ggml-org/llama.cpp

Pull Requests Commits

server : support preserving reasoning_content in assistant message (#18994)

ngxson committed 10 days ago

Verified 51fa458a

mla : make the V tensor a view of K (#18986)

ggerganov committed 10 days ago

Verified a5eaa1d6

CUDA: fix alignment check for FA (#19023)

JohannesGaessler committed 10 days ago

Verified e2baf021

convert_hf_to_gguf.py: refactor modify_tensors to call super (#18866)

am17an committed 10 days ago

Verified e34d6d03

opencl: enable the general fp mm for non-cont input and as a fallback for specialized kqv kernel for adreno (#18970)

lhez committed 10 days ago

Verified 9c96465f

server: do not log certain endpoints (avoid log spam) (#19028)

ngxson committed 10 days ago

Verified 4e595b25

quant : manual overrides of tensor types take precedence (#18952)

ggerganov committed 11 days ago

Verified 0e4ebeb0

release: update github api (#19022)

taronaeo committed 11 days ago

Verified 8b308407

mtmd : update docs to use llama_model_n_embd_inp (#18999)

ngxson committed 11 days ago

Verified 9eb5bfec

server: Reorder methods in `server-task.cpp` (#19016)

openingnow committed 11 days ago

Verified c6926d1d

CUDA: add gqa_ratio 4 for GLM 4.7 flash (#18953)

am17an committed 11 days ago

Verified b70d2510

opencl: add TRI op support (#18979)

shaofeiqi committed 11 days ago

Verified 5516b9c1

ggml-zdnn : mark zDNN buffers as non-host (#18967)

AlekseiNikiforovIBM committed 11 days ago

Verified 94242a62

ci : update GitHub Actions versions [no ci] (#18935)

pgoslatara committed 11 days ago

Verified 6b99a223

convert : add Devstral-2 (Ministral3ForCausalLM) arch (#18972)

emsi committed 11 days ago

Verified 77078e80

jinja: support none|string (#18995)

pwilkin committed 11 days ago

Verified c301172f

fix: Use `tabular-nums` for chat message statistics (#18915)

nathanlesage committed 12 days ago

Verified 3802d3c7

llama : clarify nemotron-h.cpp comment about RoPE [no ci] (#18997)

danbev committed 12 days ago

Verified 9da3dcd7

vulkan: Remove transfer_ctx, do everything in compute_ctx. (#18945)

jeffbolznv committed 12 days ago

Verified bd544c94

common : improve error message when HTTPS is missing but required (#18987)

angt committed 12 days ago

Verified 14be5a39

server: /v1/responses (partial) (#18486)

openingnow committed 12 days ago

Verified fbbf3ad1

vulkan: support flash attention GQA/split_k with small batches (#18938)

jeffbolznv committed 12 days ago

Verified 33f890e5

Revert "vulkan: force full subgroups for flash attention to fix intel subgroup crash (#17356)" (#18831)

rillomas committed 12 days ago

Verified 067b8d7a

vulkan: Use mul_mat_vec_id for small values of n (#18918)

jeffbolznv committed 12 days ago

Verified 50b7f076

memory : add llama_memory_hybrid_iswa (#18601)

tdakhran committed 12 days ago

Verified ad8d85bd

Fix GLM 4.7 Lite MoE gating func (#18980)

pwilkin committed 12 days ago

Verified 12a4a47e

gguf: display strerrno when cant load a model (#18884)

teto committed 12 days ago

Verified 37c35f0e

CUDA: Fix builds for older CCCL versions by ifdefing strided_iterator (#18964)

ORippler committed 12 days ago

Verified 5bd341c9

common, server : use the same User-Agent by default (#18957)

angt committed 13 days ago

Verified 1c7cf94b

cli : fix reasoning responses in CLI (#18961)

ngxson committed 13 days ago

Verified 2c1f1996

Newer Older