Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

AR-Diffusion data.name and exp.name #75

Open
AIgods opened this issue Feb 24, 2024 · 2 comments
Open

AR-Diffusion data.name and exp.name #75

AIgods opened this issue Feb 24, 2024 · 2 comments

Comments

@AIgods
Copy link

AIgods commented Feb 24, 2024

in AR Difusion, I would like to ask why in train.sh, in the training script for the commongen dataset, data.name is not commongen, but iwslt14_tok, and What does exp.name mean

Commongen

FILE_NAME = commongen
STEP = 40000

torchrun --nproc_per_node=8 --nnodes=1 ./train_utils/trainer_main.py
model.name='bert-base-uncased' batch_size=384 grad_accum=1
total_steps=$STEP exp.name=$FILE_NAME
data.name=iwslt14_tok tgt_len=54 max_pos_len=128 lr=3e-4 lr_step=40000
intermediate_size=512 num_attention_heads=8 dropout=0.2
in_channels=64 out_channels=64 time_channels=64
eval_interval=3000 log_interval=1000
schedule_sampler='xy_uniform' time_att=True att_strategy='txl' \

@hwaseem04
Copy link

Afaik, exp.name is something that you give as name for the experiment you are doing. Ideally it can be any arbitary string value.

@wutong4012
Copy link
Collaborator

Sorry, there is a mistake here, data.name should really be replaced with commongen.

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants