issues
search
CoinCheung
/
gdGPT
Train llm (bloom, llama, baichuan2-7b, chatglm3-6b) with deepspeed pipeline mode. Faster than zero/zero++/fsdp.
Apache License 2.0
88
stars
8
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Questions about TiedLayerSpec
#31
josephwong14wkh
opened
2 months ago
0
Multi-node model training
#30
pipijiev12
opened
4 months ago
3
support mixtral-8x7b, and a bit fix to match new version transformers
#29
CoinCheung
closed
5 months ago
0
Dev
#28
CoinCheung
closed
6 months ago
0
It was found that the deepspeed folder exists. Has the deepspeed source code been modified in this project?
#27
daneren
opened
6 months ago
3
add support for chatglm3-6b
#26
CoinCheung
closed
7 months ago
0
Is there any plan to support chatglm2 or yi
#25
yangzhipeng1108
closed
7 months ago
2
can it make Lora sft?
#24
ReverseSystem001
opened
8 months ago
2
support baichuan2-7b
#23
CoinCheung
closed
8 months ago
0
update to deepspeed 0.11.1
#22
CoinCheung
closed
8 months ago
0
scaled dot product for bloom
#21
CoinCheung
closed
9 months ago
0
Dev
#20
CoinCheung
closed
9 months ago
0
add max_z_loss
#19
CoinCheung
closed
9 months ago
0
Any plan to incorporate tensor parallelism or zero data parallelism?
#18
GeneZC
opened
10 months ago
2
看了您的碎碎念
#17
mc112611
closed
10 months ago
0
small fix
#16
CoinCheung
closed
10 months ago
0
speed up and refine
#15
CoinCheung
closed
10 months ago
0
small fix and refine
#14
CoinCheung
closed
11 months ago
0
init weight
#13
CoinCheung
closed
11 months ago
0
ninja -v指令出错导致transformer_inference.so文件缺失
#12
Debouter
opened
11 months ago
4
Dev
#11
CoinCheung
closed
11 months ago
0
Dev
#10
CoinCheung
closed
11 months ago
0
refine dataset format
#9
CoinCheung
closed
11 months ago
0
Dev
#8
CoinCheung
closed
11 months ago
0
llama support flash-attention
#7
CoinCheung
closed
11 months ago
0
use deepspeed 0.10.0
#6
CoinCheung
closed
11 months ago
0
add speed to readme
#5
CoinCheung
closed
1 year ago
0
Dev
#4
CoinCheung
closed
1 year ago
0
Dev
#3
CoinCheung
closed
1 year ago
0
Dev
#2
CoinCheung
closed
1 year ago
0
refine readme
#1
CoinCheung
closed
1 year ago
0