ggerganov/llama.cpp

Pull Requests Commits

ggml: refactor cross entropy loss CPU impl. (ggml/976)

JohannesGaessler committed 1 year ago

Verified eee39bdc

metal : fix compute pass descriptor autorelease crash (#9718)

jmousseau committed 1 year ago

Verified 5d5ab1e5

ggml-backend : add device description to CPU backend (#9720)

slaren committed 1 year ago

Verified a7ad5535

ggml: unify backend logging mechanism (#9709)

bandoti committed 1 year ago

Verified d6fe7abf

convert : handle tokenizer merges format from transformers 4.45 (#9696)

compilade committed 1 year ago

Verified e3c355ba

rpc : enable vulkan (#9714)

rgerganov committed 1 year ago

Verified 841713e1

Fixed dequant precision issues in Q4_1 and Q5_1 (#9711)

OuadiElfarouki committed 1 year ago

Verified 56399714

ggml-backend : add device and backend reg interfaces (#9707)

slaren committed 1 year ago

Verified c83ad6d0

llama : reduce compile time and binary size (#9712)

ngxson committed 1 year ago

Verified a39ab216

[SYCL] Initial cmake support of SYCL for AMD GPUs (#9658)

Alcpz committed 1 year ago

Verified f536f4c4

vulkan : do not use tensor->extra (#9407)

rgerganov committed 1 year ago

Verified 00b7317e

gguf-split : improve --split and --merge logic (#9619)

kylo5aby committed 1 year ago

Verified 76b37d15

examples : remove benchmark (#9704)

ggerganov committed 1 year ago

Verified 148844fe

Update README.md (#9591)

32bitmicro committed 1 year ago

Verified 3f1ae2e3

ggerganov committed 1 year ago

Verified f1b8c427

test: fix OPT_STEP_ADAMW for test-backend-ops (ggml/974)

JohannesGaessler committed 1 year ago

Verified e98c1c18

vulkan : mul_mat: fix UB with small warps (ggml/952)

smeso committed 1 year ago

Verified cb000205

ggml : fix ggml_cast (ggml/973)

iboB committed 1 year ago

Verified 6c532248

ggml: fix gradient allocation logic (ggml/966)

JohannesGaessler committed 1 year ago

Verified 7254cdf7

metal : reduce command encoding overhead (#9698)

ggerganov committed 1 year ago

Verified cad341d8

llama : print correct model type for Llama 3.2 1B and 3B

ggerganov committed 1 year ago

Verified a90484c6

convert : refactor rope_freqs generation (#9396)

compilade committed 1 year ago

Verified 1927378b

Fix Docker ROCM builds, use AMDGPU_TARGETS instead of GPU_TARGETS (#9641)

serhii-nakon committed 1 year ago

Verified 6f1d9d71

ci : reduce severity of unused Pyright ignore comments (#9697)

compilade committed 1 year ago

Verified 511636df

py : update transfomers version (#9694)

Vaibhavs10 committed 1 year ago

Verified 08a43d05

flake.lock: Update (#9680)

ggerganov committed 1 year ago

Verified ace4f4be

console : utf-8 fix for windows stdin (#9690)

hasaranga committed 1 year ago

Verified 8277a817

ggml : define missing HWCAP flags (#9684)

ggerganov committed 1 year ago

Verified c919d5db

ggerganov committed 1 year ago

Verified d0b1d663

CUDA: remove bad assert (ggml/972)

JohannesGaessler committed 1 year ago

Verified aaa40999

Newer Older