ngxson/llama.cpp

Pull Requests Commits

llama : remove check flash_attn with lora

ngxson committed 344 days ago

8dbd0880

llama : rename missed batch params/vars to ubatch (#10059)

danbev committed 344 days ago

Verified 6369f867

llama : update llama_model API names (#11063)

ggerganov committed 344 days ago

Verified 47182dd0

tokenize : escape the prompt (#11058)

ggerganov committed 344 days ago

Verified 3e6e7a6b

mmap : fix fileno macro clash (#11076)

ggerganov committed 344 days ago

Verified ae2f606b

llama : use LLAMA_TOKEN_NULL (#11062)

ggerganov committed 344 days ago

Verified 727368c6

llama : use _impl suffix instead of _internal (#11060)

ggerganov committed 344 days ago

Verified 5047dd35

CUDA: add BF16 support (#11093)

JohannesGaessler committed 344 days ago

Verified 46e3556e

Vulkan: Add device-specific blacklist for coopmat for the AMD proprietary driver (#11074)

0cc4m committed 345 days ago

Verified b56f079e

llama : Add support for DeepSeek V3 (#11049)

fairydreaming committed 345 days ago

Verified 9394bbd4

[GGML][RPC] Support for models with non-512-aligned tensors over RPC. (#11047)

matt23654 committed 345 days ago

Verified f922a9c5

llama : add support for the cohere2 model architecture (#10900)

dranger003 committed 345 days ago

Verified 46be9422

ggerganov committed 345 days ago

78c67851

ggml : do not install metal source when embed library (ggml/1054)

ggerganov committed 345 days ago

5e3b08d6

ggml : improve inputs log sched_print_assignments (ggml/1053)

danbev committed 345 days ago

db68c93b

fix: Vulkan shader gen binary path (#11037)

giladgd committed 346 days ago

Verified c31fc8b9

common : disable KV cache shifting automatically for unsupported models (#11053)

MollySophia committed 347 days ago

Verified 4b0c638b

metal : avoid uint (#11019)

ggerganov committed 347 days ago

Verified e7da954e

llama : refactor `src/llama.cpp` (#10902)

ggerganov committed 347 days ago

Verified f66f5829

server: bench: minor fixes (#10765)

phymbert committed 347 days ago

Verified 2f0ee84b

server : allow using LoRA adapters per-request (#10994)

ngxson committed 348 days ago

Verified 0da5d860

readme : add llama-swap to infrastructure section (#11032)

mostlygeek committed 348 days ago

Verified a45433ba

ggml : fixes for AVXVNNI instruction set with MSVC and Clang (#11027)

Srihari-mcw committed 349 days ago

Verified 0827b2c1

server : clean up built-in template detection (#11026)

ngxson committed 349 days ago

Verified 45095a61

server : add OAI compat for /v1/completions (#10974)

ngxson committed 350 days ago

Verified 5896c652

convert : fix Llama-3_1-Nemotron-51B rope settings (#11008)

ymcki committed 350 days ago

Verified bc7b1f86

common, examples, ggml : fix MSYS2 GCC compiler errors and warnings when building with LLAMA_CURL=ON and GGML_OPENCL=ON (#11013)

peter277 committed 350 days ago

Verified 6e1531ac

vulkan: optimize mul_mat for small values of N (#10991)

jeffbolznv committed 350 days ago

Verified 716bd6de

android : fix llama_batch free (#11014)

ag2s20150909 committed 351 days ago

Verified c250ecb3

vulkan: im2col and matmul optimizations for stable diffusion (#10942)

jeffbolznv committed 352 days ago

Verified a813badb

Older