv_bias     = false)
            : num_heads(num_heads) {
            int64_t head_dim = dim / num_heads;
            blocks["qkv"]    = std::shared_ptr<GGMLBlock>(new Linear(dim, dim * 3, qkv_bias));
            blocks["norm"]   = std::shared_ptr<GGMLBlock>(new QKNorm(head_dim));
            blocks["proj"]   = std::shared_ptr<GGMLBlock>(new Linear(dim, dim));
        }

        std::vector<struct ggml_tensor*> pre_attention(struct ggml_context* ctx, struct ggml_tensor* x) {
            auto qkv_proj = std::dynamic_pointer_cast<Linear>(blocks["qkv"]);
            auto norm     = std::dynamic_pointer_cast<QKNorm>(blocks["norm"]);

            auto qkv         = qkv_proj->forward(ctx, x);
            auto qkv_vec     = split_qkv(ctx, qkv);
            int64_t head_dim = qkv_vec[0]->ne[0] / num_heads;
            auto q           = ggml_reshape_4d(ctx, qkv_vec[0], head_dim, num_heads, qkv_vec[0]->ne[1], qkv_vec[0]->ne[2]);
            auto k           = ggml_reshape_4d(ctx, qkv_vec[1], head_dim, num_heads, qkv_vec[1]->ne[1], qkv_vec[1]->ne[2]);
            auto v           = ggml_reshape_4d(ctx, qkv_vec[2], head_dim, num_heads, qkv_vec[2]->ne[1], qkv_vec[2]->ne[2]);
            q                = norm->query_norm(ctx, q);
            k                = norm->key_norm(ctx, k);
            return {q, k, v};
        }

        struct ggml_tensor* post_attention(struct ggml_context* ctx, struct ggml_tensor* x) {
            auto proj = std::dynamic_pointer_cast<Linear>(blocks["proj"]);

            x = proj->forward(ctx, x);  // [N, n_token, dim]
            return x;
        }

        struct ggml_tensor* forward(struct ggml_context* ctx, struct ggml_tensor* x, struct ggml_tensor* pe) {
            // x: [N, n_token, dim]
            // pe: [n_token, d_head/2, 2, 2]
            // return [N, n_token, dim]
            auto qkv = pre_attention(ctx, x);                       // q,k,v: [N, n_token, n_head, d_head]
            x        = attention(ctx, qkv[0], qkv[1], qkv[2], pe);  // [N, n_token, dim]
            x        = post_attention(ctx, x);                      // [N, n_token, dim]
            return x;
        }
    };

    struct ModulationOut {
        ggml_tensor* shift = NULL;
        ggml_tensor* scale = NULL;
        ggml_tensor* gate  = NULL;

        ModulationOut(ggml_tensor* shift = NULL, ggml_tensor* scale = NULL, ggml_tensor* gate = NULL)
            : shift(shift), scale(scale), gate(gate) {}
    };

    struct Modulation : public GGMLBlock {
    public:
        bool is_double;
        int multiplier;

    public:
        Modulation(int64_t dim, bool is_double)
            : is_double(is_double) {
            multiplier    = is_double ? 6 : 3;
            blocks["lin"] = std::shared_ptr<GGMLBlock>(new Linear(dim, dim * multiplier));
        }

        std::vector<ModulationOut> forward(struct ggml_context* ctx, struct ggml_tensor* vec) {
            // x: [N, dim]
            // return: [ModulationOut, ModulationOut]
            auto lin = std::dynamic_pointer_cast<Linear>(blocks["lin"]);

            auto out = ggml_silu(ctx, vec);
            out      = lin->forward(ctx, out);  // [N, multiplier*dim]

            auto m = ggml_reshape_3d(ctx, out, vec->ne[0], multiplier, vec->ne[1]);  // [N, multiplier, dim]
            m      = ggml_cont(ctx, ggml_permute(ctx, m, 0, 2, 1, 3));               // [multiplier, N, dim]

            int64_t offset = m->nb[1] * m->ne[1];
            auto shift_0   = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 0);  // [N, dim]
            auto scale_0   = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 1);  // [N, dim]
            auto gate_0    = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 2);  // [N, dim]

            if (is_double) {
                auto shift_1 = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 3);  // [N, dim]
                auto scale_1 = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 4);  // [N, dim]
                auto gate_1  = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 5);  // [N, dim]
                return {ModulationOut(shift_0, scale_0, gate_0), ModulationOut(shift_1, scale_1, gate_1)};
            }

            return {ModulationOut(shift_0, scale_0, gate_0), ModulationOut()};
        }
    };

