embeddinggemma-2-GGUF / PROVENANCE.json
homerquan's picture
Add standard Q4_K_M text model and Q8_0 vision/audio companion
336c50f verified
Raw History Blame Contribute Delete
23 kB
{
"publisher": "MirrorNeuronLab",
"repository": "MirrorNeuronLab/embeddinggemma-2-GGUF",
"created_utc": "2026-11-01T15:08:34.179820+00:00",
"original_developer": "Google DeepMind",
"source": {
"repo_id": "google/embeddinggemma-2",
"revision": "f9972ac22e696a105c482202646979e33e4b5a1e",
"files": [
{
"path": ".gitattributes",
"size_bytes": 1301,
"sha256": "bcdccfaa66bc1240c7192dbd374cfb39b695bb1d8ff6b7ce0113aa5e9ac21a44",
"upstream_lfs_sha256": null
},
{
"path": "1_Pooling/config.json",
"size_bytes": 90,
"sha256": "8759bdf7c77efc7df7723f64856a593c8943b71ee38baf2a88771fbaf78438f9",
"upstream_lfs_sha256": null
},
{
"path": "2_Normalize/config.json",
"size_bytes": 97,
"sha256": "cdb09dfca347a56aa2d691744e38d5ad3c7cbc2834e7181272b9a15328b82524",
"upstream_lfs_sha256": null
},
{
"path": "README.md",
"size_bytes": 22234,
"sha256": "2c058c304c68221f1de80979594036eff3972b2bce686bb7b0f43621a6b0b011",
"upstream_lfs_sha256": null
},
{
"path": "chat_template.jinja",
"size_bytes": 1016,
"sha256": "4b852efc0b9960283e735363331e6f325b33bc74bdbaa076f595bc4e9b94d85e",
"upstream_lfs_sha256": null
},
{
"path": "config.json",
"size_bytes": 4455,
"sha256": "b8f1e9931b57fbc054acdb445c41765d55b0074c58d145fa82839941ad1b5bb3",
"upstream_lfs_sha256": null
},
{
"path": "config_sentence_transformers.json",
"size_bytes": 1565,
"sha256": "031e56a498d33c349ab489a21885bcfe25b4fcba841149dc99e1e90d4a7c28f5",
"upstream_lfs_sha256": null
},
{
"path": "model.safetensors",
"size_bytes": 1488915288,
"sha256": "197a32965d4b1105faf060417baa899e193fb73cd401f42ec9295234d5553d79",
"upstream_lfs_sha256": "197a32965d4b1105faf060417baa899e193fb73cd401f42ec9295234d5553d79"
},
{
"path": "modules.json",
"size_bytes": 413,
"sha256": "3d02572a0455b832de67fb8e63a54981bc7e8b46e337c95e917bd8122a533bfd",
"upstream_lfs_sha256": null
},
{
"path": "preprocessor_config.json",
"size_bytes": 511,
"sha256": "ea2ae257e901064abdd98dceb19f2b0da06af600bed15e0f99f5c85c37ee9d78",
"upstream_lfs_sha256": null
},
{
"path": "processor_config.json",
"size_bytes": 1788,
"sha256": "168f6a08522f3ce5dea596d94d003af2fd691742d4f41fe1f9d8cce76bfbf69c",
"upstream_lfs_sha256": null
},
{
"path": "sentence_bert_config.json",
"size_bytes": 747,
"sha256": "b1bcd9f2dce3ae863b359e87d0710b5dbc3314a59ecb4e2f97c7778fc8e4b228",
"upstream_lfs_sha256": null
},
{
"path": "tokenizer.json",
"size_bytes": 32170510,
"sha256": "4d777ef5bdc1aa36227abdfb77c3e49e7b9c892d16e1b6bda41c393504828be4",
"upstream_lfs_sha256": "4d777ef5bdc1aa36227abdfb77c3e49e7b9c892d16e1b6bda41c393504828be4"
},
{
"path": "tokenizer.model",
"size_bytes": 4689013,
"sha256": "e594c8a90eb08d8bda498ff4747977dc827ae0c3c56b5c0d41a605a22d02ef03",
"upstream_lfs_sha256": "e594c8a90eb08d8bda498ff4747977dc827ae0c3c56b5c0d41a605a22d02ef03"
},
{
"path": "tokenizer_config.json",
"size_bytes": 1599,
"sha256": "17bd5d6e9364ca49a534e1502076593317c298d4a663623091ed45388f004874",
"upstream_lfs_sha256": null
}
]
},
"llama_cpp": {
"repository": "https://github.com/ggml-org/llama.cpp",
"revision": "dde879bc4f954b809892fd0d660515f42536394a",
"working_tree_clean": true
},
"conversion": {
"intermediate": "BF16",
"text_quantization": "Q4_K_M",
"importance_matrix": null,
"additional_training": false,
"companion_quantization": "Q8_0",
"companion_modalities": [
"vision",
"audio"
],
"tokenizer_compatibility": "Transformers 5 config keys mapped in memory to Transformers 4.57.6 arguments; all 262144 token strings and IDs checked against tokenizer.json; source and llama.cpp files unchanged.",
