tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-18 00:27:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

context : move encode/decode to llama-context.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2025-02-12 11:23:38 +02:00
parent 02ef4be975
commit b52b79b048
3 changed files with 48 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
src/llama-context.h
View File

@ -45,7 +45,30 @@ struct llama_context {
virtual ggml_context_ptr init();
// decode a batch of tokens by evaluating the transformer
// in case of unsuccessful decoding (error or warning),
// the kv_cache state will be returned to its original state
// (for non-recurrent models) or cleaned (for recurrent models)
//
// - lctx: llama context
// - inp_batch: batch to evaluate
//
// return 0 on success
// return positive int on warning
// return negative int on error
//
virtual int decode(llama_batch & inp_batch) = 0;
// encode a batch of tokens by evaluating the encoder part of the transformer
//
// - lctx: llama context
// - batch: batch to evaluate
//
// return 0 on success
// return positive int on warning
// return negative int on error
//
virtual int encode(llama_batch & inp_batch) = 0;
// graph build API (generic)