    __STATIC_INLINE__ struct ggml_tensor* modulate(struct ggml_context* ctx,
                                                   struct ggml_tensor* x,
                                                   struct ggml_tensor* shift,
                                                   struct ggml_tensor* scale) {
        // x: [N, L, C]
        // scale: [N, C]
        // shift: [N, C]
        scale = ggml_reshape_3d(ctx, scale, scale->ne[0], 1, scale->ne[1]);  // [N, 1, C]
        shift = ggml_reshape_3d(ctx, shift, shift->ne[0], 1, shift->ne[1]);  // [N, 1, C]
        x     = ggml_add(ctx, x, ggml_mul(ctx, x, scale));
        x     = ggml_add(ctx, x, shift);
        return x;
    }

    struct DoubleStreamBlock : public GGMLBlock {
    public:
        DoubleStreamBlock(int64_t hidden_size,
                          int64_t num_heads,
                          float mlp_ratio,
                          bool qkv_bias = false) {
            int64_t mlp_hidden_dim = hidden_size * mlp_ratio;
            blocks["img_mod"]      = std::shared_ptr<GGMLBlock>(new Modulation(hidden_size, true));
            blocks["img_norm1"]    = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-6f, false));
            blocks["img_attn"]     = std::shared_ptr<GGMLBlock>(new SelfAttention(hidden_size, num_heads, qkv_bias));

            blocks["img_norm2"] = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-6f, false));
            blocks["img_mlp.0"] = std::shared_ptr<GGMLBlock>(new Linear(hidden_size, mlp_hidden_dim));
            // img_mlp.1 is nn.GELU(approximate="tanh")
            blocks["img_mlp.2"] = std::shared_ptr<GGMLBlock>(new Linear(mlp_hidden_dim, hidden_size));

            blocks["txt_mod"]   = std::shared_ptr<GGMLBlock>(new Modulation(hidden_size, true));
            blocks["txt_norm1"] = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-6f, false));
            blocks["txt_attn"]  = std::shared_ptr<GGMLBlock>(new SelfAttention(hidden_size, num_heads, qkv_bias));

            blocks["txt_norm2"] = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-6f, false));
            blocks["txt_mlp.0"] = std::shared_ptr<GGMLBlock>(new Linear(hidden_size, mlp_hidden_dim));
            // img_mlp.1 is nn.GELU(approximate="tanh")
            blocks["txt_mlp.2"] = std::shared_ptr<GGMLBlock>(new Linear(mlp_hidden_dim, hidden_size));
        }

        std::pair<struct ggml_tensor*, struct ggml_tensor*> forward(struct ggml_context* ctx,
                                                                    struct ggml_tensor* img,
                                                                    struct ggml_tensor* txt,
                                                                    struct ggml_tensor* vec,
                                                                    struct ggml_tensor* pe) {
            // img: [N, n_img_token, hidden_size]
            // txt: [N, n_txt_token, hidden_size]
            // pe: [n_img_token + n_txt_token, d_head/2, 2, 2]
            // return: ([N, n_img_token, hidden_size], [N, n_txt_token, hidden_size])

            auto img_mod   = std::dynamic_pointer_cast<Modulation>(blocks["img_mod"]);
            auto img_norm1 = std::dynamic_pointer_cast<LayerNorm>(blocks["img_norm1"]);
            auto img_attn  = std::dynamic_pointer_cast<SelfAttention>(blocks["img_attn"]);

            auto img_norm2 = std::dynamic_pointer_cast<LayerNorm>(blocks["img_norm2"]);
            auto img_mlp_0 = std::dynamic_pointer_cast<Linear>(blocks["img_mlp.0"]);
            auto img_mlp_2 = std::dynamic_pointer_cast<Linear>(blocks["img_mlp.2"]);

            auto txt_mod   = std::dynamic_pointer_cast<Modulation>(blocks["txt_mod"]);
            auto txt_norm1 = std::dynamic_pointer_cast<LayerNorm>(blocks["txt_norm1"]);
            auto txt_attn  = std::dynamic_pointer_cast<SelfAttention>(blocks["txt_attn"]);

            auto txt_norm2 = std::dynamic_pointer_cast<LayerNorm>(blocks["txt_norm2"]);
            auto txt_mlp_0 = std::dynamic_pointer_cast<Linear>(blocks["txt_mlp.0"]);
            auto txt_mlp_2 = std::dynamic_pointer_cast<Linear>(blocks["txt_mlp.2"]);

            auto img_mods          = img_mod->forward(ctx, vec);
            ModulationOut img_mod1 = img_mods[0];
            ModulationOut img_mod2 = img_mods[1];
            auto txt_mods          = txt_mod->forward(ctx, vec);
            ModulationOut txt_mod1 = txt_mods[0];
            ModulationOut txt_mod2 = txt_mods[1];

            // prepare image for attention
            auto img_modulated = img_norm1->forward(ctx, img);
            img_modulated      = Flux::modulate(ctx, img_modulated, img_mod1.shift, img_mod1.scale);
            auto img_qkv       = img_attn->pre_attention(ctx, img_modulated);  // q,k,v: [N, n_img_token, n_head, d_head]
            auto img_q         = img_qkv[0];
            auto img_k         = img_qkv[1];
            auto img_v         = img_qkv[2];

            // prepare txt for attention
            auto txt_modulated = txt_norm1->forward(ctx, txt);
            txt_modulated      = Flux::modulate(ctx, txt_modulated, txt_mod1.shift, txt_mod1.scale);
            auto txt_qkv       = txt_attn->pre_attention(ctx, txt_modulated);  // q,k,v: [N, n_txt_token, n_head, d_head]
            auto txt_q         = txt_qkv[0];
            auto txt_k         = txt_qkv[1];
            auto txt_v         = txt_qkv[2];

