@@ -4981,63 +4981,46 @@ void llama_context_enc_dec::perf_reset() {
49814981
49824982//
49834983// state save/load
4984+ // TODO: for now dump just the decoder state, in the future dump both
49844985//
49854986
49864987size_t llama_context_enc_dec::state_get_size () {
4987- GGML_ABORT ( " TODO: implement " );
4988+ return ctx_dec-> state_get_size ( );
49884989}
49894990
49904991size_t llama_context_enc_dec::state_get_data ( uint8_t * dst, size_t size) {
4991- GGML_UNUSED (dst);
4992- GGML_UNUSED (size);
4993- GGML_ABORT (" TODO: implement" );
4992+ return ctx_dec->state_get_data (dst, size);
49944993}
49954994
49964995size_t llama_context_enc_dec::state_set_data (const uint8_t * src, size_t size) {
4997- GGML_UNUSED (src);
4998- GGML_UNUSED (size);
4999- GGML_ABORT (" TODO: implement" );
4996+ return ctx_dec->state_set_data (src, size);
50004997}
50014998
50024999size_t llama_context_enc_dec::state_seq_get_size (llama_seq_id seq_id) {
5003- GGML_UNUSED (seq_id);
5004- GGML_ABORT (" TODO: implement" );
5000+ return ctx_dec->state_seq_get_size (seq_id);
50055001}
50065002
50075003size_t llama_context_enc_dec::state_seq_get_data (llama_seq_id seq_id, uint8_t * dst, size_t size) {
5008- GGML_UNUSED (seq_id);
5009- GGML_UNUSED (dst);
5010- GGML_UNUSED (size);
5011- GGML_ABORT (" TODO: implement" );
5004+ return ctx_dec->state_seq_get_data (seq_id, dst, size);
50125005}
50135006
50145007size_t llama_context_enc_dec::state_seq_set_data (llama_seq_id seq_id, const uint8_t * src, size_t size) {
5015- GGML_UNUSED (seq_id);
5016- GGML_UNUSED (src);
5017- GGML_UNUSED (size);
5018- GGML_ABORT (" TODO: implement" );
5008+ return ctx_dec->state_seq_set_data (seq_id, src, size);
50195009}
50205010
50215011bool llama_context_enc_dec::state_load_file (
50225012 const char * filepath,
50235013 llama_token * tokens_out,
50245014 size_t n_token_capacity,
50255015 size_t * n_token_count_out) {
5026- GGML_UNUSED (filepath);
5027- GGML_UNUSED (tokens_out);
5028- GGML_UNUSED (n_token_capacity);
5029- GGML_UNUSED (n_token_count_out);
5030- GGML_ABORT (" TODO: implement" );
5016+ return ctx_dec->state_load_file (filepath, tokens_out, n_token_capacity, n_token_count_out);
50315017}
50325018
50335019bool llama_context_enc_dec::state_save_file (
50345020 const char * filepath,
50355021 const llama_token * tokens,
50365022 size_t n_token_count) {
5037- GGML_UNUSED (filepath);
5038- GGML_UNUSED (tokens);
5039- GGML_UNUSED (n_token_count);
5040- GGML_ABORT (" TODO: implement" );
5023+ return ctx_dec->state_save_file (filepath, tokens, n_token_count);
50415024}
50425025
50435026size_t llama_context_enc_dec::state_seq_load_file (
@@ -5046,24 +5029,15 @@ size_t llama_context_enc_dec::state_seq_load_file(
50465029 llama_token * tokens_out,
50475030 size_t n_token_capacity,
50485031 size_t * n_token_count_out) {
5049- GGML_UNUSED (seq_id);
5050- GGML_UNUSED (filepath);
5051- GGML_UNUSED (tokens_out);
5052- GGML_UNUSED (n_token_capacity);
5053- GGML_UNUSED (n_token_count_out);
5054- GGML_ABORT (" TODO: implement" );
5032+ return ctx_dec->state_seq_load_file (seq_id, filepath, tokens_out, n_token_capacity, n_token_count_out);
50555033}
50565034
50575035size_t llama_context_enc_dec::state_seq_save_file (
50585036 llama_seq_id seq_id,
50595037 const char * filepath,
50605038 const llama_token * tokens,
50615039 size_t n_token_count) {
5062- GGML_UNUSED (seq_id);
5063- GGML_UNUSED (filepath);
5064- GGML_UNUSED (tokens);
5065- GGML_UNUSED (n_token_count);
5066- GGML_ABORT (" TODO: implement" );
5040+ return ctx_dec->state_seq_save_file (seq_id, filepath, tokens, n_token_count);
50675041}
50685042
50695043//
0 commit comments