How to get the stage 2 checkpoint path for 3_sft.sh

NVlabs / VILA

VILA - a multi-image visual language model with training, inference and evaluation recipe, deployable from cloud to edge (Jetson Orin and laptops)

Apache License 2.0

2.01k stars 161 forks source link

How to get the stage 2 checkpoint path for 3_sft.sh #143

Open Qnancy opened 1 month ago

Qnancy commented 1 month ago

I want to start training my own fine-tuning dataset from the stage 2 of VILA1.5-3b. I noticed in 3_sft.sh that there is a comment for the output of the stage 2 script:

# OUTPUT of stage 2 script
STAGE2_PATH=$1
# Final output checkpoint path
OUTPUT=$2

I would like to directly use your stage 2 checkpoints. How can I obtain them? looking forward to your reply.

ruojinhhhhhh commented 3 weeks ago

same question... do you get the access?

Qnancy commented 3 weeks ago

same question... do you get the access?

No one has replied to me yet...

98986oiuoy commented 1 week ago

where is 3_sft.sh???

ruojinhhhhhh commented 1 week ago

. scripts/v1_5/8b

Lyken17 commented 4 days ago

Usually you can continue training based stage3 checkpoints and will get higher performance.

For stage2 checkpoints, we are under legal process to release them.