            // run actual attention
            auto q = ggml_concat(ctx, txt_q, img_q, 2);  // [N, n_txt_token + n_img_token, n_head, d_head]
            auto k = ggml_concat(ctx, txt_k, img_k, 2);  // [N, n_txt_token + n_img_token, n_head, d_head]
            auto v = ggml_concat(ctx, txt_v, img_v, 2);  // [N, n_txt_token + n_img_token, n_head, d_head]

            auto attn         = attention(ctx, q, k, v, pe);                          // [N, n_txt_token + n_img_token, n_head*d_head]
            attn              = ggml_cont(ctx, ggml_permute(ctx, attn, 0, 2, 1, 3));  // [n_txt_token + n_img_token, N, hidden_size]
            auto txt_attn_out = ggml_view_3d(ctx,
                                             attn,
                                             attn->ne[0],
                                             attn->ne[1],
                                             txt->ne[1],
                                             attn->nb[1],
                                             attn->nb[2],
                                             0);                                              // [n_txt_token, N, hidden_size]
            txt_attn_out      = ggml_cont(ctx, ggml_permute(ctx, txt_attn_out, 0, 2, 1, 3));  // [N, n_txt_token, hidden_size]
            auto img_attn_out = ggml_view_3d(ctx,
                                             attn,
                                             attn->ne[0],
                                             attn->ne[1],
                                             img->ne[1],
                                             attn->nb[1],
                                             attn->nb[2],
                                             attn->nb[2] * txt->ne[1]);                       // [n_img_token, N, hidden_size]
            img_attn_out      = ggml_cont(ctx, ggml_permute(ctx, img_attn_out, 0, 2, 1, 3));  // [N, n_img_token, hidden_size]

            // calculate the img bloks
            img = ggml_add(ctx, img, ggml_mul(ctx, img_attn->post_attention(ctx, img_attn_out), img_mod1.gate));

            auto img_mlp_out = img_mlp_0->forward(ctx, Flux::modulate(ctx, img_norm2->forward(ctx, img), img_mod2.shift, img_mod2.scale));
            img_mlp_out      = ggml_gelu_inplace(ctx, img_mlp_out);
            img_mlp_out      = img_mlp_2->forward(ctx, img_mlp_out);

            img = ggml_add(ctx, img, ggml_mul(ctx, img_mlp_out, img_mod2.gate));

            // calculate the txt bloks
            txt = ggml_add(ctx, txt, ggml_mul(ctx, txt_attn->post_attention(ctx, txt_attn_out), txt_mod1.gate));

            auto txt_mlp_out = txt_mlp_0->forward(ctx, Flux::modulate(ctx, txt_norm2->forward(ctx, txt), txt_mod2.shift, txt_mod2.scale));
            txt_mlp_out      = ggml_gelu_inplace(ctx, txt_mlp_out);
            txt_mlp_out      = txt_mlp_2->forward(ctx, txt_mlp_out);

            txt = ggml_add(ctx, txt, ggml_mul(ctx, txt_mlp_out, txt_mod2.gate));

            return {img, txt};
        }
    };

    struct SingleStreamBlock : public GGMLBlock {
    public:
        int64_t num_heads;
        int64_t hidden_size;
        int64_t mlp_hidden_dim;

    public:
        SingleStreamBlock(int64_t hidden_size,
                          int64_t num_heads,
                          float mlp_ratio = 4.0f,
                          float qk_scale  = 0.f)
            : hidden_size(hidden_size), num_heads(num_heads) {
            int64_t head_dim = hidden_size / num_heads;
            float scale      = qk_scale;
            if (scale <= 0.f) {
                scale = 1 / sqrt((float)head_dim);
            }
            mlp_hidden_dim = hidden_size * mlp_ratio;

            blocks["linear1"]  = std::shared_ptr<GGMLBlock>(new Linear(hidden_size, hidden_size * 3 + mlp_hidden_dim));
            blocks["linear2"]  = std::shared_ptr<GGMLBlock>(new Linear(hidden_size + mlp_hidden_dim, hidden_size));
            blocks["norm"]     = std::shared_ptr<GGMLBlock>(new QKNorm(head_dim));
            blocks["pre_norm"] = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-6f, false));
            // mlp_act is nn.GELU(approximate="tanh")
            blocks["modulation"] = std::shared_ptr<GGMLBlock>(new Modulation(hidden_size, false));
        }

        struct ggml_tensor* forward(struct ggml_context* ctx,
                                    struct ggml_tensor* x,
                                    struct ggml_tensor* vec,
                                    struct ggml_tensor* pe) {
            // x: [N, n_token, hidden_size]
            // pe: [n_token, d_head/2, 2, 2]
            // return: [N, n_token, hidden_size]

            auto linear1    = std::dynamic_pointer_cast<Linear>(blocks["linear1"]);
            auto linear2    = std::dynamic_pointer_cast<Linear>(blocks["linear2"]);
            auto norm       = std::dynamic_pointer_cast<QKNorm>(blocks["norm"]);
            auto pre_norm   = std::dynamic_pointer_cast<LayerNorm>(blocks["pre_norm"]);
            auto modulation = std::dynamic_pointer_cast<Modulation>(blocks["modulation"]);

            auto mods         = modulation->forward(ctx, vec);
            ModulationOut mod = mods[0];

