ggml-org/llama.cpp

Pull Requests Commits

CUDA: mul_mat_v support for batch sizes > 1 (#14262)

JohannesGaessler committed 89 days ago

Verified defe2158

kv-cells : fix tracking of seq_pos (#14339)

ggerganov committed 89 days ago

Verified 7b50d589

vulkan: update windows SDK in CI (#14334)

jeffbolznv committed 89 days ago

Verified 3a9457df

quantize : handle user-defined pruning of whole layers (blocks) (#13037)

EAddario committed 89 days ago

Verified fa4a9f2a

gguf-py : fix SpecialVocab parsing when post_processor is null (#14330)

CISC committed 90 days ago

Verified 238005c2

run : avoid double tokenization (#14327)

retr0reg committed 90 days ago

Verified 66aba7ac

examples : fix is_first logic for tokenization (#14329)

ggerganov committed 90 days ago

Verified f1f5e82d

HIP: enable vec fattn on RDNA4 (#14323)

IMbackK committed 90 days ago

Verified af3373f1

mtmd : fix Pixtral OOM with large images by capping image_size to 1024 (#14326)

yuiseki committed 90 days ago

Verified 5d5c066d

common : use std::string_view now that we target c++17 (#14319)

CISC committed 90 days ago

Verified 40bfa04c

CUDA: add mean operation (#14313)

am17an committed 90 days ago

Verified aa064b2e

gguf-py : fix Qwen3-Embedding eos token (#14314)

CISC committed 91 days ago

Verified aa0ef5c5

Add support for VK_EXT_debug_utils to add labels to Vulkan objects. (#13792)

mtavenrath committed 91 days ago

Verified bb16041c

gguf-py : fix TemplateProcessing pair when bos/eos is missing (#14312)

CISC committed 91 days ago

Verified 58cba76a

metal : fix thread-safety (#14300)

ggerganov committed 91 days ago

Verified 67ae5312

memory : rename interface to llama_memory_context_i (#14296)

ggerganov committed 91 days ago

Verified 692e3cdd

convert : fix Llama 4 conversion (#14311)

danielhanchen committed 91 days ago

Verified b23fa0b3

ggerganov committed 92 days ago

06cbedfc

Add `ggml_roll` (ggml/1274)

Acly committed 92 days ago

b7147673

docs : fix the link to llama.h (#14293)

david20571015 committed 92 days ago

Verified d860dd99

CUDA: add conv_2d_transpose (#14287)

am17an committed 92 days ago

Verified c959f462

lint : remove trailing whitepace (#14304)

CISC committed 92 days ago

Verified 22015b20

vocab : prevent tokenizer overflow (#14301)

retr0reg committed 92 days ago

Verified dd6e6d0b

sycl: add usage of enqueue_functions extension (#14244)

s-Nick committed 92 days ago

Verified 8308f98c

Implement GGML_CPU_ALL_VARIANTS for PowerPC (#14286)

ckastner committed 92 days ago

Verified 6369be07

llama : improve sep token handling (#14272)

CISC committed 92 days ago

Verified 88fc854b

cuda : synchronize graph capture and cublas handle destruction (#14288)

slaren committed 92 days ago

Verified e28c1b93

ggml : fix repack work size for mul_mat_id (#14292)

ggerganov committed 92 days ago

Verified d27b3ca1

ggml: Update KleidiAI to v1.9.0 (#14277)

chaxu01 committed 92 days ago

Verified 9230dbe2

model : more uniform output id handling (#14275)

ggerganov committed 92 days ago

Verified 812939a9

Newer Older