ggml-org/llama.cpp

Pull Requests Commits
completion : fix prompt cache for recurrent models (#19045)
ggerganov committed 7 days ago
Verified 080b1619
readme: update RWKV7 model links (#19061)
MollySophia committed 7 days ago
Verified 1243f93a
llama: fix integer type consistency in split helpers (#18894)
MaheshJakkala committed 7 days ago
Verified 24bc2383
common : use two decimal places for float arg help messages (#19048)
danbev committed 7 days ago
Verified 16639ba2
convert : fix conversion for inheriting models that were bypassing modify_tensors (#19064)
bartowski1182 committed 7 days ago
Verified 9981c301
common: moved self-spec impl to ngram-map
srogmann committed 7 days ago
cb3a4027
llama-fit-params: keep explicit --ctx-size 0 (#19070)
JohannesGaessler committed 7 days ago
Verified e9fd8dca
GGUF: check that tensor size is representable (#19072)
JohannesGaessler committed 7 days ago
Verified 4e5b83b2
chat: fix language input for translategemma (#19052)
ngxson committed 8 days ago
Verified bb02f74c
server: cleanup (remove slot.batch_spec, rename)
srogmann committed 8 days ago
a1584ac8
common: add option --spec-draftless
srogmann committed 8 days ago
1e29af4e
common: add vector of speculative states
srogmann committed 8 days ago
eb43748b
common: add enum common_speculative_type
srogmann committed 8 days ago
b38eb590
common: ngram map, config self-speculative decoding
srogmann committed 8 days ago
456268fa
server: can_speculate() requires a task instance
srogmann committed 8 days ago
907d094f
common: use %zu format specifier for size_t in logging
srogmann committed 8 days ago
f1f6584c
server: replace can_speculate() with slot.can_speculate()
srogmann committed 8 days ago
917f4bb1
server: can_speculate() tests self-spec
srogmann committed 8 days ago
38f7c287
can_speculate() includes self-speculation
srogmann committed 8 days ago
e3e809cc
server: moved self-call into speculative.cpp
srogmann committed 8 days ago
1faeb628
server: introduce self-speculative decoding
srogmann committed 8 days ago
1fb2658b
CUDA: re-use MLA K data for V in MMA FA (#19057)
JohannesGaessler committed 8 days ago
Verified 8f91ca54
ggml-cuda: enable cuda-graphs for `n-cpu-moe` (#18934)
am17an committed 8 days ago
Verified 81ab64f3
ggml-hexagon: flash-attn opt (#19025)
chraac committed 8 days ago
Verified 8af1f5f4
graph : utilize `ggml_build_forward_select()` to avoid reallocations (#18898)
ggerganov committed 9 days ago
Verified 557515be
[SYCL] use malloc to support both iGPU and dGPU in same time (#18992)
arthw committed 9 days ago
Verified cb6caca1
chat : fix translategemma crash on common_chat_format_example (#19019)
ngxson committed 9 days ago
Verified b5b8fa1c
model-conversion : use BUILD_DIR variable in all scripts (#19015)
danbev committed 9 days ago
Verified a14b960b
ggml-cpu: aarm64: q5_K repack gemm and gemv (and generic) implementations (i8mm) (#18860)
Alcpz committed 9 days ago
Verified 091a46cb
cli : load parser definition (#19031)
aldehir committed 9 days ago
Verified a3e81281
Terms Of Use Privacy Policy Refund Policy Impressum