ggml-org/llama.cpp

Pull Requests Commits

imatrix : avoid using imatrix.dat in README

compilade committed 1 year ago

942c55cd

imatrix : avoid loading model to convert or combine imatrix

compilade committed 1 year ago

183eeb55

imatrix : warn when writing partial data, to help guess dataset coverage

compilade committed 1 year ago

50f53b3e

imatrix : add warning when legacy format is written

compilade committed 1 year ago

42423ec4

Merge branch 'master' into compilade/imatrix-batched-chunks

compilade committed 1 year ago

0ee322cd

tests : cover lfm2 cases in test_ssm_conv (#14651)

tdakhran committed 1 year ago

Verified c31e6064

docs : add LFM2 to models section (#14650)

tdakhran committed 1 year ago

Verified 67eade1b

CUDA: add set rows for f32 and f16 (#14551)

am17an committed 1 year ago

Verified 7de5c7ca

ggerganov committed 1 year ago

8eff9554

vulkan : remove unused vars (#0)

ggerganov committed 1 year ago

3120413c

ggerganov committed 1 year ago

21553570

vulkan : implement bilinear interpolation (ggml/1291)

Acly committed 1 year ago

74bb2945

vulkan : implement ggml_roll (ggml/1290)

Acly committed 1 year ago

3e303b11

server : fix pooled embedding output (#14645)

iamlemec committed 1 year ago

Verified 0c1df14b

vulkan: support SET_ROWS (#14587)

jeffbolznv committed 1 year ago

Verified b3ad3a01

vulkan: optimizations for deepseek prompt processing (#14555)

jeffbolznv committed 1 year ago

Verified 98197e5c

model : support LiquidAI LFM2 hybrid family (#14620)

tdakhran committed 1 year ago

Verified f5e96b36

HIP : Add HIP 7.0+ compatibility for hipBLAS compute types (#14634)

slojosic-amd committed 1 year ago

Verified 756aa102

readme : add hot PRs (#14636)

ggerganov committed 1 year ago

Verified aaa088d8

llama : move enum llama_vocab_pre_type to implementation (#14631)

ggerganov committed 1 year ago

Verified 0d5375d5

vocab : add midm-2.0 model pre-tokenizer (#14626)

Bing-su committed 1 year ago

Verified 576c82ed

model : Granite Four (#13550)

gabe-l-hart committed 1 year ago

Verified 0aedae00

opencl: add tiled mul_mat_f16_f32 (#14535)

rmatif committed 1 year ago

Verified 6bdda139

opencl: add `set_rows` for `f16` and `f32` (#14547)

lhez committed 1 year ago

Verified 0b885577

Smoldocling support (#14597)

ryan-mangeno committed 1 year ago

Verified 4bb625b7

Docs: script to auto-generate ggml operations docs (#14598)

am17an committed 1 year ago

Verified 11ee0fea

cmake : do not search for curl libraries by ourselves (#14613)

EZForever committed 1 year ago

Verified a4575513

SYCL: Initial set_rows kernel implementation (#14562)

qnixsynapse committed 1 year ago

Verified 704bb7a7

llama : minor coding style fix for smollm3 (#14605)

ngxson committed 1 year ago

Verified 435a6d10

cmake : bump llguidance version to v1.0.1 (#14609)

EZForever committed 1 year ago

Verified f9a867f5

Older