Feature Extraction
Transformers
Safetensors
lfm2
fill-mask
encoder-only
multimodal
image-text-retrieval
image-text-matching
siglip2
lfm2.5
gptq
custom_code
compressed-tensors
Instructions to use konic-labs/LFM2.5-multimodal-encoder-230M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use konic-labs/LFM2.5-multimodal-encoder-230M with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="konic-labs/LFM2.5-multimodal-encoder-230M", trust_remote_code=True)# Load model directly from transformers import AutoTokenizer, AutoModelForMaskedLM tokenizer = AutoTokenizer.from_pretrained("konic-labs/LFM2.5-multimodal-encoder-230M", trust_remote_code=True) model = AutoModelForMaskedLM.from_pretrained("konic-labs/LFM2.5-multimodal-encoder-230M", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 2,105 Bytes
ff3d404 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 | {
"model": "outputs/monet-100k/compression-gptq-int4-clean",
"vision_model": "google/siglip2-base-patch16-256",
"device": "cuda:0",
"dtype": "torch.float32",
"setup": {
"model_load_ms": 1752.9155970041757,
"vision_load_ms": 689.1005889992812
},
"model_parameter_count": 230963108,
"model_bytes": 370447122,
"pairs": "/data/datasets/monet-100k/test_pairs.jsonl",
"projector_checkpoint": "outputs/monet-100k/retrieval-matching-clean/multimodal_task_adapter.pt",
"text": {
"iterations": 10,
"warmup": 3,
"latency_ms_mean": 16.75748469933751,
"latency_ms_median": 16.738547001295956,
"latency_ms_p95": 17.114150999987032,
"latency_ms_min": 16.403875000833068,
"latency_ms_max": 17.172412997751962,
"peak_vram_bytes": 1049502720,
"batch_size": 4,
"sequence_length": 53,
"hidden_shape": [
4,
53,
1024
],
"logits_shape": [
4,
53,
65536
],
"finite_hidden": true,
"finite_logits": true
},
"multimodal": {
"body_only": {
"iterations": 10,
"warmup": 3,
"latency_ms_mean": 18.0679445998976,
"latency_ms_median": 18.043951498839306,
"latency_ms_p95": 18.232377005915623,
"latency_ms_min": 17.939090001164004,
"latency_ms_max": 18.2529970043106,
"peak_vram_bytes": 2531130880
},
"vision_plus_body": {
"iterations": 10,
"warmup": 3,
"latency_ms_mean": 45.672791600372875,
"latency_ms_median": 45.429460500599816,
"latency_ms_p95": 46.00368000683375,
"latency_ms_min": 44.96823700173991,
"latency_ms_max": 47.790171003725845,
"peak_vram_bytes": 2552086016
},
"vision_features_shape": [
4,
256,
768
],
"image_tokens_shape": [
4,
32,
1024
],
"fused_shape": [
4,
86,
1024
],
"hidden_shape": [
4,
86,
1024
],
"image_marker_position": 1,
"finite_hidden": true
},
"note": "Multimodal path uses the supplied trained projector; matching-head scoring is not timed."
}
|