issues
search
microsoft
/
mttl
Building modular LMs with parameter-efficient fine-tuning.
MIT License
70
stars
6
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Sparse masks
#108
oleksost
opened
13 hours ago
0
Not sure about the details of arrow routing
#107
Weifan1226
opened
4 days ago
2
mixeval evaluator
#106
sordonia
opened
4 days ago
0
Modular policies support
#105
pclucas14
opened
6 days ago
0
Basic wrap fn
#104
pclucas14
closed
6 days ago
0
Huggingface trainer no delegate
#103
sordonia
closed
6 days ago
0
vector selector
#102
pclucas14
closed
6 days ago
0
[WIP] Circular Import
#101
matheper
closed
1 week ago
0
FIX: Filter ExpertContainer
#100
sordonia
closed
1 week ago
0
use gpt-neo-125m for testing
#99
matheper
closed
1 week ago
0
Packed batch
#98
sordonia
closed
2 weeks ago
5
Can we remove "modify_layers" argument? "Modify_modules" can be sufficient?
#97
oleksost
closed
1 week ago
1
Rename CoalescedLoRAExpertContainer to LoRAExpertContainer, e.g. remove LoraExpertContainer
#96
sordonia
opened
2 weeks ago
0
Huggingface trainer and huge refactoring of models and pretty much everything, lol
#95
sordonia
closed
2 days ago
0
Not a bug but a feature
#94
pclucas14
closed
2 weeks ago
0
PEER MoE
#93
oleksost
closed
1 week ago
0
Em topic
#92
sordonia
closed
2 weeks ago
0
Moe training fixes
#91
oleksost
closed
3 weeks ago
0
[Draft] Split `qkv_proj` into `q_proj`, `k_proj` and `v_proj`
#90
pclucas14
closed
2 days ago
1
HardPromptConfig takes tokenizer which is a complex arg
#89
sordonia
opened
3 weeks ago
0
Update test_routed_multi_expert_model.py
#88
sordonia
closed
3 weeks ago
0
Remove decorator for coalesced
#87
sordonia
closed
3 weeks ago
0
DRAFT: Load from peft
#86
sordonia
opened
3 weeks ago
0
CLEAN: Move all the functions from projects/modular_llm into mttl
#85
sordonia
closed
3 weeks ago
0
warmup no decay scheduler
#84
pclucas14
closed
3 weeks ago
1
Config refactor
#83
sordonia
closed
3 weeks ago
1
Speedup tests
#82
sordonia
closed
3 weeks ago
0
make datamodule registrable
#81
sordonia
closed
3 weeks ago
0
MTTL installer
#80
matheper
closed
3 weeks ago
0
Fix circular import
#79
sordonia
closed
4 weeks ago
0
Dockerfile torch 24 07
#78
matheper
closed
3 weeks ago
1
Upgrade base image and python version
#77
matheper
closed
1 month ago
0
A bit more cleaning
#76
sordonia
closed
1 month ago
0
Selectors cache
#75
sordonia
closed
1 month ago
0
Fix task to expert
#74
sordonia
closed
1 month ago
0
fix tests
#73
sordonia
closed
1 month ago
0
May I get the trained LoRA lib to reproduce the paper:Modular LLM?
#72
Weifan1226
closed
3 weeks ago
3
Registrable
#71
sordonia
closed
1 month ago
0
cluster by embedding
#70
shuishen112
closed
12 hours ago
0
Adds info container as forward context
#69
sordonia
closed
1 month ago
0
Fixed import and circular dependency
#68
matheper
closed
1 month ago
0
Mod-Chatbot
#67
sordonia
opened
1 month ago
0
Code restructuring 1
#66
matheper
closed
1 month ago
0
[WIP] Sequence packing
#65
pclucas14
closed
1 month ago
0
From pretrained library
#64
sordonia
closed
1 month ago
0
Dpo training (Do not merge)
#63
shuishen112
opened
1 month ago
2
Coalesced container fix
#62
pclucas14
closed
1 month ago
0
Black and isort formatters
#61
matheper
closed
1 month ago
0
Run tests in parallel
#60
matheper
closed
1 month ago
0
rename folder and move expert_containers one level up
#59
sordonia
closed
1 month ago
0
Next