From 739880b79fe49cacfdae1c421bf7f12d38a8cd61 Mon Sep 17 00:00:00 2001 From: quidscio Date: Thu, 1 Jan 2026 23:03:15 -0500 Subject: [PATCH] fix: avoid Nemotron v2 MoE crash (llama.cpp #18309, 0b0ceb5) --- llama/llama.cpp/src/llama-model.cpp | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/llama/llama.cpp/src/llama-model.cpp b/llama/llama.cpp/src/llama-model.cpp index 00cd579e0..1e478640c 100644 --- a/llama/llama.cpp/src/llama-model.cpp +++ b/llama/llama.cpp/src/llama-model.cpp @@ -5196,9 +5196,6 @@ bool llama_model::load_tensors(llama_model_loader & ml) { const int64_t n_group = hparams.ssm_n_group; const int64_t d_in_proj = 2*d_inner + 2*n_group*d_state + n_ssm_head; - const int64_t n_ff_exp = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used; - const int64_t n_ff_shexp = hparams.n_ff_shexp; - // embeddings tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0); @@ -5250,6 +5247,9 @@ bool llama_model::load_tensors(llama_model_loader & ml) { layer.bo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "bias", i), {n_embd}, TENSOR_NOT_REQUIRED); } else { if (n_expert != 0) { + const int64_t n_ff_exp = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used; + const int64_t n_ff_shexp = hparams.n_ff_shexp; + layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), { n_embd, n_expert}, 0); layer.ffn_exp_probs_b = create_tensor(tn(LLM_TENSOR_FFN_EXP_PROBS_B, "bias", i), {n_expert }, 0);