Here is the rewritten article:
Introduction
The Hugging Face Hub, a vast repository of over 750K public models, has been continuously growing since its inception, offering diverse pre-trained models for various machine learning frameworks. As technology advances, the community increasingly demands interoperability across tools and formats. With significant developments happening in this aspect, users of the Transformer library will now have better integration opportunities with the broader KerasNLP spectrum.
Initial Limitations
With KerasNLP releasing its latest integration, compatibility has extended to around 33 transformer models from a pool of nearly 400 thousand available public models on Hugging Face Hub. Even so, users faced substantial limitations regarding the accessible architecture space in Keras NLP’s framework, necessitating Keras model downloads to experiment with fine-tuned Hugging Face library models on Keras based backends.
<span class="hljs-keyword">from</span> keras_nlp.models <span class="hljs-keyword">element</span>
<span class/span">='height:<*s*data:*key*&: "/>:&&t*' / /';
GemmaCausalLM (from)</*
<a href="use-a-wider-range-of-frameworksi>
Keras models can be smoothly transferred, allowing Keras NLP users to choose a significantly broader model ecosystem without hindrances of platform dependency constraints.One of the earliest KerasNLP-related releases brought a total pool of thirty-three available libraries
Data for Fine-tuning:
models@./upload
gemma2: ~200,~m,
```
Keras-based models will take part and fine-tunable
## What’s New in This Post?
•
KerasNLP - New version of Hugging Face (Keras/transformer)
With 40 models.
New to Keras Models:
KerasLM
KerasTF New
to TensorFlow Keras NLP.