ggerganov/llama.cpp

Pull Requests Commits

convert : update transformers requirements (#16866)

RodriMora committed 5 days ago

Verified ce18efea

server : bump request URI max length to 32768 (#16862)

chansikpark committed 5 days ago

Verified 16724b5b

server : remove n_past (#16818)

ggerganov committed 5 days ago

Verified b52edd25

cpu: introduce chunking for repack matmuls and enable matmul-id chunking on ARM64 (#16833)

max-krasnyansky committed 5 days ago

Verified 517b7170

common: fix typo in cli help text (#16864)

sbera77 committed 5 days ago

Verified 835e918d

model: add support for qwen3vl series (#16780)

JJJYmmm committed 5 days ago

Verified d261223d

cpu: introduce chunking for flash attention (#16829)

max-krasnyansky committed 5 days ago

Verified dcca0d3a

model: Add support for CogVLM model (#15002)

Tianyue-Zhao committed 5 days ago

Verified bacddc04

cuda : fix argsort with 64k+ rows (#16849)

CISC committed 5 days ago

Verified 229bf686

llama : use std::abs instead of abs (#16853)

kaetemi committed 6 days ago

Verified d7395115

vulkan: Handle argsort with a large number of rows (#16851)

jeffbolznv committed 6 days ago

Verified 052df28b

Hide latency of bias and gate-loading (#16847)

ORippler committed 6 days ago

Verified 8b11deea

vulkan: Fuse rope+set_rows (#16769)

jeffbolznv committed 6 days ago

Verified b9ce9401

llama: fix ASAN error with M-RoPE (#16848)

ngxson committed 6 days ago

Verified 3464bdac

llama: store mrope data in KV cell (#16825)

ngxson committed 6 days ago

Verified e3af5563

vulkan: Update topk_moe fusion to handle gpt's late softmax (#16656)

jeffbolznv committed 6 days ago

Verified 10fcc412

Vulkan MMQ Integer Dot Refactor and K-Quant support (#16536)

0cc4m committed 6 days ago

Verified bcf5bda6

Hexagon Op queue & dispatch optimizations (#16820)

max-krasnyansky committed 6 days ago

Verified 3eb2be1c

CUDA: use fastdiv in set-rows (#16834)

am17an committed 6 days ago

Verified e41bcce8

vendor : sync minja (#16500)

CISC committed 6 days ago

Verified 144a4ce8

vulkan: Call ggml_vk_buffer_write_2d from ggml_vk_buffer_copy (#16793)

jeffbolznv committed 6 days ago

Verified f549b000

CUDA: Fix bug in topk-moe for gpt-oss (#16821)

am17an committed 6 days ago

Verified 9a3ea685

sycl: add RMS_NORM_BACK operation support (#16808)

YaelLogic committed 7 days ago

Verified 338074c3

cuda: add SET operation support (#16804)

YaelGitAccount committed 7 days ago

Verified 851553ea

memory : remove KV cache size padding (#16812)

ggerganov committed 7 days ago

Verified 85a7d867

llama-bench : clarify benchmarked parts of the computation (#16823)

ggerganov committed 7 days ago

Verified a8ca18b4

initialise buffer.device in ggml_hexagon_session (#16816)

l3utterfly committed 7 days ago

Verified 8284efc3

embedding: add raw option for --embd-output-format (#16541)

SamMalayek committed 7 days ago

Verified 1c1409e1

llama: consistent ctx <-> buf order for KV cache (#16746)

JohannesGaessler committed 7 days ago

Verified 7a0e900e

grammar : support array references in json schema (#16792)

aldehir committed 7 days ago

Verified 280d97be

Newer Older