Skip to content
New issue

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

Sign up for GitHub

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

Already on GitHub? Sign in to your account

add cool down config #161

Merged
merged 1 commit into from
Nov 15, 2024
Merged

add cool down config #161

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
38 changes: 38 additions & 0 deletions configs/10B/H100_cooldown.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
name_model = "10B"
project = "10B_zero_band"
wandb_resume = false

[train]
micro_bs = 1
ac_ckpt = true

[optim]
sched_type = "wsd-sqrt"
batch_size = 128 #1M tokens bs
warmup_steps = 1000
num_stable_steps = 74700
Copy link
Collaborator Author

@samsja samsja Nov 15, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we are at 74500 right now

total_steps = 90400
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(90400 - 74700 ) * 8192 * 128* 14 / 1e9 = 230.5B tokens.

We are at 770B right now. Picking 90400 and not 90300 to be sure we hit the 1T = 100% on the dashboard

lr = 7.5e-5

adam_betas1 = 0.9
adam_betas2 = 0.95
weight_decay = 0.1

z_loss = true

[data]
seq_length = 8192
dataset_name_or_paths = "/data/datasets/fineweb-edu,/data/datasets/fineweb,/data/datasets/StackV1-popular"
dataset_ratio = "80:10:10"
num_workers = 4
reverse_data_files = false

[diloco]
inner_steps = 100
compression = "uint8"

[ckpt]
interval = 100
topk = 40
path = "/data/10B"
remote_data_path = "/data/10B_data_ckpt"