ggml-org/llama.cpp

llama : refactor the llm.build_xxx functions

Chiusa

#5239 aperta il 31 gen 2024

 (3 commenti) (3 reazioni) (0 assegnatari)C++ (21.764 fork)batch import
good first issuerefactoringroadmap

Metriche repository

Star
 (124.134 stelle)
Metriche merge PR
 (Merge medio 6g 8h) (389 PR mergiate in 30 g)

Descrizione

Now that we support a large amount of architectures, we can clearly see the patterns when constructing the compute graphs - i.e. optional biases, different norm types, QKV vs Q+K+V, etc.

We should deduplicate the copy-paste portions in functions such as llm.build_llama(), llm.build_falcon(), etc.

The advantage of the current code is that it is easy to look into the graph of a specific architecture. When we refactor this, we will lose this convenience to some extend. So we should think about making this refactoring in such a way that we don't completely obscure which parts of the graph belong to which architectures

Open for ideas and suggestions how to do this best

Guida contributor