            auto x_mod   = Flux::modulate(ctx, pre_norm->forward(ctx, x), mod.shift, mod.scale);
            auto qkv_mlp = linear1->forward(ctx, x_mod);                            // [N, n_token, hidden_size * 3 + mlp_hidden_dim]
            qkv_mlp      = ggml_cont(ctx, ggml_permute(ctx, qkv_mlp, 2, 0, 1, 3));  // [hidden_size * 3 + mlp_hidden_dim, N, n_token]

            auto qkv = ggml_view_3d(ctx,
                                    qkv_mlp,
                                    qkv_mlp->ne[0],
                                    qkv_mlp->ne[1],
                                    hidden_size * 3,
                                    qkv_mlp->nb[1],
                                    qkv_mlp->nb[2],
                                    0);                                     // [hidden_size * 3 , N, n_token]
            qkv      = ggml_cont(ctx, ggml_permute(ctx, qkv, 1, 2, 0, 3));  // [N, n_token, hidden_size * 3]
            auto mlp = ggml_view_3d(ctx,
                                    qkv_mlp,
                                    qkv_mlp->ne[0],
                                    qkv_mlp->ne[1],
                                    mlp_hidden_dim,
                                    qkv_mlp->nb[1],
                                    qkv_mlp->nb[2],
                                    qkv_mlp->nb[2] * hidden_size * 3);      // [mlp_hidden_dim , N, n_token]
            mlp      = ggml_cont(ctx, ggml_permute(ctx, mlp, 1, 2, 0, 3));  // [N, n_token, mlp_hidden_dim]

            auto qkv_vec     = split_qkv(ctx, qkv);  // q,k,v: [N, n_token, hidden_size]
            int64_t head_dim = hidden_size / num_heads;
            auto q           = ggml_reshape_4d(ctx, qkv_vec[0], head_dim, num_heads, qkv_vec[0]->ne[1], qkv_vec[0]->ne[2]);  // [N, n_token, n_head, d_head]
            auto k           = ggml_reshape_4d(ctx, qkv_vec[1], head_dim, num_heads, qkv_vec[1]->ne[1], qkv_vec[1]->ne[2]);  // [N, n_token, n_head, d_head]
            auto v           = ggml_reshape_4d(ctx, qkv_vec[2], head_dim, num_heads, qkv_vec[2]->ne[1], qkv_vec[2]->ne[2]);  // [N, n_token, n_head, d_head]
            q                = norm->query_norm(ctx, q);
            k                = norm->key_norm(ctx, k);
            auto attn        = attention(ctx, q, k, v, pe);  // [N, n_token, hidden_size]

            auto attn_mlp = ggml_concat(ctx, attn, ggml_gelu_inplace(ctx, mlp), 0);  // [N, n_token, hidden_size + mlp_hidden_dim]
            auto output   = linear2->forward(ctx, attn_mlp);                         // [N, n_token, hidden_size]

            output = ggml_add(ctx, x, ggml_mul(ctx, output, mod.gate));
            return output;
        }
    };

    struct LastLayer : public GGMLBlock {
    public:
        LastLayer(int64_t hidden_size,
                  int64_t patch_size,
                  int64_t out_channels) {
            blocks["norm_final"]         = std::shared_ptr<GGMLBlock>(new LayerNorm(hidden_size, 1e-06f, false));
            blocks["linear"]             = std::shared_ptr<GGMLBlock>(new Linear(hidden_size, patch_size * patch_size * out_channels));
            blocks["adaLN_modulation.1"] = std::shared_ptr<GGMLBlock>(new Linear(hidden_size, 2 * hidden_size));
        }

        struct ggml_tensor* forward(struct ggml_context* ctx,
                                    struct ggml_tensor* x,
                                    struct ggml_tensor* c) {
            // x: [N, n_token, hidden_size]
            // c: [N, hidden_size]
            // return: [N, n_token, patch_size * patch_size * out_channels]
            auto norm_final         = std::dynamic_pointer_cast<LayerNorm>(blocks["norm_final"]);
            auto linear             = std::dynamic_pointer_cast<Linear>(blocks["linear"]);
            auto adaLN_modulation_1 = std::dynamic_pointer_cast<Linear>(blocks["adaLN_modulation.1"]);

            auto m = adaLN_modulation_1->forward(ctx, ggml_silu(ctx, c));  // [N, 2 * hidden_size]
            m      = ggml_reshape_3d(ctx, m, c->ne[0], 2, c->ne[1]);       // [N, 2, hidden_size]
            m      = ggml_cont(ctx, ggml_permute(ctx, m, 0, 2, 1, 3));     // [2, N, hidden_size]

            int64_t offset = m->nb[1] * m->ne[1];
            auto shift     = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 0);  // [N, hidden_size]
            auto scale     = ggml_view_2d(ctx, m, m->ne[0], m->ne[1], m->nb[1], offset * 1);  // [N, hidden_size]

            x = Flux::modulate(ctx, norm_final->forward(ctx, x), shift, scale);
            x = linear->forward(ctx, x);

            return x;
        }
    };

    struct FluxParams {
        int64_t in_channels         = 64;
        int64_t vec_in_dim          = 768;
        int64_t context_in_dim      = 4096;
        int64_t hidden_size         = 3072;
        float mlp_ratio             = 4.0f;
        int64_t num_heads           = 24;
        int64_t depth               = 19;
        int64_t depth_single_blocks = 38;
        std::vector<int> axes_dim   = {16, 56, 56};
        int64_t axes_dim_sum        = 128;
        int theta                   = 10000;
        bool qkv_bias               = true;
        bool guidance_embed         = true;
    };

