From 7f51953d7d2f359939a763549e93d51f0347e569 Mon Sep 17 00:00:00 2001 From: Oliver Simons Date: Tue, 13 Oct 2026 12:34:47 +0200 Subject: [PATCH] model : allow Qwen3.8 Flash Next to accept image embds for MTP (#30397) Cont of #30257, closes #30307 --- src/models/qwen4exp.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/models/qwen4exp.cpp b/src/models/qwen4exp.cpp index c1771fcad..1d151dc21 100644 --- a/src/models/qwen4exp.cpp +++ b/src/models/qwen4exp.cpp @@ -526,7 +526,7 @@ llama_model_qwen4exp::graph::graph(const llama_model & model, const llm_graph_pa llama_model_qwen4exp::graph_mtp::graph_mtp(const llama_model & model, const llm_graph_params & params) : graph(model, params, no_build{}) { GGML_ASSERT(hparams.n_layer_nextn == 1 && "qwen4exp MTP has a single block"); - GGML_ASSERT(ubatch.token && "qwen4exp MTP requires token input"); + ASSERT_EMBD_OR_TOKEN(ubatch); const int64_t hc = hparams.dsv4_hc_mult; GGML_ASSERT(hparams.n_embd_out() == (uint32_t) (n_embd*hc) && "qwen4exp MTP hidden width mismatch");