Instructions to use zeromodels/deberta_base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- ZeroModels
How to use zeromodels/deberta_base with ZeroModels:
# pip install -U zeromodels # ZeroModels is pure Keras 3, so pick a backend: "jax", "torch" or "tensorflow". import os os.environ["KERAS_BACKEND"] = "jax" from zeromodels import AutoZModel # AutoZModel reads the repo's model_type and loads the matching class. # For a task head use the matching loader, e.g. AutoZMImageClassify / AutoZMDetect / # AutoZMSemanticSegment / AutoZMTextGenerate (see zeromodels.auto). model = AutoZModel.from_weights("zeromodels/deberta_base") - Keras
How to use zeromodels/deberta_base with Keras:
# !pip install -U keras tensorflow huggingface_hub # Keras needs TensorFlow installed to read "hf://" paths, so the tensorflow backend is selected here; # "jax" and "torch" also work for computation once TensorFlow is installed. import os os.environ["KERAS_BACKEND"] = "tensorflow" import keras model = keras.saving.load_model("hf://zeromodels/deberta_base") - Notebooks
- Google Colab
- Kaggle
Migrate to zeromodels (rename kf_*.json -> zm_*.json, fix refs in config + README, ensure tag + badge)
24db011 verified Download zm_config.json from zeromodels/deberta_base: direct link, hf CLI and curl.
- Browser
- Download file 625 Bytes
-
https://huggingface.co/zeromodels/deberta_base/resolve/main/zm_config.json
- Command line
-
hf download hf://zeromodels/deberta_base/zm_config.json
-
curl -L -o zm_config.json https://huggingface.co/zeromodels/deberta_base/resolve/main/zm_config.json
625 Bytes
| { | |
| "library_name": "zeromodels", | |
| "zeromodels_version": "1.2.1", | |
| "model_module": "zeromodels.models.deberta", | |
| "model_class": "DebertaModel", | |
| "variant": "deberta_base", | |
| "weights": "model.weights.h5", | |
| "schema_version": 2, | |
| "weight_dtype": "float32", | |
| "model_type": "deberta", | |
| "text_config": { | |
| "vocab_size": 50265, | |
| "embed_dim": 768, | |
| "num_layers": 12, | |
| "num_heads": 12, | |
| "mlp_dim": 3072, | |
| "max_position_embeddings": 512, | |
| "max_relative_positions": 512, | |
| "pos_att_type": [ | |
| "c2p", | |
| "p2c" | |
| ], | |
| "hidden_act": "gelu", | |
| "layer_norm_eps": 1e-07, | |
| "pad_token_id": 0 | |
| } | |
| } |