    struct Flux : public GGMLBlock {
    public:
        std::vector<float> linspace(float start, float end, int num) {
            std::vector<float> result(num);
            float step = (end - start) / (num - 1);
            for (int i = 0; i < num; ++i) {
                result[i] = start + i * step;
            }
            return result;
        }

        std::vector<std::vector<float>> transpose(const std::vector<std::vector<float>>& mat) {
            int rows = mat.size();
            int cols = mat[0].size();
            std::vector<std::vector<float>> transposed(cols, std::vector<float>(rows));
            for (int i = 0; i < rows; ++i) {
                for (int j = 0; j < cols; ++j) {
                    transposed[j][i] = mat[i][j];
                }
            }
            return transposed;
        }

        std::vector<float> flatten(const std::vector<std::vector<float>>& vec) {
            std::vector<float> flat_vec;
            for (const auto& sub_vec : vec) {
                flat_vec.insert(flat_vec.end(), sub_vec.begin(), sub_vec.end());
            }
            return flat_vec;
        }

        std::vector<std::vector<float>> rope(const std::vector<float>& pos, int dim, int theta) {
            assert(dim % 2 == 0);
            int half_dim = dim / 2;

            std::vector<float> scale = linspace(0, (dim * 1.0f - 2) / dim, half_dim);

            std::vector<float> omega(half_dim);
            for (int i = 0; i < half_dim; ++i) {
                omega[i] = 1.0 / std::pow(theta, scale[i]);
            }

            int pos_size = pos.size();
            std::vector<std::vector<float>> out(pos_size, std::vector<float>(half_dim));
            for (int i = 0; i < pos_size; ++i) {
                for (int j = 0; j < half_dim; ++j) {
                    out[i][j] = pos[i] * omega[j];
                }
            }

            std::vector<std::vector<float>> result(pos_size, std::vector<float>(half_dim * 4));
            for (int i = 0; i < pos_size; ++i) {
                for (int j = 0; j < half_dim; ++j) {
                    result[i][4 * j]     = std::cos(out[i][j]);
                    result[i][4 * j + 1] = -std::sin(out[i][j]);
                    result[i][4 * j + 2] = std::sin(out[i][j]);
                    result[i][4 * j + 3] = std::cos(out[i][j]);
                }
            }

            return result;
        }

        // Generate IDs for image patches and text
        std::vector<std::vector<float>> gen_ids(int h, int w, int patch_size, int bs, int context_len) {
            int h_len = (h + (patch_size / 2)) / patch_size;
            int w_len = (w + (patch_size / 2)) / patch_size;

            std::vector<std::vector<float>> img_ids(h_len * w_len, std::vector<float>(3, 0.0));

            std::vector<float> row_ids = linspace(0, h_len - 1, h_len);
            std::vector<float> col_ids = linspace(0, w_len - 1, w_len);

            for (int i = 0; i < h_len; ++i) {
                for (int j = 0; j < w_len; ++j) {
                    img_ids[i * w_len + j][1] = row_ids[i];
                    img_ids[i * w_len + j][2] = col_ids[j];
                }
            }

            std::vector<std::vector<float>> img_ids_repeated(bs * img_ids.size(), std::vector<float>(3));
            for (int i = 0; i < bs; ++i) {
                for (int j = 0; j < img_ids.size(); ++j) {
                    img_ids_repeated[i * img_ids.size() + j] = img_ids[j];
                }
            }

            std::vector<std::vector<float>> txt_ids(bs * context_len, std::vector<float>(3, 0.0));
            std::vector<std::vector<float>> ids(bs * (context_len + img_ids.size()), std::vector<float>(3));
            for (int i = 0; i < bs; ++i) {
                for (int j = 0; j < context_len; ++j) {
                    ids[i * (context_len + img_ids.size()) + j] = txt_ids[j];
                }
                for (int j = 0; j < img_ids.size(); ++j) {
                    ids[i * (context_len + img_ids.size()) + context_len + j] = img_ids_repeated[i * img_ids.size() + j];
                }
            }

            return ids;
        }

        // Generate positional embeddings
        std::vector<float> gen_pe(int h, int w, int patch_size, int bs, int context_len, int theta, const std::vector<int>& axes_dim) {
            std::vector<std::vector<float>> ids       = gen_ids(h, w, patch_size, bs, context_len);
            std::vector<std::vector<float>> trans_ids = transpose(ids);
            size_t pos_len                            = ids.size();
            int num_axes                              = axes_dim.size();
            for (int i = 0; i < pos_len; i++) {
                // std::cout << trans_ids[0][i] << " " << trans_ids[1][i] << " " << trans_ids[2][i] << std::endl;
            }

            int emb_dim = 0;
            for (int d : axes_dim)
                emb_dim += d / 2;

