From 08f49598c189d7fbf42bcbdc62ca3b3bf1758771 Mon Sep 17 00:00:00 2001 From: Pascal Date: Thu, 12 Nov 2026 12:14:10 +0100 Subject: [PATCH] mtmd: require a 2x2 spatial merge for the Qwen-VL projectors (#32130) The Qwen2-VL, Qwen2.5-VL, Qwen3-VL and Ling3-VL graphs build their input with a temporal merge that assumes a 2x2 patch grid, and the Qwen2-VL merger reshapes by 4. Any other spatial_merge_size from the GGUF reaches a GGML_ASSERT at the first image request and aborts the server. Reject a spatial_merge_size other than 2 when loading the hparams. --- tools/mtmd/mtmd-model.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tools/mtmd/mtmd-model.cpp b/tools/mtmd/mtmd-model.cpp index 8a76e11b5..e00043818 100644 --- a/tools/mtmd/mtmd-model.cpp +++ b/tools/mtmd/mtmd-model.cpp @@ -662,6 +662,10 @@ struct mtmd_model_loader { hparams.image_resize_pad = PAD_NONE; // the HF processor resizes to the smart_resize size, no padding hparams.image_round_half_even = true; // smart_resize uses Python round() get_u32(KEY_SPATIAL_MERGE_SIZE, hparams.n_merge, false); + if (hparams.n_merge != 2) { + // the temporal merge input and the merger use a 2x2 patch grid + throw std::runtime_error(mtmd_format("%s: spatial_merge_size (%d) must be 2\n", __func__, hparams.n_merge)); + } get_u32(KEY_WIN_ATTN_PATTERN, hparams.n_wa_pattern, model.proj_type == PROJECTOR_TYPE_QWEN25VL); // only 2.5 requires it // ref: https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct/blob/main/preprocessor_config.json hparams.set_limit_image_tokens(8, 4096);