Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-17 16:25:01 +02:00
Code Issues Packages Projects Releases Wiki Activity

revert qwen vl warmup size

Browse Source
This commit is contained in:
LostRuins Concedo
2025-11-09 02:24:49 +08:00
parent d6a2ad8455
commit 4fc022a51f
1 changed files with 11 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/mtmd/clip.cpp
+11 -9
View File
@@ -2872,17 +2872,19 @@ struct clip_model_loader {
get_u32(KEY_SPATIAL_MERGE_SIZE, hparams.n_merge, false);
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
if (model.proj_type==PROJECTOR_TYPE_QWEN25VL && q25vl_migrated) {
if (model.proj_type==PROJECTOR_TYPE_QWEN25VL && q25vl_migrated) {
hparams.n_wa_pattern = 8;
}
hparams.set_limit_image_tokens(8, 4096);
hparams.set_warmup_n_tokens(46*46); // avoid OOM on warmup
const int warn_min_pixels = 1024 * hparams.n_merge * hparams.n_merge * hparams.patch_size * hparams.patch_size;
if (hparams.image_min_pixels < warn_min_pixels) {
LOG_WRN("%s: Qwen-VL models require at minimum 1024 image tokens to function correctly on grounding tasks\n", __func__);
LOG_WRN("%s: if you encounter problems with accuracy, try adding --image-min-tokens 1024\n", __func__);
LOG_WRN("%s: more info: https://github.com/ggml-org/llama.cpp/issues/16842\n\n", __func__);
}
// hparams.set_limit_image_tokens(8, 4096);
// hparams.set_warmup_n_tokens(46*46); // avoid OOM on warmup
hparams.set_limit_image_tokens(8, 2048);
hparams.set_warmup_n_tokens(256); // kcpp revert qwen vision size
// const int warn_min_pixels = 1024 * hparams.n_merge * hparams.n_merge * hparams.patch_size * hparams.patch_size;
// if (hparams.image_min_pixels < warn_min_pixels) {
// LOG_WRN("%s: Qwen-VL models require at minimum 1024 image tokens to function correctly on grounding tasks\n", __func__);
// LOG_WRN("%s: if you encounter problems with accuracy, try adding --image-min-tokens 1024\n", __func__);
// LOG_WRN("%s: more info: https://github.com/ggml-org/llama.cpp/issues/16842\n\n", __func__);
// }
} break;
case PROJECTOR_TYPE_LLAMA4:
{