            std::vector<std::vector<float>> emb(bs * pos_len, std::vector<float>(emb_dim * 2 * 2, 0.0));
            int offset = 0;
            for (int i = 0; i < num_axes; ++i) {
                std::vector<std::vector<float>> rope_emb = rope(trans_ids[i], axes_dim[i], theta);  // [bs*pos_len, axes_dim[i]/2 * 2 * 2]
                for (int b = 0; b < bs; ++b) {
                    for (int j = 0; j < pos_len; ++j) {
                        for (int k = 0; k < rope_emb[0].size(); ++k) {
                            emb[b * pos_len + j][offset + k] = rope_emb[j][k];
                        }
                    }
                }
                offset += rope_emb[0].size();
            }

            return flatten(emb);
        }

    public:
        FluxParams params;
        Flux() {}
        Flux(FluxParams params)
            : params(params) {
            int64_t out_channels = params.in_channels;
            int64_t pe_dim       = params.hidden_size / params.num_heads;

            blocks["img_in"]    = std::shared_ptr<GGMLBlock>(new Linear(params.in_channels, params.hidden_size, true));
            blocks["time_in"]   = std::shared_ptr<GGMLBlock>(new MLPEmbedder(256, params.hidden_size));
            blocks["vector_in"] = std::shared_ptr<GGMLBlock>(new MLPEmbedder(params.vec_in_dim, params.hidden_size));
            if (params.guidance_embed) {
                blocks["guidance_in"] = std::shared_ptr<GGMLBlock>(new MLPEmbedder(256, params.hidden_size));
            }
            blocks["txt_in"] = std::shared_ptr<GGMLBlock>(new Linear(params.context_in_dim, params.hidden_size, true));

            for (int i = 0; i < params.depth; i++) {
                blocks["double_blocks." + std::to_string(i)] = std::shared_ptr<GGMLBlock>(new DoubleStreamBlock(params.hidden_size,
                                                                                                                params.num_heads,
                                                                                                                params.mlp_ratio,
                                                                                                                params.qkv_bias));
            }

            for (int i = 0; i < params.depth_single_blocks; i++) {
                blocks["single_blocks." + std::to_string(i)] = std::shared_ptr<GGMLBlock>(new SingleStreamBlock(params.hidden_size,
                                                                                                                params.num_heads,
                                                                                                                params.mlp_ratio));
            }

            blocks["final_layer"] = std::shared_ptr<GGMLBlock>(new LastLayer(params.hidden_size, 1, out_channels));
        }

        struct ggml_tensor* patchify(struct ggml_context* ctx,
                                     struct ggml_tensor* x,
                                     int64_t patch_size) {
            // x: [N, C, H, W]
            // return: [N, h*w, C * patch_size * patch_size]
            int64_t N = x->ne[3];
            int64_t C = x->ne[2];
            int64_t H = x->ne[1];
            int64_t W = x->ne[0];
            int64_t p = patch_size;
            int64_t h = H / patch_size;
            int64_t w = W / patch_size;

            GGML_ASSERT(h * p == H && w * p == W);

            x = ggml_reshape_4d(ctx, x, p, w, p, h * C * N);       // [N*C*h, p, w, p]
            x = ggml_cont(ctx, ggml_permute(ctx, x, 0, 2, 1, 3));  // [N*C*h, w, p, p]
            x = ggml_reshape_4d(ctx, x, p * p, w * h, C, N);       // [N, C, h*w, p*p]
            x = ggml_cont(ctx, ggml_permute(ctx, x, 0, 2, 1, 3));  // [N, h*w, C, p*p]
            x = ggml_reshape_3d(ctx, x, p * p * C, w * h, N);      // [N, h*w, C*p*p]
            return x;
        }

        struct ggml_tensor* unpatchify(struct ggml_context* ctx,
                                       struct ggml_tensor* x,
                                       int64_t h,
                                       int64_t w,
                                       int64_t patch_size) {
            // x: [N, h*w, C*patch_size*patch_size]
            // return: [N, C, H, W]
            int64_t N = x->ne[2];
            int64_t C = x->ne[0] / patch_size / patch_size;
            int64_t H = h * patch_size;
            int64_t W = w * patch_size;
            int64_t p = patch_size;

            GGML_ASSERT(C * p * p == x->ne[0]);

            x = ggml_reshape_4d(ctx, x, p * p, C, w * h, N);       // [N, h*w, C, p*p]
            x = ggml_cont(ctx, ggml_permute(ctx, x, 0, 2, 1, 3));  // [N, C, h*w, p*p]
            x = ggml_reshape_4d(ctx, x, p, p, w, h * C * N);       // [N*C*h, w, p, p]
            x = ggml_cont(ctx, ggml_permute(ctx, x, 0, 2, 1, 3));  // [N*C*h, p, w, p]
            x = ggml_reshape_4d(ctx, x, W, H, C, N);               // [N, C, h*p, w*p]

            return x;
        }

        struct ggml_tensor* forward_orig(struct ggml_context* ctx,
                                         struct ggml_tensor* img,
                                         struct ggml_tensor* txt,
                                         struct ggml_tensor* timesteps,
                                         struct ggml_tensor* y,
                                         struct ggml_tensor* guidance,
                                         struct ggml_tensor* pe) {
            auto img_in      = std::dynamic_pointer_cast<Linear>(blocks["img_in"]);
            auto time_in     = std::dynamic_pointer_cast<MLPEmbedder>(blocks["time_in"]);
            auto vector_in   = std::dynamic_pointer_cast<MLPEmbedder>(blocks["vector_in"]);
            auto txt_in      = std::dynamic_pointer_cast<Linear>(blocks["txt_in"]);
            auto final_layer = std::dynamic_pointer_cast<LastLayer>(blocks["final_layer"]);

