minllama-assignment
minllama-assignment copied to clipboard
Results
2
minllama-assignment issues
Sort by
recently updated
recently updated
newest added
In `structure.md` llama.Attention.forward -> llama.Attention.compute_query_key_value_scores llama.Llama.forward -> llama.LlamaLayer.forward Rectified some functions in `To be implemented` section In `rope_test.py` Use Tuple from typing instead of tuple as it is not subscriptable...
there is no causal mask in the attention layer. Is it because the model is designed for classification rather than generation?