"adapter_sha256": "9591828d8b0bfa92da1725c2dc27c2bc01e8b6a47cb7402afa6c26ed3c114f6d"
},
"files": [
{
"path": "work/embeddinggemma-2-BF16.gguf",
"size_bytes": 557950176,
"sha256": "68bae29d62fb8c7d23e98d21fd4662753ddd636e6b62b8a70dfe059a9844f216",
"tensor_count": 413,
"tensor_types": {
"BF16": 219,
"F32": 194
},
"tensor_prefixes": {
"token_embd": 1,
"output": 1,
"blk": 408,
"output_norm": 1,
"per_layer_model_proj": 1,
"per_layer_proj_norm": 1
},
"metadata": {
"general.architecture": "gemma-embedding2",
"general.type": "model",
"general.name": "embeddinggemma-2",
"general.size_label": "271M",
"general.license": "apache-2.0",
"general.tags": [
"embedding",
"feature-extraction",
"sentence-transformers",
"multimodal-embedding",
"multimodal",
"vision",
"audio",
"video",
"image-feature-extraction",
"audio-feature-extraction",
"video-feature-extraction",
"sentence-similarity",
"feature-extraction"
],
"general.languages": [
"multilingual"
],
"gemma-embedding2.block_count": 24,
"gemma-embedding2.context_length": 262144,
"gemma-embedding2.embedding_length": 512,
"gemma-embedding2.feed_forward_length": 2048,
"gemma-embedding2.attention.head_count": 4,
"gemma-embedding2.attention.head_count_kv": [
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1
],
"gemma-embedding2.rope.freq_base": 1000000.0,
"gemma-embedding2.rope.freq_base_swa": 10000.0,
"gemma-embedding2.attention.layer_norm_rms_epsilon": 9.999999974752427e-07,
"gemma-embedding2.attention.key_length": 512,
"gemma-embedding2.attention.value_length": 512,
"general.file_type": 32,
"gemma-embedding2.attention.sliding_window": 1024,
"gemma-embedding2.attention.shared_kv_layers": 0,
"gemma-embedding2.embedding_length_per_layer_input": 512,
"gemma-embedding2.attention.sliding_window_pattern": [
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false
],
"gemma-embedding2.attention.key_length_swa": 256,
"gemma-embedding2.attention.value_length_swa": 256,
"gemma-embedding2.rope.dimension_count": 512,
"gemma-embedding2.rope.dimension_count_swa": 256,
"gemma-embedding2.embedding_length_out": 768,
"gemma-embedding2.attention.causal": false,
"gemma-embedding2.pooling_type": 1,
"general.quantization_version": 2
}
},
{
"path": "release/embeddinggemma-2-Q4_K_M.gguf",
"size_bytes": 181694688,
"sha256": "d610afe26dd783739275646d51d4d85b4167820fd051600696d7a49e6f68609f",
"tensor_count": 413,
"tensor_types": {
"Q6_K": 25,
"F32": 194,
"BF16": 1,
"Q4_K": 193
},
"tensor_prefixes": {
"output": 1,
"output_norm": 1,
"per_layer_model_proj": 1,
"per_layer_proj_norm": 1,
"token_embd": 1,
"blk": 408
},
"metadata": {
"general.architecture": "gemma-embedding2",
"general.type": "model",
"general.name": "embeddinggemma-2",
"general.size_label": "271M",
"general.license": "apache-2.0",
"general.tags": [
"embedding",
"feature-extraction",
"sentence-transformers",
"multimodal-embedding",
"multimodal",
"vision",
"audio",
"video",
"image-feature-extraction",
"audio-feature-extraction",
"video-feature-extraction",
"sentence-similarity",
"feature-extraction"
],
"general.languages": [
"multilingual"
],
"gemma-embedding2.block_count": 24,
"gemma-embedding2.context_length": 262144,
"gemma-embedding2.embedding_length": 512,
"gemma-embedding2.feed_forward_length": 2048,
"gemma-embedding2.attention.head_count": 4,
"gemma-embedding2.attention.head_count_kv": [
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1,
2,
2,
2,
2,
2,
1
],
"gemma-embedding2.rope.freq_base": 1000000.0,
"gemma-embedding2.rope.freq_base_swa": 10000.0,
"gemma-embedding2.attention.layer_norm_rms_epsilon": 9.999999974752427e-07,
"gemma-embedding2.attention.key_length": 512,
