The beam search and llama_token_get_type related functions have

[AdolfVonKleist]

recently been removed from llama.cpp and are no longer present in the latest commit (f83351f9a62a6262f1fc3d08f320033089cddfb5) This causes llama-cpp-python import to fail due to missing symbols. This commit removes the stale references from llama-cpp-python.

without these changes, llama_cpp generates the following symbol lookup errors on import:

libllama.so: undefined symbol: llama_beam_search
libllama.so: undefined symbol: llama_token_get_type. Did you mean: 'llama_token_get_text'?

these deletions eliminate the symbol lookup errors.

[AdolfVonKleist]

The specific pull request where llama_beam_search was removed from llama.cpp, for reference:

• llama : remove beam search ggml-org/llama.cpp#7736

[abetlen]

Thank you for the PR, the functions have been removed on main.