issues
search
chrisociepa
/
allamo
Simple, hackable and fast implementation for training/finetuning medium-sized LLaMA-based models
MIT License
152
stars
15
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Add support for FSDP2 and (experimental) TP
#16
chrisociepa
closed
1 month ago
0
Major refactor
#15
chrisociepa
closed
2 months ago
0
Add log-probability metrics to DPO training
#14
djstrong
closed
3 months ago
0
more dpo logs
#13
djstrong
closed
3 months ago
0
dpo fix
#12
djstrong
closed
3 months ago
0
Optimize SFT dataset packing: correct RoPE encoding and without cross-contamination
#11
chrisociepa
closed
3 months ago
0
Add support for weighted token-level loss and adaptive learning rate during masked token instruction training
#10
chrisociepa
closed
7 months ago
0
Bulk update with new features and improvements
#9
chrisociepa
closed
1 year ago
0
Consider adding support for openlm-research/open_llama_7b and _3b
#8
RDouglasSharp
closed
8 months ago
1
Import_llama_weights.py code bug
#7
RDouglasSharp
closed
1 year ago
1
Sample json file mentioned in README.md is not part of repository
#6
RDouglasSharp
closed
1 year ago
1
[feature request] usage of trained model in python script
#5
phineas-pta
closed
1 year ago
2
text classification?
#4
arpitest
closed
1 year ago
2
finetune for chatbot
#3
wallon-ai
closed
1 year ago
2
Where to find all files mentioned?
#2
qizzzh
closed
1 year ago
1
train.txt format
#1
AAnirudh07
closed
1 year ago
2