            img      = img_in->forward(ctx, img);
            auto vec = time_in->forward(ctx, ggml_nn_timestep_embedding(ctx, timesteps, 256, 10000, 1000.f));

            if (params.guidance_embed) {
                GGML_ASSERT(guidance != NULL);
                auto guidance_in = std::dynamic_pointer_cast<MLPEmbedder>(blocks["guidance_in"]);
                // bf16 and fp16 result is different
                auto g_in = ggml_nn_timestep_embedding(ctx, guidance, 256, 10000, 1000.f);
                vec       = ggml_add(ctx, vec, guidance_in->forward(ctx, g_in));
            }

            vec = ggml_add(ctx, vec, vector_in->forward(ctx, y));
            txt = txt_in->forward(ctx, txt);

            for (int i = 0; i < params.depth; i++) {
                auto block = std::dynamic_pointer_cast<DoubleStreamBlock>(blocks["double_blocks." + std::to_string(i)]);

                auto img_txt = block->forward(ctx, img, txt, vec, pe);
                img          = img_txt.first;   // [N, n_img_token, hidden_size]
                txt          = img_txt.second;  // [N, n_txt_token, hidden_size]
            }

            auto txt_img = ggml_concat(ctx, txt, img, 1);  // [N, n_txt_token + n_img_token, hidden_size]
            for (int i = 0; i < params.depth_single_blocks; i++) {
                auto block = std::dynamic_pointer_cast<SingleStreamBlock>(blocks["single_blocks." + std::to_string(i)]);

                txt_img = block->forward(ctx, txt_img, vec, pe);
            }

            txt_img = ggml_cont(ctx, ggml_permute(ctx, txt_img, 0, 2, 1, 3));  // [n_txt_token + n_img_token, N, hidden_size]
            img     = ggml_view_3d(ctx,
                                   txt_img,
                                   txt_img->ne[0],
                                   txt_img->ne[1],
                                   img->ne[1],
                                   txt_img->nb[1],
                                   txt_img->nb[2],
                                   txt_img->nb[2] * txt->ne[1]);           // [n_img_token, N, hidden_size]
            img     = ggml_cont(ctx, ggml_permute(ctx, img, 0, 2, 1, 3));  // [N, n_img_token, hidden_size]

            img = final_layer->forward(ctx, img, vec);  // (N, T, patch_size ** 2 * out_channels)

            return img;
        }

        struct ggml_tensor* forward(struct ggml_context* ctx,
                                    struct ggml_tensor* x,
                                    struct ggml_tensor* timestep,
                                    struct ggml_tensor* context,
                                    struct ggml_tensor* y,
                                    struct ggml_tensor* guidance,
                                    struct ggml_tensor* pe) {
            // Forward pass of DiT.
            // x: (N, C, H, W) tensor of spatial inputs (images or latent representations of images)
            // timestep: (N,) tensor of diffusion timesteps
            // context: (N, L, D)
            // y: (N, adm_in_channels) tensor of class labels
            // guidance: (N,)
            // pe: (L, d_head/2, 2, 2)
            // return: (N, C, H, W)

            GGML_ASSERT(x->ne[3] == 1);

            int64_t W          = x->ne[0];
            int64_t H          = x->ne[1];
            int64_t patch_size = 2;
            int pad_h          = (patch_size - H % patch_size) % patch_size;
            int pad_w          = (patch_size - W % patch_size) % patch_size;
            x                  = ggml_pad(ctx, x, pad_w, pad_h, 0, 0);  // [N, C, H + pad_h, W + pad_w]

            // img = rearrange(x, "b c (h ph) (w pw) -> b (h w) (c ph pw)", ph=patch_size, pw=patch_size)
            auto img = patchify(ctx, x, patch_size);  // [N, h*w, C * patch_size * patch_size]

            auto out = forward_orig(ctx, img, context, timestep, y, guidance, pe);  // [N, h*w, C * patch_size * patch_size]

            // rearrange(out, "b (h w) (c ph pw) -> b c (h ph) (w pw)", h=h_len, w=w_len, ph=2, pw=2)
            out = unpatchify(ctx, out, (H + pad_h) / patch_size, (W + pad_w) / patch_size, patch_size);  // [N, C, H + pad_h, W + pad_w]

            return out;
        }
    };

    struct FluxRunner : public GGMLRunner {
    public:
        FluxParams flux_params;
        Flux flux;
        std::vector<float> pe_vec;  // for cache

