thek0tyara/llama-cpp-turboquant
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 4

model: support GLM-OCR (#19677)

Browse source 
* model: support GLM-OCR

* Update convert_hf_to_gguf.py

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>

---------

Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com>
This commit is contained in:
Xuan-Son Nguyen 2026-02-18 17:51:40 +01:00 • committed by GitHub
parent e99f1083a0
commit eeef3cfced
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
8 changed files with 122 additions and 43 deletions
Download patch file Download diff file Expand all files Collapse all files

17
src/models/glm4.cpp
View file

@ -29,7 +29,10 @@ llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params
ggml_tensor * inp_out_ids = build_inp_out_ids();
for (int il = 0; il < n_layer; ++il) {
// Only process up to last layer (skip final NextN layer)
// Final layer tensors are loaded but not processed in forward pass
const int n_transformer_layers = n_layer - hparams.nextn_predict_layers;
for (int il = 0; il < n_transformer_layers; ++il) {
ggml_tensor * inpSA = inpL;
// Pre-attention norm
@ -100,7 +103,7 @@ llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params
model.layers[il].wo, NULL,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids) {
if (il == n_transformer_layers - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
}
@ -130,9 +133,13 @@ llm_build_glm4::llm_build_glm4(const llama_model & model, const llm_graph_params
cur = build_norm(cur, model.layers[il].ffn_post_norm, NULL, LLM_NORM_RMS, il);
cb(cur, "post_mlp_norm", il);
}
// Add residual connection after post-MLP norm
inpL = ggml_add(ctx0, cur, ffn_inp);
cb(inpL, "l_out", il);
cur = ggml_add(ctx0, cur, ffn_inp);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
// input for next layer
inpL = cur;
}
// Final norm
cur = build_norm(inpL, model.output_norm, NULL, LLM_NORM_RMS, -1);