"gemma-embedding2.attention.value_length": 512,
"gemma-embedding2.attention.sliding_window": 1024,
"gemma-embedding2.attention.shared_kv_layers": 0,
"gemma-embedding2.embedding_length_per_layer_input": 512,
"gemma-embedding2.attention.sliding_window_pattern": [
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false,
true,
true,
true,
true,
true,
false
],
"gemma-embedding2.attention.key_length_swa": 256,
"gemma-embedding2.attention.value_length_swa": 256,
"gemma-embedding2.rope.dimension_count": 512,
"gemma-embedding2.rope.dimension_count_swa": 256,
"gemma-embedding2.embedding_length_out": 768,
"gemma-embedding2.attention.causal": false,
"gemma-embedding2.pooling_type": 1,
"general.quantization_version": 2,
"general.file_type": 15
}
},
{
"path": "release/mmproj-embeddinggemma-2-Q8_0.gguf",
"size_bytes": 554821056,
"sha256": "0531309472a6bc6f4b146c3736eee7795ec9db1903c25490d3c937490cef6dd8",
"tensor_count": 963,
"tensor_types": {
"F32": 716,
"Q8_0": 247
},
"tensor_prefixes": {
"a": 751,
"mm": 2,
"v": 210
},
"metadata": {
"general.architecture": "clip",
"general.type": "mmproj",
"general.name": "embeddinggemma-2",
"general.size_label": "473M",
"general.license": "apache-2.0",
"general.tags": [
"embedding",
"feature-extraction",
"sentence-transformers",
"multimodal-embedding",
"multimodal",
"vision",
"audio",
"video",
"image-feature-extraction",
"audio-feature-extraction",
"video-feature-extraction",
"sentence-similarity",
"feature-extraction"
],
"general.languages": [
"multilingual"
],
"general.file_type": 7,
"clip.has_vision_encoder": true,
"clip.vision.projection_dim": 512,
"clip.vision.image_size": 224,
"clip.vision.patch_size": 16,
"clip.vision.embedding_length": 768,
"clip.vision.feed_forward_length": 3072,
"clip.vision.block_count": 16,
"clip.vision.attention.head_count": 12,
"clip.vision.image_mean": [
0.0,
0.0,
0.0
],
"clip.vision.image_std": [
1.0,
1.0,
1.0
],
"clip.has_audio_encoder": true,
"clip.audio.projection_dim": 512,
"clip.audio.embedding_length": 1024,
"clip.audio.feed_forward_length": 4096,
"clip.audio.block_count": 12,
"clip.audio.attention.head_count": 8,
"clip.vision.projector_type": "gemma4v",
"clip.vision.attention.layer_norm_epsilon": 9.999999974752427e-07,
"clip.use_gelu": true,
"clip.audio.projector_type": "gemma4a",
"clip.audio.num_mel_bins": 128,
"clip.audio.attention.layer_norm_epsilon": 9.999999974752427e-07,
"general.quantization_version": 2
}
}
],
"machine": {
"name": "NVIDIA DGX Spark",
"architecture": "aarch64",
"gpu": "NVIDIA GB10",
"unified_memory": "128 GB nominal",
"python": "3.12.3",
"compiler": "GCC 13.3.0",
"cmake": "3.28.3",
"cuda_toolkit": "13.0.88",
"nvidia_driver": "580.178.04"
},
"docker": {
"engine": "29.6.2",
"client": "v1.2.9",
"server": "v1.2.9",
"image": "docker/model-runner@sha256:9967b6c1092ff01e7cf68119ab25067a1f58e3d1b06873bc16fe1e758bcd914f",
"llama_cpp_revision": "dde879bc4f954b809892fd0d660515f42536394a",
"llama_cpp_build": "b12032 / 0.8.0-dev",
"cuda_architecture": "121",
"local_model": "mirrorneuronlab/embeddinggemma-2:q4_k_m",
"endpoint": "http://127.0.0.1:12435",
"configuration": "-ngl 999 --pooling mean -b 2048 -ub 2048 --parallel 1 -t 10; context 2048; embedding mode",
"gpu_evidence": "Docker logged gpuSupport=true, backend arguments include -ngl 999, NVIDIA-SMI reports /opt/llama/llama-server GPU processes.",
"backend_version_note": "The stock images checked bundled older engines without this architecture. Validation used the official current v1.2.9 controller with the pinned upstream CUDA backend mounted read-only. The controller displays the backend version as unknown because its parser does not recognize llama.cpp's new version string."
