Skip to content

model: fix all granite family parameter counts #28643

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ggerganov merged 3 commits into from
Sep 10, 2026
+16 −4
Merged

model: fix all granite family parameter counts #28643

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
27b87b6
model: fix all granite family parameter counts
taronaeo Sep 9, 2026
58592dd
model: fix additional include, add missing `A` prefix for active experts
taronaeo Sep 9, 2026
1c8101c
model: fix code alignment, rm unused 40 block case
taronaeo Sep 9, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions src/llama-model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -936,6 +936,7 @@ const char * llm_type_name(llm_type type) {
case LLM_TYPE_17B_128E: return "17Bx128E (Maverick)";
case LLM_TYPE_A13B: return "A13B";
case LLM_TYPE_1B_A400M: return "1B.A400M";
case LLM_TYPE_3B_A800M: return "3B.A800M";
case LLM_TYPE_7B_A1B: return "7B.A1B";
case LLM_TYPE_8B_A1B: return "8B.A1B";
case LLM_TYPE_7_9B_A1_3B: return "7.9B.A1.3B";
Expand All @@ -946,6 +947,7 @@ const char * llm_type_name(llm_type type) {
case LLM_TYPE_26B_A4B: return "26B.A4B";
case LLM_TYPE_30B_A3B: return "30B.A3B";
case LLM_TYPE_31B_A3_5B: return "31B.A3.5B";
case LLM_TYPE_32B_A9B: return "32B.A9B";
case LLM_TYPE_35B_A3B: return "35B.A3B";
case LLM_TYPE_48B_A3B: return "48B.A3B";
case LLM_TYPE_75B_A9B: return "75B.A9B";
Expand Down
2 changes: 2 additions & 0 deletions src/llama-model.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,6 +117,7 @@ enum llm_type {
LLM_TYPE_17B_128E, // llama4 Maverick
LLM_TYPE_A13B,
LLM_TYPE_1B_A400M, // Granite3 MoE
LLM_TYPE_3B_A800M, // Granite3 MoE
LLM_TYPE_7B_A1B,
LLM_TYPE_8B_A1B, // lfm2moe
LLM_TYPE_7_9B_A1_3B, // Ling-3.0-tiny
Expand All @@ -127,6 +128,7 @@ enum llm_type {
LLM_TYPE_26B_A4B, // Gemma4
LLM_TYPE_30B_A3B,
LLM_TYPE_31B_A3_5B,
LLM_TYPE_32B_A9B, // Granite4 Hybrid
LLM_TYPE_35B_A3B, // Qwen3.5
LLM_TYPE_48B_A3B, // Kimi Linear
LLM_TYPE_75B_A9B, // Nemotron 3 Puzzle
Expand Down
2 changes: 1 addition & 1 deletion src/models/granite-hybrid.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ void llama_model_granite_hybrid::load_arch_hparams(llama_model_loader & ml) {
case 768: type = LLM_TYPE_350M; break;
case 1536: type = (hparams.n_ff() == 512 ? LLM_TYPE_7B_A1B : LLM_TYPE_1B); break;
case 2048: case 2560: type = LLM_TYPE_3B; break;
case 4096: type = LLM_TYPE_32B; break;
case 4096: type = LLM_TYPE_32B_A9B; break;
default: type = LLM_TYPE_UNKNOWN;
}

Expand Down
3 changes: 1 addition & 2 deletions src/models/granite-moe.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,8 +9,7 @@ void llama_model_granite_moe::load_arch_hparams(llama_model_loader & ml) {

switch (hparams.n_layer()) {
case 24: type = LLM_TYPE_1B_A400M; break;
case 32: type = LLM_TYPE_3B; break;
case 40: type = LLM_TYPE_3B; break;
case 32: type = LLM_TYPE_3B_A800M; break;
// Add additional layer/vocab/etc checks here for other model sizes
default: type = LLM_TYPE_UNKNOWN;
}
Expand Down
11 changes: 10 additions & 1 deletion src/models/granite.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,16 @@ void llama_model_granite::load_arch_hparams(llama_model_loader & ml) {

switch (hparams.n_layer()) {
case 32: type = LLM_TYPE_3B; break;
case 40: type = LLM_TYPE_3B; break;
case 40: {
switch (hparams.n_embd) {
case 2048: type = LLM_TYPE_2B; break;
case 2560: type = LLM_TYPE_3B; break;
case 4096: type = LLM_TYPE_8B; break;
default: type = LLM_TYPE_UNKNOWN;
}
break;
}
case 64: type = LLM_TYPE_30B; break;
// Add additional layer/vocab/etc checks here for other model sizes
default: type = LLM_TYPE_UNKNOWN;
}
Expand Down
Loading