ochafik/llama.cpp

Pull Requests Commits

grammars: linked stacks to avoid gazillions of vector copies

ochafik committed 1 year ago

cb2c09ce

grammar: renames

ochafik committed 1 year ago

1d77d02d

grammar: faster llama_grammar_copy

ochafik committed 1 year ago

d8eb1674

grammars: early exit when no next_candidates to reject

ochafik committed 1 year ago

82dcbef1

grammars: cache decoded tokens

ochafik committed 1 year ago

e449ab24

ci: add ubuntu latest release and fix missing build number (mac & ubuntu) (#6748)

loonerin committed 1 year ago

Verified 0e4802b2

server: static: upstream upgrade (#6765)

phymbert committed 1 year ago

Verified 637e9a86

Implement the OLMo architecture (#6741)

nopperl committed 1 year ago

Verified 9958c81b

train : add general name (#6752)

teleprint-me committed 1 year ago

Verified 8b1b1f49

fix wrong parameter in cmd in readme-sycl.md (#6755)

arthw committed 1 year ago

Verified bca40e98

ggml : group all experts in a single ggml_mul_mat_id (#6505)

slaren committed 1 year ago

Verified 0d56246f

convert : support models with multiple chat templates (#6588)

CISC committed 1 year ago

Verified 03c0946d

Qwen2 : assume tied weights if lm_head/output weights is missing (#6738)

jklj077 committed 1 year ago

Verified e11b2e6e

llama : fix compatibility with old 2 expert models (#6735)

slaren committed 1 year ago

Verified c71bfd73

llamafile : tmp disable + build sgemm.o when needed (#6716)

ggerganov committed 1 year ago

Verified 3b8f1ec4

readme : add UI (#6724)

yaroslavyaroslav committed 1 year ago

Verified 8dd1ec8b

convert : fix autoawq gemma (#6704)

dengzheng-cloud committed 1 year ago

Verified facb8b56

llama : make general.name optional (#6709)

ggerganov committed 1 year ago

Verified 532c1737

ggml : fix llamafile sgemm wdata offsets (#6710)

ggerganov committed 1 year ago

Verified 666867b7

ggml : add llamafile sgemm (#6414)

jart committed 1 year ago

Verified 8cc91dc6

llama : add StableLM2 12B (#6635)

ashishdatta committed 1 year ago

Verified dbceec87

llama : add qwen2moe (#6074)

simonJJJ committed 1 year ago

Verified f4dea7da

gritlm : add --outdir option to hf.sh script (#6699)

danbev committed 1 year ago

Verified 8a56075b

perplexity : require positive --ctx-size arg (#6695)

ggerganov committed 1 year ago

Verified 58227ffd

gguf : add special tokens metadata for FIM/Infill (#6689)

danbev committed 1 year ago

Verified 4fbd8098

`main`: add --json-schema / -j flag (#6659)

ochafik committed 1 year ago

Verified 7593639c

llama : fix restoring the number of outputs from state files (#6687)

compilade committed 1 year ago

Verified 132f5579

server : revert "minor layout improvements" (#6684)

phymbert committed 1 year ago

Verified 3272896d

swift : linux support (#6590)

spprichard committed 1 year ago

Verified 7fc16a2c

fix mul_mat_id() for new input, make the ut pass (#6682)

NeoZhangJianyu committed 1 year ago

Verified 17e98d4c

Older