CODEWITHSUNDEEP
pythonkagglegemma
Marecky
Marecky

Reputation: 2106

Trying to run Gemma on Kaggle and got issue 'keras_nlp.models' has no attribute 'GemmaCausalLM'

I'm trying to run Gemma on Keras with this model https://www.kaggle.com/models/keras/gemma/frameworks/keras/variations/gemma_instruct_7b_en And I'm reproducing the Example available on "Model Card" on above page. When I run this code:

gemma_lm = keras_nlp.models.GemmaCausalLM.from_preset("gemma_instruct_7b_en")
gemma_lm.generate("Keras is a", max_length=30)

# Generate with batched prompts.
gemma_lm.generate(["Keras is a", "I want to say"], max_length=30)

I get this error:

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[2], line 1
----> 1 gemma_lm = keras_nlp.models.GemmaCausalLM.from_preset("gemma_instruct_7b_en")
      2 gemma_lm.generate("Keras is a", max_length=30)
      4 # Generate with batched prompts.
AttributeError: module 'keras_nlp.models' has no attribute 'GemmaCausalLM'

How can I fix this?

Upvotes: 3

Views: 1680

Answers (2)

JOSMAR BARBOSA - M4NOV3Y
JOSMAR BARBOSA - M4NOV3Y

Reputation: 81

For me did not resolve restart the Kernel and I think that's related to keras_nlp and Gemma (maybe some versioning issue) I needed to update my keras_nlp to 0.8.1

pip install -U keras-nlp

Upvotes: 0

gsamaras
gsamaras

Reputation: 73414

Try restarting the kernel.

Note that the same behavior occurs with the 2B Gemma model, i.e. gemma_lm = keras_nlp.models.GemmaCausalLM.from_preset("gemma_2b_en").

I am not sure why restarting the kernel works, but it's worth noting that this doesn't happen in Google Colab.

Upvotes: 3

Related Questions