Add gemma lightning example for single L40 GPU

[qingquansong]

Summary

Add gemma lightning example for single L40 GPU

Testing Done

• Hardware Type:
• run make test to ensure correctness
• run make checkstyle to ensure code style
• run make test-convergence to ensure convergence

[qingquansong]

@yundai424 @ByronHsu Waiting for #111 to check in and then can patch fused CE to test memory

[yundai424]

@qingquansong do you think if we should just replace llama with gemma2 in this lightning example, and fix the config so that it works with L40 GPU, just to minimize the code 🤔 -- and the example can be used out of the box as lightning studio example

[qingquansong]

@qingquansong do you think if we should just replace llama with gemma2 in this lightning example, and fix the config so that it works with L40 GPU, just to minimize the code 🤔 -- and the example can be used out of the box as lightning studio example

SG! Or maybe set the default one to be Gemma and provide a llama example in readme with config changes? (basically wanna keep the small changes (tokenizer and fsdp) to support llama in the pro lightning testing