 Georgi Gerganov
|
d32e03f449
server : add SWA checkpoints (#15293)
|
há 5 meses atrás |
 Jonathan Graehl
|
5cdb27e091
finetune: SGD optimizer, more CLI args (#13873)
|
há 5 meses atrás |
 compilade
|
ee3a9fcf88
context : fix index overflow on huge outputs (#15080)
|
há 6 meses atrás |
|
Georgi Gerganov
|
a4569c41fd
llama : enable LLAMA_SET_ROWS=1 by default (#14959)
|
há 6 meses atrás |
|
Georgi Gerganov
|
ba42794c9e
graph : fix equal_seq() check (#14986)
|
há 6 meses atrás |
 Daniel Bevenius
|
ca0ef2dddb
llama : clarify comment about pp and tg graphs [no ci] (#14895)
|
há 6 meses atrás |
|
Georgi Gerganov
|
c1dbea752a
context : restore preemptive sched reset when LLAMA_SET_ROWS=0 (#14870)
|
há 6 meses atrás |
|
Georgi Gerganov
|
e4868d16d2
context : perform output reorder lazily upon access after sync (#14853)
|
há 6 meses atrás |
|
Georgi Gerganov
|
d498af3d5a
graph : avoid huge warm-up graphs for MoE models (#14753)
|
há 6 meses atrás |
|
Georgi Gerganov
|
8f974bc1e9
graph : refactor context to not pass gf explicitly (#14629)
|
há 6 meses atrás |
|
Georgi Gerganov
|
01612b7409
llama : reuse compute graphs (#14482)
|
há 6 meses atrás |
|
Georgi Gerganov
|
225e7a1438
llama : add high-throughput mode (#14363)
|
há 6 meses atrás |
 Aman Gupta
|
9c9e4fc635
llama-context: add ability to get logits (#14672)
|
há 6 meses atrás |
|
Georgi Gerganov
|
7b50d589a8
kv-cells : fix tracking of seq_pos (#14339)
|
há 7 meses atrás |
|
Georgi Gerganov
|
692e3cdd0a
memory : rename interface to llama_memory_context_i (#14296)
|
há 7 meses atrás |
|
Georgi Gerganov
|
4c9fdfbe15
ubatch : new splitting logic (#14217)
|
há 7 meses atrás |
|
Georgi Gerganov
|
d3e64b9f49
llama : rework embeddings logic (#14208)
|
há 7 meses atrás |
|
Georgi Gerganov
|
c311ac664d
cparams : rename LLAMA_MAX_PARALLEL_SEQUENCES to LLAMA_MAX_SEQ (#14188)
|
há 7 meses atrás |
|
Georgi Gerganov
|
b9912ac570
batch : auto-gen positions + verify multi-sequence input (#14177)
|
há 7 meses atrás |
|
Georgi Gerganov
|
60c666347b
batch : rework llama_batch_allocr (#14153)
|
há 8 meses atrás |
|
Georgi Gerganov
|
f6e1a7aa87
context : simplify output counting logic during decode (#14142)
|
há 8 meses atrás |
|
Georgi Gerganov
|
c3ee46fab4
batch : remove logits_all flag (#14141)
|
há 8 meses atrás |
|
Georgi Gerganov
|
9596506965
kv-cache : fix split_equal handling in unified implementation (#14130)
|
há 8 meses atrás |
|
compilade
|
a20b2b05bc
context : round n_tokens to next multiple of n_seqs when reserving (#14140)
|
há 8 meses atrás |
|
Georgi Gerganov
|
745aa5319b
llama : deprecate llama_kv_self_ API (#14030)
|
há 8 meses atrás |
|
Georgi Gerganov
|
487a5e0401
context : fix SWA-related warning for multiple sequences (#14045)
|
há 8 meses atrás |
 Sigbjørn Skjæret
|
d17a809ef0
llama : support multiple classifier outputs and labels (#13940)
|
há 8 meses atrás |
|
Georgi Gerganov
|
7f37b6cf1e
memory : migrate from llama_kv_cache to more generic llama_memory (#14006)
|
há 8 meses atrás |
|
Georgi Gerganov
|
9e31bec4fd
context : fix pos_min initialization upon error decode (#14008)
|
há 8 meses atrás |
|
Georgi Gerganov
|
3e63a58ef7
kv-cache : refactor the update/defrag mechanism (#13988)
|
há 8 meses atrás |
