Upload astropt/p32k01/hparams.txt with huggingface_hub
Browse files- astropt/p32k01/hparams.txt +43 -0
astropt/p32k01/hparams.txt
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
AstroPT-0097.4M
|
| 2 |
+
time: 1740652910
|
| 3 |
+
log_via_wandb: True
|
| 4 |
+
log_emissions: False
|
| 5 |
+
out_dir: logs/astropt070M_p32k01
|
| 6 |
+
eval_interval: 100
|
| 7 |
+
log_interval: 10
|
| 8 |
+
checkpoint_interval: 500
|
| 9 |
+
eval_iters: 100
|
| 10 |
+
eval_only: False
|
| 11 |
+
always_save_checkpoint: False
|
| 12 |
+
init_from: scratch
|
| 13 |
+
hf_url: smith42/galaxies
|
| 14 |
+
stream_hf_dataset: False
|
| 15 |
+
gradient_accumulation_steps: 40
|
| 16 |
+
batch_size: 64
|
| 17 |
+
spiral: True
|
| 18 |
+
block_size: 64
|
| 19 |
+
image_size: 256
|
| 20 |
+
num_workers: 64
|
| 21 |
+
n_layer: 12
|
| 22 |
+
n_head: 12
|
| 23 |
+
n_embd: 768
|
| 24 |
+
n_chan: 3
|
| 25 |
+
dropout: 0.0
|
| 26 |
+
patch_size: 32
|
| 27 |
+
bias: False
|
| 28 |
+
attn_type: causal
|
| 29 |
+
k_ratio: 0.01
|
| 30 |
+
learning_rate: 0.0006
|
| 31 |
+
max_iters: 3500
|
| 32 |
+
weight_decay: 0.1
|
| 33 |
+
beta1: 0.9
|
| 34 |
+
beta2: 0.95
|
| 35 |
+
grad_clip: 1.0
|
| 36 |
+
decay_lr: True
|
| 37 |
+
warmup_iters: 2000
|
| 38 |
+
lr_decay_iters: 3300.0000000000005
|
| 39 |
+
min_lr: 5.9999999999999995e-05
|
| 40 |
+
backend: nccl
|
| 41 |
+
device: cuda
|
| 42 |
+
dtype: bfloat16
|
| 43 |
+
compile: True
|