ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

llama : fix OLMo-2-0325-32B-Instruct K-norm size (#12400)

Browse source
This commit is contained in:
Sigbjørn Skjæret 2025-03-16 18:46:36 +01:00 • committed by GitHub
parent dc079cfdff
commit 8ba95dca20
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 4 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

5
src/llama-model.cpp
View file

@ -1005,6 +1005,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
case 16: type = LLM_TYPE_1B; break; case 16: type = LLM_TYPE_1B; break;
case 32: type = LLM_TYPE_7B; break; case 32: type = LLM_TYPE_7B; break;
case 40: type = LLM_TYPE_13B; break; case 40: type = LLM_TYPE_13B; break;
case 64: type = LLM_TYPE_32B; break;
default: type = LLM_TYPE_UNKNOWN; default: type = LLM_TYPE_UNKNOWN;
} }
} break; } break;
@ -2726,6 +2727,8 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
} break; } break;
case LLM_ARCH_OLMO2: case LLM_ARCH_OLMO2:
{ {
const int64_t n_embd_head = n_embd / n_head;
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0); tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
// output // output
@ -2740,7 +2743,7 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
layer.wv = create_tensor(tn(LLM_TENSOR_ATTN_V, "weight", i), {n_embd, n_embd_gqa}, 0); layer.wv = create_tensor(tn(LLM_TENSOR_ATTN_V, "weight", i), {n_embd, n_embd_gqa}, 0);
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd, n_embd}, 0); layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd, n_embd}, 0);
layer.attn_q_norm = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM, "weight", i), {n_embd}, 0); layer.attn_q_norm = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM, "weight", i), {n_embd}, 0);
layer.attn_k_norm = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM, "weight", i), {n_embd}, 0); layer.attn_k_norm = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM, "weight", i), {n_head_kv * n_embd_head}, 0);
layer.attn_post_norm = create_tensor(tn(LLM_TENSOR_ATTN_POST_NORM, "weight", i), {n_embd}, 0); layer.attn_post_norm = create_tensor(tn(LLM_TENSOR_ATTN_POST_NORM, "weight", i), {n_embd}, 0);
layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0); layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);