tianyu-l

@rlrs It seems HF's llama implementation is different from the official llama's. We'll need to understand why that's the case. HF: https://github.com/huggingface/transformers/blob/main/src/transformers/models/llama/modeling_llama.py#L184 meta-llama: https://github.com/meta-llama/llama3/blob/main/llama/model.py#L65 asked here: https://github.com/huggingface/transformers/issues/30872

@rlrs > This also seems to mean that loading weights from https://huggingface.co/meta-llama/Meta-Llama-3-8B does not work. If I understand correctly, there are two ways you can download weights from HF. The...

not sure why the 1D compile test is failing...