ExLLama.EmbeddingOptions (LLama CPP Nif Wrapper v0.4.1)

Copy Markdown

Options for ExLLama.Model.embeddings/3 embedding passes.

  • n_threads — threads for the embedding pass (default 1).
  • n_threads_batch — threads for the prompt/batch pass (default 1).

Note: the current embeddings NIF bridge creates its embeddings-enabled context internally and does not yet forward these thread options — they are reserved API surface. Embedding behavior (pooling, causal masking) is tuned via the 0.4.0 ExLLama.SessionOptions / ExLLama.ContextParams fields pooling_type and attention_type.