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

mtmd : add vision support for llama 4 (#13282)

Browse source 
* wip llama 4 conversion

* rm redundant __init__

* fix conversion

* fix conversion

* test impl

* try this

* reshape patch_embeddings_0

* fix view

* rm ffn_post_norm

* cgraph ok

* f32 for pos embd

* add image marker tokens

* Llama4UnfoldConvolution

* correct pixel shuffle

* fix merge conflicts

* correct

* add debug_graph

* logits matched, but it still preceives the image incorrectly

* fix style

* add image_grid_pinpoints

* handle llama 4 preprocessing

* rm load_image_size

* rm unused line

* fix

* small fix 2

* add test & docs

* fix llava-1.6 test

* test: add notion of huge models

* add comment

* add warn about degraded quality
This commit is contained in:
Xuan-Son Nguyen 2025-05-19 13:04:14 +02:00 • committed by GitHub
parent f71f40a284
commit 92ecdcc06a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
9 changed files with 424 additions and 82 deletions
Download patch file Download diff file Expand all files Collapse all files

22
tools/mtmd/tests.sh
View file

@ -21,6 +21,13 @@ if [ "${1:-}" = "big" ]; then
echo "Include BIG models..."
fi
RUN_HUGE_TESTS=false
if [ "${1:-}" = "huge" ]; then
RUN_HUGE_TESTS=true
RUN_BIG_TESTS=true
echo "Include BIG models..."
fi
###############
arr_bin=()
@ -42,7 +49,7 @@ add_test "llama-mtmd-cli" "ggml-org/SmolVLM2-500M-Video-Instruct-GGUF:Q8_0"
add_test "llama-mtmd-cli" "ggml-org/gemma-3-4b-it-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "THUDM/glm-edge-v-5b-gguf:Q4_K_M"
add_test "llama-mtmd-cli" "second-state/Llava-v1.5-7B-GGUF:Q2_K" "vicuna"
add_test "llama-mtmd-cli" "cjpais/llava-1.6-mistral-7b-gguf:Q3_K" "vicuna"
add_test "llama-mtmd-cli" "cjpais/llava-1.6-mistral-7b-gguf:Q3_K_M" "vicuna"
add_test "llama-mtmd-cli" "ibm-research/granite-vision-3.2-2b-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "second-state/MiniCPM-Llama3-V-2_5-GGUF:Q2_K" # model from openbmb is corrupted
add_test "llama-mtmd-cli" "openbmb/MiniCPM-V-2_6-gguf:Q2_K"
@ -60,10 +67,17 @@ if [ "$RUN_BIG_TESTS" = true ]; then
add_test "llama-mtmd-cli" "ggml-org/Qwen2-VL-7B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/Qwen2.5-VL-3B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/Qwen2.5-VL-7B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/InternVL3-8B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/InternVL3-14B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/InternVL3-8B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/InternVL3-14B-Instruct-GGUF:Q4_K_M"
# add_test "llama-mtmd-cli" "ggml-org/Qwen2.5-VL-32B-Instruct-GGUF:Q4_K_M" # does not work on my mac M3 Ultra
# add_test "llama-mtmd-cli" "ggml-org/Qwen2.5-VL-72B-Instruct-GGUF:Q4_K_M" # too big
fi
# to test the huge models, run: ./tests.sh huge
# this will run both the big and huge models
# huge models are > 32B parameters
if [ "$RUN_HUGE_TESTS" = true ]; then
add_test "llama-mtmd-cli" "ggml-org/Qwen2.5-VL-72B-Instruct-GGUF:Q4_K_M"
add_test "llama-mtmd-cli" "ggml-org/Llama-4-Scout-17B-16E-Instruct-GGUF:IQ1_S"
fi
# these models always give the wrong answer, not sure why