ggml-org/llama.cpp

Pull Requests Commits

kompute : fix ggml_add kernel

ggerganov committed 2 years ago

Verified 14532151

fix supported ops for kompute backend

cebtenzzre committed 2 years ago

610394ff

never try to evaluate an empty command buffer

cebtenzzre committed 2 years ago

7addf2b8

sync op_rope_f16 with recent op_rope_f32 changes

cebtenzzre committed 2 years ago

16bc3c3b

actually fix this assertion

cebtenzzre committed 2 years ago

0f1a958a

clean up old backend code

cebtenzzre committed 2 years ago

a97935e0

kompute : fix rope_f32 and scale ops (#5008)

ggerganov committed 2 years ago

Verified 696faa86

kompute : ignore exceptions in ggml_vk_available_devices (#12)

cebtenzzre committed 2 years ago

02b9bafe

add sanity check and fix kompute teardown order

cebtenzzre committed 2 years ago

de9b0bbb

attempt to get test-backend-ops working

cebtenzzre committed 2 years ago

50579f27

fix assertion failure

cebtenzzre committed 2 years ago

8a99f698

kompute : initial attempt at ggml-backend v2 support

cebtenzzre committed 2 years ago

d5670d6e

cebtenzzre committed 2 years ago

1eb8804c

Merge branch 'master' of https://github.com/ggerganov/llama.cpp into ceb/nomic-vulkan

cebtenzzre committed 2 years ago

3773e1af

Merge commit 'd232aca5a73b290e218a2e48b91023d5e994203f' into ceb/nomic-vulkan

cebtenzzre committed 2 years ago

ae6d6824

sync xxd commands with GPT4All llama.cpp.cmake

cebtenzzre committed 2 years ago

904c563d

llama : add additional suffixes for model params (#4834)

mofosyne committed 2 years ago

Verified 57d016ba

llama : recognize 1B phi models (#4847)

teleprint-me committed 2 years ago

Verified 329ff615

clip : support more quantization types (#4846)

cmp-nct committed 2 years ago

Verified d34633d8

Python script to compare commits with llama-bench (#4844)

JohannesGaessler committed 2 years ago

Verified 4f56458d

convert.py : fix vanilla LLaMA model conversion (#4818)

teleprint-me committed 2 years ago

Verified 6efb8eb3

llava-cli : don't crash if --image flag is invalid (#4835)

jart committed 2 years ago

Verified 36e5a08b

metal : improve dequantize precision to match CPU (#4836)

ggerganov committed 2 years ago

Verified 4dccb38d

scripts : improve get-pg.sh (#4838)

ggerganov committed 2 years ago

Verified 9a818f7c

readme : add 3rd party collama reference to UI list (#4840)

iohub committed 2 years ago

Verified 18adb4e9

scripts : script to get Paul Graham essays in txt format (#4838)

ggerganov committed 2 years ago

Verified d9653894

server : update readme about token probs (#4777)

ibehnam committed 2 years ago

Verified 128de358

server : add api-key flag to documentation (#4832)

Zsapi committed 2 years ago

Verified 8c583303

ggml : fix vld1q_s8_x4 32-bit compat (#4828)

ggerganov committed 2 years ago

Verified 18c2e175

CUDA: faster softmax via shared memory + fp16 math (#4742)

JohannesGaessler committed 2 years ago

Verified 8f900abf

Older