 Georgi Gerganov
|
cd5e3b5754
server : support unified cache across slots (#16736)
|
vor 2 Monaten |
 Johannes Gäßler
|
e789095502
llama: print memory breakdown on exit (#15860)
|
vor 4 Monaten |
 Diego Devesa
|
9777032dcc
llama : separate compute buffer reserve from fattn check (#15696)
|
vor 4 Monaten |
|
Georgi Gerganov
|
8a4280ce43
kv-cache : remove LLAMA_SET_ROWS checks (#15505)
|
vor 4 Monaten |
|
Georgi Gerganov
|
cd36b5e5c7
llama : remove deprecated llama_kv_self API (#15472)
|
vor 5 Monaten |
|
Georgi Gerganov
|
d32e03f449
server : add SWA checkpoints (#15293)
|
vor 5 Monaten |
 Jonathan Graehl
|
5cdb27e091
finetune: SGD optimizer, more CLI args (#13873)
|
vor 5 Monaten |
|
Georgi Gerganov
|
a4569c41fd
llama : enable LLAMA_SET_ROWS=1 by default (#14959)
|
vor 5 Monaten |
|
Georgi Gerganov
|
ba42794c9e
graph : fix equal_seq() check (#14986)
|
vor 5 Monaten |
|
Georgi Gerganov
|
c1dbea752a
context : restore preemptive sched reset when LLAMA_SET_ROWS=0 (#14870)
|
vor 6 Monaten |
|
Georgi Gerganov
|
e4868d16d2
context : perform output reorder lazily upon access after sync (#14853)
|
vor 6 Monaten |
|
Georgi Gerganov
|
8f974bc1e9
graph : refactor context to not pass gf explicitly (#14629)
|
vor 6 Monaten |
|
Georgi Gerganov
|
01612b7409
llama : reuse compute graphs (#14482)
|
vor 6 Monaten |
|
Georgi Gerganov
|
692e3cdd0a
memory : rename interface to llama_memory_context_i (#14296)
|
vor 7 Monaten |
|
Georgi Gerganov
|
4c9fdfbe15
ubatch : new splitting logic (#14217)
|
vor 7 Monaten |
|
Georgi Gerganov
|
60c666347b
batch : rework llama_batch_allocr (#14153)
|
vor 7 Monaten |
|
Georgi Gerganov
|
7f37b6cf1e
memory : migrate from llama_kv_cache to more generic llama_memory (#14006)
|
vor 7 Monaten |
|
Georgi Gerganov
|
3e63a58ef7
kv-cache : refactor the update/defrag mechanism (#13988)
|
vor 7 Monaten |
|
Georgi Gerganov
|
3f55f781f1
llama : auto-batch preparation (#13845)
|
vor 7 Monaten |
|
Georgi Gerganov
|
12d0188c0d
kv-cache : refactor + add llama_memory_state_i (#13746)
|
vor 7 Monaten |
|
Johannes Gäßler
|
10d2af0eaa
llama/ggml: add LLM training support (#10544)
|
vor 8 Monaten |
|
Georgi Gerganov
|
51fb96b1ff
context : remove logits_all flag (#13284)
|
vor 8 Monaten |
|
Georgi Gerganov
|
c642bc014c
kv-cache : separate recurrent vs non-recurrent impl (#12799)
|
vor 8 Monaten |
|
Diego Devesa
|
295354ea68
llama : fix K-shift with quantized K and BLAS backend (#13113)
|
vor 9 Monaten |
 fairydreaming
|
8fcb563613
Load all MoE experts during warmup (#11571)
|
vor 10 Monaten |
|
Georgi Gerganov
|
84d5475541
llama : fix Gemma3 SWA KV cache shift (#12373)
|
vor 10 Monaten |
|
Georgi Gerganov
|
e0dbec0bc6
llama : refactor llama_context, llama_kv_cache, llm_build_context (#12181)
|
vor 10 Monaten |
|
Georgi Gerganov
|
afa8a9ec9b
llama : add `llama_vocab`, functions -> methods, naming (#11110)
|
vor 1 Jahr |
|
Georgi Gerganov
|
f66f582927
llama : refactor `src/llama.cpp` (#10902)
|
vor 1 Jahr |
