ggerganov/llama.cpp

Pull Requests Commits

llama : minor indentation during tensor loading

ggerganov committed 1 year ago

Verified f55b6473

llama : fix t5 uses of n_head and n_ff

compilade committed 1 year ago

18e92879

Merge branch 'master' into openelm

compilade committed 1 year ago

c6ac1984

llama : use const ref for print_f and fix division by zero

compilade committed 1 year ago

269e07bb

build: Export hf-to-gguf as snakecase

ditsuke committed 1 year ago

51d2ebad

doc: Add context for why we add an explicit pytorch source

ditsuke committed 1 year ago

1e920018

chore: Remove rebase artifacts

ditsuke committed 1 year ago

01a5f065

chore: Fixup requirements and build

ditsuke committed 1 year ago

07786a61

chore: ignore all __pychache__

ditsuke committed 1 year ago

de14e2ea

fix: Update script paths in CI scripts

ditsuke committed 1 year ago

82192291

fix: Actually include scripts in build

ditsuke committed 1 year ago

b1c3f26e

build(python): Package scripts with pip-0517 compliance

ditsuke committed 1 year ago

b0a46993

Merge branch 'master' into pr/7359

ggerganov committed 1 year ago

Verified 199d0fb0

llama : handle n_head == 0

ggerganov committed 1 year ago

Verified 3fe395d2

Inference support for T5 and FLAN-T5 model families (#5763)

fairydreaming committed 1 year ago

Verified 807b0c49

Merge branch 'master' into pr/7359

ggerganov committed 1 year ago

Verified 22a648f8

llama : do not print hparams for vocab-only models

ggerganov committed 1 year ago

Verified 9971c38a

llama : fix save/load state

ggerganov committed 1 year ago

Verified b59ddf94

llama : use std::array for per-layer hparams

ggerganov committed 1 year ago

Verified 29ab5a0e

tests : add _CRT_SECURE_NO_WARNINGS for WIN32 (#8231)

danbev committed 1 year ago

Verified f8c4c073

llama : suppress unref var in Windows MSVC (#8150)

danbev committed 1 year ago

Verified 402d6fef

convert : fix gemma v1 tokenizer convert (#8248)

ggerganov committed 1 year ago

Verified 20fc3804

[SYCL] Remove unneeded semicolons (#8280)

AidanBeltonS committed 1 year ago

Verified f6190247

Define and optimize RDNA1 (#8085)

daniandtheweb committed 1 year ago

Verified d23287f1

ppl : fix n_seq_max for perplexity (#8277)

slaren committed 1 year ago

Verified 5f2d4e60

fix phi 3 conversion (#8262)

ngxson committed 1 year ago

Verified 916248af

fix typo (#8267)

foldl committed 1 year ago

Verified f8d6a238

Dequant improvements rebase (#8255)

AidanBeltonS committed 1 year ago

Verified fadde671

fix: add missing short command line argument -mli for multiline-input (#8261)

MistApproach committed 1 year ago

Verified a27152b6

Adding step to `clean` target to remove legacy binary names to reduce upgrade / migration confusion arising from #7809. (#8257)

HanClinto committed 1 year ago

Verified 3e2618bc

Older