Instructions to use litert-community/TIPSv2-B14-LiteRT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LiteRT
How to use litert-community/TIPSv2-B14-LiteRT with LiteRT:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Download quantization.json from litert-community/TIPSv2-B14-LiteRT: direct link, hf CLI and curl.
- Browser
- Download file 2.4 kB
-
https://huggingface.co/litert-community/TIPSv2-B14-LiteRT/resolve/main/quantization.json
- Command line
-
hf download hf://litert-community/TIPSv2-B14-LiteRT/quantization.json
-
curl -L -o quantization.json https://huggingface.co/litert-community/TIPSv2-B14-LiteRT/resolve/main/quantization.json
2.4 kB
| { | |
| "224": { | |
| "provenance": { | |
| "model": "google/tipsv2-b14@ed1e4dc6b74bf3935ae099e9d5eb30fa96528454", | |
| "task": "image-encoder", | |
| "image_size": 224, | |
| "calibration_npy_sha256": "8b2056a4e49b11429475d3e7875a3e8170fb5e7a960105bd88a6ae37158fdcff", | |
| "evaluation_npy_sha256": "50a43ede0096253b921b783dcf6b29faa8e7fac5335ce0465d51e0d6b5d391ca", | |
| "calibration_samples": 500, | |
| "fp32_source_sha256": "9805c17592169a2b714a8d92292b8c37cba812b4820cbfb49f11526e2ef71810", | |
| "versions": { | |
| "torch": "2.14.1+cu130", | |
| "torchao": "0.18.0+cu130", | |
| "transformers": "5.21.0", | |
| "ai-edge-litert": "2.3.0" | |
| }, | |
| "calibration_device": "cuda", | |
| "cuda_tf32": false, | |
| "activation_observer": "TorchAO asymmetric per-tensor min/max", | |
| "weight_encoding": "symmetric INT8 [-127,127], per output channel", | |
| "output_ranges": "FP32 projection outputs after LayerScale folding", | |
| "gpu_compat": false | |
| }, | |
| "checkpoint_sha256": "c6aee4e8910a0d285f2a134dcfae7c5c88aca6ce0e0da3fe0eb534425808e9f8", | |
| "activation_bits": 16, | |
| "accumulator_bits": 32, | |
| "region_activation_bits": 16, | |
| "public_output_quantization": true | |
| }, | |
| "448": { | |
| "provenance": { | |
| "model": "google/tipsv2-b14@ed1e4dc6b74bf3935ae099e9d5eb30fa96528454", | |
| "task": "image-encoder", | |
| "image_size": 448, | |
| "calibration_npy_sha256": "dbfbb90e3ab2d626616bc98def2f3b36d643aaf46414c3ee5e2cafbf2f3b3b9e", | |
| "evaluation_npy_sha256": "ae027374527a30e04a0cb8f4356ba6ce27b59e6017e6de035b8c187c8f87bc42", | |
| "calibration_samples": 500, | |
| "fp32_source_sha256": "66461a9660294a3359eed8410d38ba72d36cf337991751db589025dcb4620e83", | |
| "versions": { | |
| "torch": "2.14.1+cu130", | |
| "torchao": "0.18.0+cu130", | |
| "transformers": "5.21.0", | |
| "ai-edge-litert": "2.3.0" | |
| }, | |
| "calibration_device": "cuda", | |
| "cuda_tf32": false, | |
| "activation_observer": "TorchAO asymmetric per-tensor min/max", | |
| "weight_encoding": "symmetric INT8 [-127,127], per output channel", | |
| "output_ranges": "FP32 projection outputs after LayerScale folding", | |
| "gpu_compat": false | |
| }, | |
| "checkpoint_sha256": "690ab47e34d20ed494e8303a4afcdda6ea630a44c1eeebb64ff3aa96e52d8f0a", | |
| "activation_bits": 16, | |
| "accumulator_bits": 32, | |
| "region_activation_bits": 16, | |
| "public_output_quantization": true | |
| } | |
| } | |