        FluxRunner(ggml_backend_t backend,
                   ggml_type wtype,
                   SDVersion version = VERSION_FLUX_DEV)
            : GGMLRunner(backend, wtype) {
            if (version == VERSION_FLUX_SCHNELL) {
                flux_params.guidance_embed = false;
            }
            flux = Flux(flux_params);
            flux.init(params_ctx, wtype);
        }

        std::string get_desc() {
            return "flux";
        }

        void get_param_tensors(std::map<std::string, struct ggml_tensor*>& tensors, const std::string prefix) {
            flux.get_param_tensors(tensors, prefix);
        }

        struct ggml_cgraph* build_graph(struct ggml_tensor* x,
                                        struct ggml_tensor* timesteps,
                                        struct ggml_tensor* context,
                                        struct ggml_tensor* y,
                                        struct ggml_tensor* guidance) {
            GGML_ASSERT(x->ne[3] == 1);
            struct ggml_cgraph* gf = ggml_new_graph_custom(compute_ctx, FLUX_GRAPH_SIZE, false);

            x         = to_backend(x);
            context   = to_backend(context);
            y         = to_backend(y);
            timesteps = to_backend(timesteps);
            if (flux_params.guidance_embed) {
                guidance = to_backend(guidance);
            }

            pe_vec      = flux.gen_pe(x->ne[1], x->ne[0], 2, x->ne[3], context->ne[1], flux_params.theta, flux_params.axes_dim);
            int pos_len = pe_vec.size() / flux_params.axes_dim_sum / 2;
            // LOG_DEBUG("pos_len %d", pos_len);
            auto pe = ggml_new_tensor_4d(compute_ctx, GGML_TYPE_F32, 2, 2, flux_params.axes_dim_sum / 2, pos_len);
            // pe->data = pe_vec.data();
            // print_ggml_tensor(pe);
            // pe->data = NULL;
            set_backend_tensor_data(pe, pe_vec.data());

            struct ggml_tensor* out = flux.forward(compute_ctx,
                                                   x,
                                                   timesteps,
                                                   context,
                                                   y,
                                                   guidance,
                                                   pe);

            ggml_build_forward_expand(gf, out);

            return gf;
        }

        void compute(int n_threads,
                     struct ggml_tensor* x,
                     struct ggml_tensor* timesteps,
                     struct ggml_tensor* context,
                     struct ggml_tensor* y,
                     struct ggml_tensor* guidance,
                     struct ggml_tensor** output     = NULL,
                     struct ggml_context* output_ctx = NULL) {
            // x: [N, in_channels, h, w]
            // timesteps: [N, ]
            // context: [N, max_position, hidden_size]
            // y: [N, adm_in_channels] or [1, adm_in_channels]
            // guidance: [N, ]
            auto get_graph = [&]() -> struct ggml_cgraph* {
                return build_graph(x, timesteps, context, y, guidance);
            };

            GGMLRunner::compute(get_graph, n_threads, false, output, output_ctx);
        }

        void test() {
            struct ggml_init_params params;
            params.mem_size   = static_cast<size_t>(20 * 1024 * 1024);  // 20 MB
            params.mem_buffer = NULL;
            params.no_alloc   = false;

            struct ggml_context* work_ctx = ggml_init(params);
            GGML_ASSERT(work_ctx != NULL);

            {
                // cpu f16:
                // cuda f16: nan
                // cuda q8_0: pass
                auto x = ggml_new_tensor_4d(work_ctx, GGML_TYPE_F32, 16, 16, 16, 1);
                ggml_set_f32(x, 0.01f);
                // print_ggml_tensor(x);

                std::vector<float> timesteps_vec(1, 999.f);
                auto timesteps = vector_to_ggml_tensor(work_ctx, timesteps_vec);

                std::vector<float> guidance_vec(1, 3.5f);
                auto guidance = vector_to_ggml_tensor(work_ctx, guidance_vec);

                auto context = ggml_new_tensor_3d(work_ctx, GGML_TYPE_F32, 4096, 256, 1);
                ggml_set_f32(context, 0.01f);
                // print_ggml_tensor(context);

                auto y = ggml_new_tensor_2d(work_ctx, GGML_TYPE_F32, 768, 1);
                ggml_set_f32(y, 0.01f);
                // print_ggml_tensor(y);

                struct ggml_tensor* out = NULL;

                int t0 = ggml_time_ms();
                compute(8, x, timesteps, context, y, guidance, &out, work_ctx);
                int t1 = ggml_time_ms();

                print_ggml_tensor(out);
                LOG_DEBUG("flux test done in %dms", t1 - t0);
            }
        }

        static void load_from_file_and_test(const std::string& file_path) {
            // ggml_backend_t backend    = ggml_backend_cuda_init(0);
            ggml_backend_t backend           = ggml_backend_cpu_init();
            ggml_type model_data_type        = GGML_TYPE_Q8_0;
            std::shared_ptr<FluxRunner> flux = std::shared_ptr<FluxRunner>(new FluxRunner(backend, model_data_type));
            {
                LOG_INFO("loading from '%s'", file_path.c_str());

                flux->alloc_params_buffer();
                std::map<std::string, ggml_tensor*> tensors;
                flux->get_param_tensors(tensors, "model.diffusion_model");

                ModelLoader model_loader;
                if (!model_loader.init_from_file(file_path, "model.diffusion_model.")) {
                    LOG_ERROR("init model loader from file failed: '%s'", file_path.c_str());
                    return;
                }

                bool success = model_loader.load_tensors(tensors, backend);

                if (!success) {
                    LOG_ERROR("load tensors from model loader failed");
                    return;
                }

                LOG_INFO("flux model loaded");
            }
            flux->test();
        }
    };

}  // namespace Flux

#endif  // __FLUX_HPP__