ngxson/llama.cpp

Pull Requests Commits

ngxson committed 207 days ago

cbb6f20a

ngxson committed 207 days ago

c2f4dc78

add rerank prompt

ngxson committed 207 days ago

c02f53d9

add embedding model tokenizer chkhsh

ngxson committed 207 days ago

8edd2cf3

rm redundant settings

ngxson committed 207 days ago

0777cd35

Merge branch 'master' into xsn/qwen3_embd_rerank

ngxson committed 207 days ago

f8facb30

correct output token position

ngxson committed 207 days ago

030dc3b0

llama : support multiple classifier outputs and labels (#13940)

CISC committed 207 days ago

Verified d17a809e

ngxson committed 208 days ago

e0eb4b8b

Merge branch 'master' into xsn/qwen3_embd_rerank

ngxson committed 208 days ago

f8fd4409

gguf-py : add add_classifier_output_labels method to writer (#14031)

CISC committed 208 days ago

Verified 1caae7fc

vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (#14001)

rillomas committed 208 days ago

Verified 669c13e0

ci: fix CUDA build failure on autodl cloud machines (#14005)

pockers21 committed 208 days ago

Verified 146b88e8

model : support qwen3 rerank and embeddings

ngxson committed 208 days ago

3f3b9a22

memory : migrate from llama_kv_cache to more generic llama_memory (#14006)

ggerganov committed 208 days ago

Verified 7f37b6cf

llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (#14013)

slaren committed 208 days ago

Verified 3a077146

readme : add badge (#13938)

Olexandr88 committed 208 days ago

Verified d01d112a

vocab : warn about missing mask token (#14022)

CISC committed 208 days ago

Verified 9f47fa57

context : fix pos_min initialization upon error decode (#14008)

ggerganov committed 208 days ago

Verified 9e31bec4

vulkan: automatically deduce size of push constants (#13936)

jeffbolznv committed 208 days ago

Verified 5a8ae305

ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (#13813)

etasnadi committed 209 days ago

Verified 0d398442

kv-cache : refactor the update/defrag mechanism (#13988)

ggerganov committed 209 days ago

Verified 3e63a58e

ci : remove cuda 11.7 releases, switch runner to windows 2022 (#13997)

slaren committed 209 days ago

Verified 2589ad37

releases : use dl backend for linux release, remove arm64 linux release (#13996)

slaren committed 209 days ago

Verified 48254871

llama-graph : use ggml_repeat_4d (#13998)

ngxson committed 209 days ago

Verified 3ac67535

CUDA: fix FTZ in FA for Gemma 3 (#13991)

JohannesGaessler committed 209 days ago

Verified 0b4be4c4

kv-cache : fix unified::seq_rm to work with seq_id < 0 (#13985)

ggerganov committed 209 days ago

Verified e0e806f5

vulkan: fix warnings in perf logger querypool code (#13937)

jeffbolznv committed 210 days ago

Verified 7e00e60e

docs : add "Quick start" section for new users (#13862)

ngxson committed 210 days ago

Verified ea1431b0

opencl: add `backend_synchronize` (#13939)

lhez committed 210 days ago

Verified 71e74a3a

Older