mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-20 01:31:31 +02:00
eagle3 : improve naming
This commit is contained in:
+2
-2
@@ -364,7 +364,7 @@ extern "C" {
|
||||
bool kv_unified; // use a unified buffer across the input sequences when computing the attention
|
||||
// try to disable when n_seq_max > 1 for improved performance when the sequences do not share a large prefix
|
||||
// ref: https://github.com/ggml-org/llama.cpp/pull/14363
|
||||
|
||||
|
||||
// EAGLE3 extraction configuration
|
||||
// When eagle3_model is set, layer extraction is automatically enabled
|
||||
const struct llama_model * eagle3_model; // EAGLE3 model to read extract_layers configuration from
|
||||
@@ -876,7 +876,7 @@ extern "C" {
|
||||
// Returns NULL if no features are available
|
||||
// Format: [3*n_embd, n_tokens] - use model.hparams.n_embd and batch.n_tokens for dimensions
|
||||
LLAMA_API const float * llama_get_eagle3_target_features(struct llama_context * ctx);
|
||||
|
||||
|
||||
// Set g_embeddings from EAGLE3 encoder output for decoder input
|
||||
// g_embd: pointer to encoder output embeddings
|
||||
LLAMA_API void llama_set_eagle3_g_embeddings(
|
||||
|
||||
Reference in New Issue
Block a user