Hugging Face's logo

google
/
gemma-4-E2B-it

Any-to-Any
Transformers
Safetensors
gemma4
image-text-to-text
Eval Results
Model card Files
xet
 Community
31

Litert version after fine-tuning

#17
by Abhijitmore - opened
Discussion
Abhijitmore

Do we have any way to convert finetuned model to litert, there is preconverted litert model available. but I am not able to convert once I fine-tune.
pannaga10
 Google org

Hi @Abhijitmore Apologies for late response .
Can you please share a bit more context?

  1. What framework is being used (e.g., PyTorch, TensorFlow, Keras, JAX)?
  2. How did you finetune? full finetune, LoRA/QLoRA, or something else? And did you merge the adapter weights back into the base model before attempting conversion?
  3. What's going wrong exactly? does the conversion fail with an error, or does it complete but the model behaves incorrectly on-device?
  4. If possible can you please share the script the code you are running to attempt the conversion

Thanks

Sign up or log in to comment