ggml-org/llama.cpp

Pull Requests Commits

sampling : allow null samplers

ggerganov committed 1 hour ago

Verified c3c1300f

common : restart grammar-based rejection sampling

ggerganov committed 2 hours ago

Verified 746c47e3

ggml-cpu: ARM64: repack version of q8_0 (dotprod and i8mm) (#18096)

Alcpz committed 2 hours ago

Verified 669696e0

model: fix LFM2_MOE missing tensors (#18132)

tdakhran committed 3 hours ago

Verified 982060fa

ci : clean up webui jobs (#18116)

CISC committed 4 hours ago

Verified 6853bee6

common: fix --override-kv to support comma-separated values (#18056)

ServeurpersoCom committed 4 hours ago

Verified 487674fb

HIP: Refactor mma for RDNA and CDNA (#17990)

zhang-hui-yulo committed 5 hours ago

Verified acec774e

llama.android : Rewrite Android binding (w/o cpu_features dep) (#17413)

naco-siren committed 6 hours ago

Verified 5c0d1888

arg: allow -kvu flag for llama-perplexity (#18117)

TrevorS committed 7 hours ago

Verified 4b2a4778

ggml : use WARP_SIZE/2 for argmax reduction offset (#18092)

Aadeshveer committed 10 hours ago

Verified 58062860

gguf-py : allow converting multi-tensor models from read-only locations (#18100)

ykhrustalev committed 12 hours ago

Verified 2973a65e

llama-fit-params: force disable mlock (#18103)

JohannesGaessler committed 14 hours ago

Verified d0794e89

llama-fit-params: lower ctx size for multi GPU (#18101)

JohannesGaessler committed 14 hours ago

Verified 9dcac6cf

llama-fit-params: fix underflow for dense models (#18095)

JohannesGaessler committed 14 hours ago

Verified 0e49a7b8

llama-fit-params: QoL impr. for prints/errors (#18089)

JohannesGaessler committed 15 hours ago

Verified 4164596c

model: fix LFM2 missing tensors (#18105)

ngxson committed 20 hours ago

Verified ef83fb86

llama: fix early stop in params_fit if ctx is set (#18070)

JohannesGaessler committed 1 day ago

Verified ec98e200

server: fix crash when batch > ubatch with embeddings (#17912)

yifant-code committed 1 day ago

Verified 59977eba

model-conversion : remove -fa option in model card template [no ci] (#18088)

danbev committed 1 day ago

Verified 79dbae03

arch: refactor LLM_TENSOR_NAMES (#18051)

ngxson committed 1 day ago

Verified 7f2b2f3c

arg: clarify auto kvu/np being set on server (#17997)

ngxson committed 1 day ago

Verified 7b1db3d3

Optimization: Qwen3 next autoregressive pass (#17996)

pwilkin committed 1 day ago

Verified a5251ca1

CLI: fixed adding cli and completion into docker containers, improved docs (#18003)

andrew-aladev committed 1 day ago

Verified fb644247

server: Update README.md incorrect argument (#18073)

2114L3 committed 1 day ago

Verified 5f5f9b46

model: support GLM4V vision encoder (#18042)

ngxson committed 1 day ago

Verified 3d86c6c2

model-conversion : add note about verifying previous models (#18082)

danbev committed 1 day ago

Verified 9963b81f

model-conversion : use CONVERTED_EMBEDDING_MODEL for embedding_verify_logits (#18079)

danbev committed 1 day ago

Verified db81d5ec

common : add nemotron 3 parsing (#18077)

aldehir committed 1 day ago

Verified c05aa69f

added note for old Intel hardware pre sycl (#18017)

alosslessdev committed 1 day ago

Verified 279cef27

security : add collaborator guidance (#18081)

ggerganov committed 1 day ago

Verified 5ba95754

Older