Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 06:39:25 +01:00
Code Issues Packages Projects Releases Wiki Activity

Adding Gemma 2 2B configs (#8784)

Browse Source 
* Adding Gemma 2 2B configs

Updates to Q scaling and Gemma 2 model sizes to match v2 2B model.

* Update src/llama.cpp

Co-authored-by: slaren <slarengh@gmail.com>

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
pculliton 2024-07-31 11:12:10 -04:00 committed by GitHub
parent 44d28ddd5c
commit 398ede5efe
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/llama.cpp
View File

@ -4969,6 +4969,7 @@ static void llm_load_hparams(
hparams.attn_soft_cap = true; hparams.attn_soft_cap = true;
switch (hparams.n_layer) { switch (hparams.n_layer) {
case 26: model.type = e_model::MODEL_2B; break;
case 42: model.type = e_model::MODEL_9B; break; case 42: model.type = e_model::MODEL_9B; break;
case 46: model.type = e_model::MODEL_27B; break; case 46: model.type = e_model::MODEL_27B; break;
default: model.type = e_model::MODEL_UNKNOWN; default: model.type = e_model::MODEL_UNKNOWN;
@ -11736,6 +11737,7 @@ struct llm_build_context {
// ref: https://github.com/google/gemma_pytorch/commit/03e657582d17cb5a8617ebf333c1c16f3694670e // ref: https://github.com/google/gemma_pytorch/commit/03e657582d17cb5a8617ebf333c1c16f3694670e
switch (model.type) { switch (model.type) {
case e_model::MODEL_2B:
case e_model::MODEL_9B: Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd_head_k))); break; case e_model::MODEL_9B: Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd_head_k))); break;
case e_model::MODEL_27B: Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd / n_head))); break; case e_model::MODEL_27B: Qcur = ggml_scale(ctx0, Qcur, 1.0f / sqrtf(float(n_embd / n_head))); break;
default: GGML_ABORT("fatal error"); default: GGML_ABORT("fatal error");