},
"execution_timings": [
{
"stage": "download",
"command": [
"python3",
"scripts/download_source.py"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:55:21.409343+00:00",
"finished_utc": "2026-11-01T14:55:53.401870+00:00",
"elapsed_seconds": 31.99252673998126,
"returncode": 0
},
{
"stage": "build-configure",
"command": [
"cmake",
"-S",
"llama.cpp",
"-B",
"llama.cpp/build",
"-DCMAKE_BUILD_TYPE=Release",
"-DGGML_CUDA=OFF",
"-DGGML_NATIVE=ON",
"-DLLAMA_BUILD_TESTS=OFF",
"-DLLAMA_CURL=OFF"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:55:46.632685+00:00",
"finished_utc": "2026-11-01T14:55:48.362987+00:00",
"elapsed_seconds": 1.73029996099649,
"returncode": 0
},
{
"stage": "build",
"command": [
"cmake",
"--build",
"llama.cpp/build",
"--config",
"Release",
"-j",
"16",
"--target",
"llama-quantize",
"llama-server"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:55:48.380276+00:00",
"finished_utc": "2026-11-01T14:56:48.044752+00:00",
"elapsed_seconds": 59.66447541900561,
"returncode": 0
},
{
"stage": "dependencies",
"command": [
".venv/bin/python",
"-m",
"pip",
"install",
"-r",
"llama.cpp/requirements/requirements-convert_hf_to_gguf.txt",
"huggingface_hub>=0.34.0,<1.0"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:55:49.275061+00:00",
"finished_utc": "2026-11-01T14:56:14.893959+00:00",
"elapsed_seconds": 25.618897876003757,
"returncode": 0
},
{
"stage": "dmr-image-pull",
"command": [
"docker",
"pull",
"docker/model-runner@sha256:9967b6c1092ff01e7cf68119ab25067a1f58e3d1b06873bc16fe1e758bcd914f"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:56:23.532588+00:00",
"finished_utc": "2026-11-01T14:56:27.210774+00:00",
"elapsed_seconds": 3.6781821949989535,
"returncode": 0
},
{
"stage": "convert-bf16",
"command": [
".venv/bin/python",
"llama.cpp/convert_hf_to_gguf.py",
"source",
"--outtype",
"bf16",
"--model-name",
"embeddinggemma-2",
"--outfile",
"work/embeddinggemma-2-BF16.gguf"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:56:49.790088+00:00",
"finished_utc": "2026-11-01T14:56:53.358771+00:00",
"elapsed_seconds": 3.5686840260168537,
"returncode": 1
},
{
"stage": "convert-mmproj",
"command": [
".venv/bin/python",
"llama.cpp/convert_hf_to_gguf.py",
"source",
"--mmproj",
"--outtype",
"q8_0",
"--model-name",
"embeddinggemma-2",
"--outfile",
"release/mmproj-embeddinggemma-2-Q8_0.gguf"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:56:50.417675+00:00",
"finished_utc": "2026-11-01T14:56:53.964458+00:00",
"elapsed_seconds": 3.54678301900276,
"returncode": 0
},
{
"stage": "cuda-configure",
"command": [
"cmake",
"-S",
"llama.cpp",
"-B",
"llama.cpp/build-cuda",
"-DCMAKE_BUILD_TYPE=Release",
"-DGGML_CUDA=ON",
"-DCMAKE_CUDA_ARCHITECTURES=121",
"-DGGML_NATIVE=ON",
"-DLLAMA_BUILD_TESTS=OFF",
"-DLLAMA_CURL=OFF"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:57:33.645338+00:00",
"finished_utc": "2026-11-01T14:57:36.272337+00:00",
"elapsed_seconds": 2.626998322986765,
"returncode": 0
},
{
"stage": "cuda-build",
"command": [
"cmake",
"--build",
"llama.cpp/build-cuda",
"--config",
"Release",
"-j",
"8",
"--target",
"llama-server"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:57:36.303210+00:00",
"finished_utc": "2026-11-01T15:01:19.019507+00:00",
"elapsed_seconds": 222.71629584900802,
