DarthAffe/stable-diffusion.cpp
1
0 Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-11-11 11:26:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor: unify ggml operation helpers around runner context (#2314)

Browse Source
This commit is contained in:
leejet authored and GitHub committed 2026-11-10 01:58:04 +08:00
1 parent 73fef08224
commit a3c49b13be
76 files changed
+1983 -1831
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

src/model/diffusion/lingbot_video.hpp
+7 -7
View File
@@ -365,7 +365,7 @@ namespace LingBotVideo {
}
selected_mask = ggml_reshape_2d(gctx, selected_mask, num_experts, n_token_total);
ggml_tensor* excluded_group_mask = ggml_sub(gctx, selected_mask, ggml_ext_ones_like(gctx, selected_mask));
ggml_tensor* excluded_group_mask = ggml_sub(gctx, selected_mask, ggml_ext_ones_like(ctx, selected_mask));
ggml_tensor* masked_scores = ggml_add(gctx, choice_scores, ggml_scale(gctx, excluded_group_mask, 1.0e9f));
return ggml_argsort_top_k(gctx, masked_scores, static_cast<int>(num_experts_per_tok));
}
@@ -476,7 +476,7 @@ namespace LingBotVideo {
auto table = ggml_reshape_3d(ctx->ggml_ctx, params["scale_shift_table"], hidden_size * 6, 1, 1);
auto mod = ggml_add(ctx->ggml_ctx, temb6, table);
auto mods = ggml_ext_chunk(ctx->ggml_ctx, mod, 6, 0);
auto mods = ggml_ext_chunk(ctx, mod, 6, 0);
auto shift_msa = mods[0];
auto scale_msa = mods[1];
@@ -556,13 +556,13 @@ namespace LingBotVideo {
int64_t w_len = W / config.patch_w;
int64_t n_img = t_len * h_len * w_len;
auto img = DiT::patchify_3d(ctx->ggml_ctx, x, config.patch_t, config.patch_h, config.patch_w, N, false);
auto img = DiT::patchify_3d(ctx, x, config.patch_t, config.patch_h, config.patch_w, N, false);
img = patch_embedder->forward(ctx, img);
auto txt = text_embedder->forward(ctx, context);
auto hidden_states = ggml_concat(ctx->ggml_ctx, img, txt, 1);
auto timestep_proj = ggml_ext_timestep_embedding(ctx->ggml_ctx,
auto timestep_proj = ggml_ext_timestep_embedding(ctx,
timestep,
static_cast<int>(config.freq_dim),
10000,
@@ -582,16 +582,16 @@ namespace LingBotVideo {
sd::ggml_graph_cut::mark_graph_cut(hidden_states, "lingbot_video.blocks." + std::to_string(i), "hidden_states");
}
auto final_mods = ggml_ext_chunk(ctx->ggml_ctx,
auto final_mods = ggml_ext_chunk(ctx,
norm_out_modulation->forward(ctx, ggml_silu(ctx->ggml_ctx, temb_tokens)),
2,
0);
hidden_states = norm_out->forward(ctx, hidden_states);
hidden_states = Flux::modulate(ctx, hidden_states, final_mods[0], final_mods[1], true);
hidden_states = proj_out->forward(ctx, hidden_states);
hidden_states = ggml_ext_slice(ctx->ggml_ctx, hidden_states, 1, 0, n_img);
hidden_states = ggml_ext_slice(ctx, hidden_states, 1, 0, n_img);
auto out = DiT::unpatchify_3d(ctx->ggml_ctx,
auto out = DiT::unpatchify_3d(ctx,
hidden_states,
t_len,
h_len,