Mohammed Innat

Results 89 issues of Mohammed Innat

This behaviour happens in Keras 2 but works in Keras 3. --- I tried to train a multi-output model. But it target looks like something as follows ```python y1_dummy =...

type:bug

Reopening from - https://github.com/keras-team/tf-keras/issues/138 - https://github.com/keras-team/tf-keras/issues/552 - https://github.com/tensorflow/tensorflow/issues/32809 - https://github.com/keras-team/keras/issues/12970 - https://github.com/keras-team/keras/issues/5625 - https://github.com/tensorflow/tensorflow/issues/39834 Currently the adopted solutions (from community) using `tf`. ```python import tensorflow as tf from tensorflow.python.profiler import...

type:feature

Keras-team Could you please take over the video swin checkponts and upload it to kaggle in order to make it usable in kaggle platform? 1. I have posted regarding weights....

**Short Description** By supporting [keras-v3](https://keras.io/keras_3/) the same codebases can be run on multiple backend, i.e. tensorflow, torch, and jax. **Other Information** - [ ] update packages from `tensorflow` to `keras-v3`....

enhancement
contribution-welcome
type:feature

**Short Description** Currently it supports only channel last format, i.e. `(bs, depth, height, width, channel)`. To make it compatible with `torch` ecosystem, it would be convenient to support channel last...

type:feature
contribution-welcome

**Short Description** Keras 3 introduces new preset format. ```python def vswin_tiny_backbone(): backbone=VideoSwinBackbone( input_shape=(32, 224, 224, 3), embed_dim=96, depths=[2, 2, 6, 2], num_heads=[3, 6, 12, 24], include_rescaling=False, ) backbone.load_weights( 'videoswin_tiny_kinetics400.weights.h5' )...

type:feature

It happens while using `remat`. ```python import keras from keras import layers import tensorflow as tf import numpy as np from keras import RematScope def with_remat(mode): with RematScope(mode=mode): base_model =...

stat:awaiting keras-eng
type:Bug

**Describe** FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness. Paper https://arxiv.org/abs/2205.14135 Cited by: 671 **Implementation** - PyTorch: https://github.com/Dao-AILab/flash-attention - Jax: https://github.com/lucidrains/flash-attention-jax - TensorFlow (with custom ops): https://github.com/intelligent-machine-learning/dlrover/tree/master/tfplus/tfplus/flash_attn Huggingface https://huggingface.co/docs/text-generation-inference/en/conceptual/flash_attention **Others**...

type:feature
stat:awaiting keras-eng
type:others

This was included in keras-cv. - https://github.com/keras-team/keras-cv/pull/434 - https://github.com/keras-team/keras-cv/pull/1979 [random-apply](https://github.com/keras-team/keras-cv/blob/94b0a551d03ee9de3d81663e2c1a680cb113f7f1/keras_cv/src/layers/preprocessing/random_apply.py#L22)

type:feature
stat:awaiting keras-eng

### Current Behavior: While trying to get P1, P2, P3 from EfficientNet, I got ```python backbone = EfficientNetV1B0Backbone(input_shape=(512, 512, 3)) for k, v in backbone.pyramid_level_inputs.items(): print(k, v, backbone.get_layer(name=v).output) P1 block1a_project_activation...

type:Bug
stat:awaiting response from contributor