ggerganov/llama.cpp

Pull Requests Commits

metal : pad mm results

ggerganov committed 158 days ago

Verified 16843dba

imatrix: fix oob writes if src1 is not contiguous (#13286)

JohannesGaessler committed 158 days ago

Verified 3e959f09

clip : revert the change of BOI/EOI token for GLM-edge (⚠️ breaking change) (#13259)

ngxson committed 159 days ago

Verified 36667c8e

llama : Llama-3_1-Nemotron-Ultra-253B-v1 support (#12843)

ymcki committed 159 days ago

Verified 3bf785f3

llama : move end-user examples to tools directory (#13249)

slaren committed 160 days ago

Verified 1d36b367

sync : ggml (#13268)

ggerganov committed 160 days ago

Verified b3444392

context : fix reorder logic (#13267)

ggerganov committed 160 days ago

Verified a75cb30d

ggml : Enable MMA for BF16 in llamafile_sgemm (#13148)

shalinib-ibm committed 160 days ago

Verified 3f3769ba

llama-model : support Qwen2 embedding models and pooling_mode_lasttoken (#13245)

cebtenzzre committed 160 days ago

Verified 2f567611

convert : use correct context length for nomic-embed-text-v2 (#13216)

cebtenzzre committed 160 days ago

Verified 7d212348

convert : converting mmproj for Qwen2/2.5VL from convert_hf_to_gguf (#13209)

ngxson committed 160 days ago

Verified 074e42ab

kv-cache : separate recurrent vs non-recurrent impl (#12799)

ggerganov committed 160 days ago

Verified c642bc01

llama : orion rope type is neox (#13261)

CISC committed 160 days ago

Verified cb06a3c3

llama : plamo rope type is neox (#13260)

CISC committed 160 days ago

Verified 626083fa

llama-chat : reset glmedge chat template (#13253)

piDack committed 160 days ago

Verified 2af68801

mtmd-cli : fix out_of_range when input image path is empty (#13244)

ahmedshakill committed 160 days ago

Verified e84773ab

server : add cache reuse card link to help (#13230)

ggerganov committed 160 days ago

Verified fab647e8

convert : explicitly disable trust_remote_code for AutoConfig (#13246)

ngxson committed 160 days ago

Verified dcf88600

ci: fix cross-compile sync issues (#12804)

bandoti committed 160 days ago

Verified d24d5928

rpc : avoid uninitialized memory in serialize_tensor (#13210)

justinsb committed 160 days ago

Verified 8efbdadc

ggml: Don't assert fail when tensor data changes (#13222)

jessegross committed 160 days ago

Verified f057808f

build : fix build info on windows (#13239)

slaren committed 160 days ago

Verified d7a14c42

clip : (minicpmv) Re-enable upscaling of images smaller than the CLIP image size (#13237)

lcarrere committed 161 days ago

Verified b6e4ff69

llama-chat : update GLM4 chat template (#13238)

matteoserva committed 161 days ago

Verified e0f572c8

vulkan: Add bfloat16 support (#12554)

jeffbolznv committed 161 days ago

Verified 79f26e9e

vulkan: Handle src1 batch dimension in non-contiguous mat-vec-mul shader (#13191)

jeffbolznv committed 161 days ago

Verified fc727bcd

test: non-cont. b in test-backend-ops -o MUL_MAT (#13187)

JohannesGaessler committed 161 days ago

Verified b0ecbd43

ggerganov committed 161 days ago

b1dd4d08

whisper : add check that target name exists (whisper/3103)

danbev committed 161 days ago

99881f77

ggml : suppress Windows compiler warnings (whisper/3075)

danbev committed 161 days ago

b5769d92

Older