"returncode": 0
},
{
"stage": "convert-bf16-compatible",
"command": [
".venv/bin/python",
"scripts/convert_compatible.py",
"source",
"--outtype",
"bf16",
"--model-name",
"embeddinggemma-2",
"--outfile",
"work/embeddinggemma-2-BF16.gguf"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:58:26.671428+00:00",
"finished_utc": "2026-11-01T14:58:39.271335+00:00",
"elapsed_seconds": 12.599901882989798,
"returncode": 0
},
{
"stage": "quantize",
"command": [
"llama.cpp/build/bin/llama-quantize",
"work/embeddinggemma-2-BF16.gguf",
"release/embeddinggemma-2-Q4_K_M.gguf",
"Q4_K_M"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:59:03.341119+00:00",
"finished_utc": "2026-11-01T14:59:05.780096+00:00",
"elapsed_seconds": 2.438970130024245,
"returncode": 0
},
{
"stage": "audit",
"command": [
".venv/bin/python",
"scripts/audit_gguf.py"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T14:59:48.642886+00:00",
"finished_utc": "2026-11-01T15:00:09.839507+00:00",
"elapsed_seconds": 21.196617795998463,
"returncode": 1
},
{
"stage": "audit-final",
"command": [
".venv/bin/python",
"scripts/audit_gguf.py"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:00:53.569576+00:00",
"finished_utc": "2026-11-01T15:01:15.285176+00:00",
"elapsed_seconds": 21.71559885001625,
"returncode": 0
},
{
"stage": "validate-bf16",
"command": [
".venv/bin/python",
"scripts/validate_bf16.py"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:01:15.314013+00:00",
"finished_utc": "2026-11-01T15:01:17.519620+00:00",
"elapsed_seconds": 2.2056051869876683,
"returncode": 0
},
{
"stage": "dmr-start",
"command": [
"bash",
"scripts/start_validation_runner.sh"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:02:00.812758+00:00",
"finished_utc": "2026-11-01T15:02:01.138690+00:00",
"elapsed_seconds": 0.32593187599559315,
"returncode": 0
},
{
"stage": "dmr-package",
"command": [
"env",
"MODEL_RUNNER_HOST=http://127.0.0.1:12435",
"docker",
"model",
"package",
"--gguf",
"$TASK_ROOT/release/embeddinggemma-2-Q4_K_M.gguf",
"--mmproj",
"$TASK_ROOT/release/mmproj-embeddinggemma-2-Q8_0.gguf",
"--license",
"$TASK_ROOT/release/LICENSE",
"--context-size",
"2048",
"mirrorneuronlab/embeddinggemma-2:q4_k_m"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:02:13.051800+00:00",
"finished_utc": "2026-11-01T15:02:15.412325+00:00",
"elapsed_seconds": 2.3605253199930303,
"returncode": 0
},
{
"stage": "validate-dmr",
"command": [
".venv/bin/python",
"scripts/validate_dmr.py"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:02:15.594479+00:00",
"finished_utc": "2026-11-01T15:02:17.820825+00:00",
"elapsed_seconds": 2.226341144007165,
"returncode": 0
},
{
"stage": "dmr-run",
"command": [
"env",
"MODEL_RUNNER_HOST=http://127.0.0.1:12435",
"docker",
"model",
"run",
"--detach",
"mirrorneuronlab/embeddinggemma-2:q4_k_m"
],
"cwd": "$TASK_ROOT",
"started_utc": "2026-11-01T15:05:52.619080+00:00",
"finished_utc": "2026-11-01T15:05:52.670364+00:00",
"elapsed_seconds": 0.0512839280127082,
"returncode": 0
}
],
"validation_fixtures": [
{
"name": "red-square.png",
"sha256": "66d2ca44529ac40d3d08a58d0fd0f10696eae138d288dc5cfc1351f263d4c0a3",
"size_bytes": 677
},
{
"name": "tone-440hz.wav",
"sha256": "3f806962fcdcc982acaca0e5180c701391a59c2635d82e4b0660dc148c01f528",
"size_bytes": 64044
}
]
}