Upload 2 files
Browse filespretrained planner model on openwebtext, GPT2-small architecture
owt_planner/.hydra/config.yaml
ADDED
|
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
ngpus: 4
|
| 2 |
+
tokens: 50257
|
| 3 |
+
compile: true
|
| 4 |
+
load_dir: null
|
| 5 |
+
work_dir: null
|
| 6 |
+
wandb:
|
| 7 |
+
project: openwebtext
|
| 8 |
+
run_name: pred_mask_nowe
|
| 9 |
+
id: null
|
| 10 |
+
is_resume: false
|
| 11 |
+
training:
|
| 12 |
+
batch_size: 256
|
| 13 |
+
accum: 2
|
| 14 |
+
n_iters: 1300001
|
| 15 |
+
snapshot_freq: 50000
|
| 16 |
+
log_freq: 50
|
| 17 |
+
eval_freq: 100
|
| 18 |
+
snapshot_freq_for_preemption: 10000
|
| 19 |
+
weight: standard
|
| 20 |
+
snapshot_sampling: true
|
| 21 |
+
ema: 0.9999
|
| 22 |
+
weighted_by_time: false
|
| 23 |
+
data:
|
| 24 |
+
train: openwebtext
|
| 25 |
+
valid: wikitext103
|
| 26 |
+
cache_dir: /pscratch/sd/s/sulinl/data
|
| 27 |
+
num_proc: 64
|
| 28 |
+
graph:
|
| 29 |
+
type: uniform
|
| 30 |
+
file: data
|
| 31 |
+
report_all: false
|
| 32 |
+
noise:
|
| 33 |
+
type: loglinear
|
| 34 |
+
sigma_min: 0.0001
|
| 35 |
+
sigma_max: 20
|
| 36 |
+
sampling:
|
| 37 |
+
predictor: euler
|
| 38 |
+
steps: 128
|
| 39 |
+
noise_removal: true
|
| 40 |
+
eval:
|
| 41 |
+
batch_size: 256
|
| 42 |
+
perplexity: true
|
| 43 |
+
perplexity_batch_size: 32
|
| 44 |
+
optim:
|
| 45 |
+
weight_decay: 0
|
| 46 |
+
optimizer: AdamW
|
| 47 |
+
lr: 0.0003
|
| 48 |
+
beta1: 0.9
|
| 49 |
+
beta2: 0.999
|
| 50 |
+
eps: 1.0e-08
|
| 51 |
+
warmup: 2500
|
| 52 |
+
grad_clip: 1.0
|
| 53 |
+
model:
|
| 54 |
+
name: small
|
| 55 |
+
type: ddit
|
| 56 |
+
hidden_size: 768
|
| 57 |
+
cond_dim: 128
|
| 58 |
+
length: 1024
|
| 59 |
+
n_blocks: 12
|
| 60 |
+
n_heads: 12
|
| 61 |
+
scale_by_sigma: true
|
| 62 |
+
dropout: 0.1
|
owt_planner/checkpoints-meta/checkpoint.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5023d21f1e052a1c495aeff266747a7142cfdd33250f3b8ed04e3f4d8cee7587
|
| 3 |
+
size 2095860040
|