v0.1.2: hotfix

Hotfix for T5 on later versions of transformers.

v0.1.1 - hotfixes

Hotfixes. Closes #57 and #56.

v0.1.0 - Support for T5, local models, integrated gradients, Captum integrations, and beam search decoding

Big update to Ecco! Massive contributions from @JoaoLages and @SSamDav.

• Added support of encoder-decoder models like T5
• Using Captum for feature attribution adds support to these new methods: (IntegratedGradients, Saliency, InputXGradient, DeepLift, DeepLiftShap, GuidedBackprop, GuidedGradCam, Deconvolution, LRP). This replaces the previous implementation within Ecco for Saliency and InputXGradients.
• Added support for Beam Search generation
• Added support for importing local models. Very useful for analyzing finetuned models.
• Added better support for various tokenizers in visualizations. Some more work needed on this front, still.

v0.0.15

• Added support for similarity score of activation matrices through CKA (centered kernel alignment), and Canonical Correlation Analysis methods: CCA, SVCCA, and PWCCA.
• Better loading of configs and model support (thanks @stprior)

v0.0.14

• Adds a documentation portal
• LM now has a call() function so MLMs like BERT can be supported (without requiring text generation). Closes #18
  call() and other functions now all support a batch dimension. The exception is "generate" which works on a single input sequence and not a batch. Closes #19
• Set up ground-work towards #6. BERT is now supported for activation collection and an earlier version of NMF factorization. EccoJS needs to clean up partial token characters like (##). Or, better yet, eccojs should remain dumb and we give it the tokens cleaned up and it only worries about displaying them.
• Part of the groundwork to support additional models is the model-config.yml file which should lay out how to connect ecco.LM with the underlying language model

v0.0.13 Hugging Face Transformers v4 Support

• Added Hugging Face Transformers v4 support. Closing #30.

v0.0.12

• Larger GPT2 models can now work with long sequences in GPU without running out of memory.
• Neuron activations: ability to specify capturing activations from certain layers

Thanks to contributor @nostalgebraist

v0.0.10

Alpha version

v0.0.9-alpha

• Started working on docs
• More tests. Github actions CI/CD

v0.0.8

- release v0.0.8