PR #820 Add LoRA support

Add LoRA support #820

slaren merged 15 commits into ggml-org:master from slaren:lora

slaren changed the title ~~Add lora support~~ Add LoRA support 2 years ago

slaren force pushed 2 years ago

ggerganov added research 🔬

slaren force pushed to 0d8999a2 2 years ago

slaren force pushed from 0d8999a2 to 671190b0 2 years ago

jon-chuang commented on 2023-04-13

ggerganov assigned

slaren 2 years ago

slaren force pushed from 671190b0 2 years ago

jon-chuang commented on 2023-04-15

slaren marked this pull request as ready for review 2 years ago

slaren requested a review from

sw 2 years ago

slaren requested a review from

ggerganov 2 years ago

jon-chuang commented on 2023-04-16

ggerganov commented on 2023-04-16

Add lora support

f52101e8

Export lora A matrix pre-transposed

ac3fbe49

Add support for quantized models

7136adac

Use the work buffer instead to fix MSVC build

dc657071

Update exporter and support scaling

87c518bb

Add compatibility with #801

c920f001

Support more layer types, fix memory and generation issues

c45868ba

Rebase to master

57627f0e

Add support for using a different base model

c150e1b0

Fix windows build

fc899160

Show warning when using a quantized base model

14858ba2

ggml_cpy: use the work buffer instead of alloca when quantizing

3df343b4

Only attempt to use mmap for the lora base model if it is supported

63da54e0

Reuse definitions from convert.py

0a6d5ad7

ggml_add: Add more checks

8d37db3c

slaren force pushed to 8d37db3c 2 years ago

ggerganov approved these changes on 2023-04-17

slaren merged 315a95a4 into master 2 years ago

slaren deleted the lora branch 2 years ago

Reviewers

ggerganov

jon-chuang

MillionthOdin16

Assignees

slaren

Labels

research 🔬

Milestone

No milestone

llama.cpp Add LoRA support #820 Merged

Add LoRA support #820

llama.cpp
Add LoRA support
#820

Merged