jdelony/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-06-28 04:30:15 -05:00
Code Issues Packages Projects Releases Wiki Activity

Gemma4 E2B/E4B tweaks (#1947)

Browse Source 
* Gemma4 E2B/E4B tweaks

* A few more named nodes
This commit is contained in:
Kawrakow 2026-06-10 15:28:54 +02:00 committed by GitHub
parent 4a1e2eaa69
commit c0d25e8fa1
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
src/graphs/build_gemma4.cpp
View File

@ -687,15 +687,14 @@ ggml_cgraph * llm_build_context::build_gemma4_mtp() {
static ggml_tensor * gemma4_project_per_layer_inputs(ggml_context * ctx0, const llama_model & model, const llm_build_cb & cb,
int n_embd, int n_embd_per_layer, int n_layer, int n_tokens,
ggml_tensor * inputs_embeds, ggml_tensor * inp_per_layer) {
const float per_layer_projection_scale = 1.0f / sqrtf((float) n_embd);
const float per_layer_input_scale = 1.0f / sqrtf(2.0f);
ggml_tensor * per_layer_proj = ggml_mul_mat(ctx0, model.per_layer_model_proj, inputs_embeds);
per_layer_proj = ggml_scale(ctx0, per_layer_proj, per_layer_projection_scale);
cb(per_layer_proj, "per_layer_proj", -1);
per_layer_proj = ggml_reshape_3d(ctx0, per_layer_proj, n_embd_per_layer, n_layer, n_tokens);
per_layer_proj = llm_build_context::llm_build_norm(ctx0, per_layer_proj, model.hparams,
model.per_layer_proj_norm, nullptr, LLM_NORM_RMS, cb, -1); // [n_embd_per_layer, n_layer, n_tokens]
cb(per_layer_proj, "per_layer_proj", -1);
model.per_layer_proj_norm, nullptr, LLM_NORM_RMS, cb, -1, 1.0f*n_embd); // [n_embd_per_layer, n_layer, n_tokens]
cb(per_layer_proj, "per_layer_proj_normed", -1);
inp_per_layer = ggml_add(ctx0, per_layer_proj, inp_per_layer);
inp_per_layer = ggml_scale(ctx0, inp_per_layer, per_layer_input_scale);
@ -902,19 +901,19 @@ ggml_cgraph * llm_build_context::build_gemma4() {
ggml_tensor * pe_in = cur;
cb(cur, "pe_in", il);
cur = llm_build_lora_mm(lctx, ctx0, model.layers[il].per_layer_inp_gate, cur); // [n_embd_per_layer, n_tokens]
cur = ggml_gelu(ctx0, cur);
ggml_tensor * inp_this_layer = ggml_view_2d(ctx0, inp_per_layer, inp_per_layer->ne[0], inp_per_layer->ne[1],
ggml_row_size(inp_per_layer->type, inp_per_layer->ne[0]),
il*inp_per_layer->ne[0]*inp_per_layer->ne[1]*ggml_element_size(inp_per_layer)); // [n_embd_per_layer, n_tokens]
// TODO @ngxson : improve this
if (il == n_layer - 1 && inp_out_ids) {
inp_this_layer = ggml_get_rows(ctx0, inp_this_layer, inp_out_ids);
}
cur = ggml_mul(ctx0, cur, inp_this_layer);
cur = llm_build_lora_mm(lctx, ctx0, model.layers[il].per_layer_inp_gate, cur); // [n_embd_per_layer, n_tokens]
cb(cur, "cur_gated", il);
cur = ggml_fused_mul_unary(ctx0, cur, inp_this_layer, GGML_UNARY_OP_GELU);
cb(cur, "cur_gelu", il);
cur = llm_build_lora_mm(lctx, ctx0, model.layers[il].per_layer_proj, cur); // [n_embd, n_tokens]
cb(cur, "cur_proj", il);
cur = llm_build_norm(ctx0, cur, hparams, model.layers[il].per_layer_post_norm, nullptr, LLM_NORM_RMS, cb, il);
cb(cur, "per_layer_embd_out", il);
@ -950,6 +949,7 @@ ggml_cgraph * llm_build_context::build_gemma4() {
cur = llm_build_lora_mm(lctx, ctx0, model.output, cur);
if (hparams.f_final_logit_softcapping > 0) {
cb(cur, "result_pre_softcap", -1);
cur = ggml_softcap(ctx0, cur, 1.0f / hparams.f_final_logit_softcapping, hparams.f_final_logit_softcapping);
}

5
src/llama-build-context.cpp
View File

@ -843,16 +843,15 @@ ggml_tensor * llm_build_context::llm_build_ffn(
}
if (down) {
cur = llm_build_lora_mm(lctx, ctx, down, cur);
cb(cur, "ffn_down", il);
if (lctx.model.arch == LLM_ARCH_GLM4 || lctx.model.arch == LLM_ARCH_GLM4_MOE) {
// GLM4 and GLM4_MOE seem to have numerical issues with half-precision accumulators
ggml_mul_mat_set_prec(cur, GGML_PREC_F32);
}
}
if (down_b) {
cb(cur, "ffn_down", il);
}
if (down_b) {
cur = ggml_add(ctx, cur, down_b);
cb(cur, "ffn_down_b", il);
}
if (down_s) {
cur = ggml_mul(ctx, cur, down_s);