yuhuili

Hello, I am Yuhui Li, the author of the EAGLE paper, and I am here to answer your question. > Does the high accept rate bring the promising speedup? Based...

It seems that the name of the embedding in your model is not 'embed_tokens'. You can modify it to the name of the embedding layer in your model.

This is not necessary; EAGLE's structure is independent of the target model. You can use the same cnet.py, or you can try other structures as well.

I noticed that your "n_layers" is set to 38, which makes your draft model very large. In EAGLE, the draft model consists of only one layer.

Are you running the training script we provided?

You need to modify the instruction templates (such as those in eagle/ge_data/ge_data_all_vicuna.py). Training the draft model for LLaMA 3 is our next step.

Hey @kalradivyanshu, support for LLaMA3 has now been updated.

After obtaining the result file, you can run the *[eagle/evaluation/alpha.py](https://github.com/SafeAILab/EAGLE/blob/main/eagle/evaluation/alpha.py)* file to get the acceptance rate.

You can check #5.

@shanpoyang654 The previous code had issues when using the Vicuna template. This problem has now been resolved, and you can use the latest code.