keras-nlp
keras-nlp copied to clipboard
Modular Natural Language Processing workflows with Keras
Updating the implementation of mobilenet as per torch, and then will add the timm conversion script to port weights from hf
TODO: - [ ] add weights conversion script here - https://github.com/keras-team/keras-hub/tree/master/tools/checkpoint_conversion - [ ] Verify output with real inputs - [ ] provide a working colab demo Example PR -...
This task is part of the KerasHub project ( Consolidating all of KerasCV and KerasNLP models in to one place) Models from KerasCV will be added to KerasNLP in KerasNLP...
Draft PR for transparency. ### Done - [x] Basic components - [x] CIoU Loss - [x] ImageObjectDetector Task - [x] ImageObjectDetectorPreprocessor ### Planned - [] YOLOv11 Architecture - [] Object...
Add https://huggingface.co/black-forest-labs/FLUX.1-dev to KerasHub TODO: - [ ] Add backbone - [ ] add task model - [ ] add weights conversion script - [ ] demo colab to verify...
Trying to load llama-3.2 on TPU VM v3-8 via this: ```python device_mesh = keras.distribution.DeviceMesh((1, 8), ["batch", "model"], devices=keras.distribution.list_devices()) layout_map = keras.distribution.LayoutMap(device_mesh) layout_map["token_embedding/embeddings"] = ("model", None) layout_map["decoder_block.*attention.*(query|key|value)/kernel"] = ("model", None, None)...
This adds the Differntial Binarization model for text detection. Implemented the architecture based on [ResNet50_vd](https://github.com/PaddlePaddle/PaddleOCR/blob/main/doc/doc_en/algorithm_det_db_en.md) from PaddleOCR and ported the weights.
Here’s the updated content reflecting that KerasNLP is now KerasHub: --- ## tl;dr - We have consolidated KerasNLP and KerasCV into a new **KerasHub** package. - We have renamed the...
This PR ports the core model into a Keras model and includes a weight conversion script. VAE and rest of the pipeline would make sense in a separate PR. Each...