Upload 2 files
Browse files
quiet-thunder-789--checkpoint:v0/config.yaml
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
eval:
|
| 2 |
+
eval_datasets: null
|
| 3 |
+
json_datasets: null
|
| 4 |
+
test_datasets:
|
| 5 |
+
- name: generator_mean
|
| 6 |
+
length: 96
|
| 7 |
+
generator: PATTERN
|
| 8 |
+
num_pairs: 4
|
| 9 |
+
batch_size: 96
|
| 10 |
+
num_tasks_to_show: 32
|
| 11 |
+
task_generator_kwargs:
|
| 12 |
+
num_cols: 4
|
| 13 |
+
num_rows: 4
|
| 14 |
+
pattern_size: 2
|
| 15 |
+
- name: generator_gradient_ascent_5
|
| 16 |
+
length: 96
|
| 17 |
+
generator: PATTERN
|
| 18 |
+
num_pairs: 4
|
| 19 |
+
batch_size: 96
|
| 20 |
+
inference_mode: gradient_ascent
|
| 21 |
+
inference_kwargs:
|
| 22 |
+
lr: 0.1
|
| 23 |
+
num_steps: 10
|
| 24 |
+
num_tasks_to_show: 32
|
| 25 |
+
task_generator_kwargs:
|
| 26 |
+
num_cols: 4
|
| 27 |
+
num_rows: 4
|
| 28 |
+
pattern_size: 2
|
| 29 |
+
training:
|
| 30 |
+
seed: 0
|
| 31 |
+
use_hf: true
|
| 32 |
+
kl_coeff: 0.001
|
| 33 |
+
batch_size: 128
|
| 34 |
+
learning_rate: 0.0004
|
| 35 |
+
inference_mode: mean
|
| 36 |
+
task_generator:
|
| 37 |
+
class: PATTERN
|
| 38 |
+
num_cols: 4
|
| 39 |
+
num_rows: 4
|
| 40 |
+
num_pairs: 4
|
| 41 |
+
num_workers: 16
|
| 42 |
+
pattern_size: 2
|
| 43 |
+
train_datasets: null
|
| 44 |
+
mixed_precision: false
|
| 45 |
+
total_num_steps: 200000
|
| 46 |
+
inference_kwargs: null
|
| 47 |
+
eval_every_n_logs: 20
|
| 48 |
+
log_every_n_steps: 1000
|
| 49 |
+
resume_from_checkpoint: null
|
| 50 |
+
online_data_augmentation: false
|
| 51 |
+
gradient_accumulation_steps: 1
|
| 52 |
+
save_checkpoint_every_n_logs: 200
|
| 53 |
+
decoder_transformer:
|
| 54 |
+
_target_: src_v2.models.utils.DecoderTransformerConfig
|
| 55 |
+
max_cols: 4
|
| 56 |
+
max_rows: 4
|
| 57 |
+
num_layers: 2
|
| 58 |
+
transformer_layer:
|
| 59 |
+
_target_: src_v2.models.utils.TransformerLayerConfig
|
| 60 |
+
num_heads: 6
|
| 61 |
+
dropout_rate: 0.0
|
| 62 |
+
mlp_dim_factor: 4.0
|
| 63 |
+
emb_dim_per_head: 12
|
| 64 |
+
attention_dropout_rate: 0.0
|
| 65 |
+
encoder_transformer:
|
| 66 |
+
_target_: src_v2.models.utils.EncoderTransformerConfig
|
| 67 |
+
max_cols: 4
|
| 68 |
+
max_rows: 4
|
| 69 |
+
latent_dim: 2
|
| 70 |
+
num_layers: 2
|
| 71 |
+
variational: true
|
| 72 |
+
transformer_layer:
|
| 73 |
+
_target_: src_v2.models.utils.TransformerLayerConfig
|
| 74 |
+
num_heads: 6
|
| 75 |
+
dropout_rate: 0.0
|
| 76 |
+
mlp_dim_factor: 4.0
|
| 77 |
+
emb_dim_per_head: 12
|
| 78 |
+
attention_dropout_rate: 0.0
|
| 79 |
+
latent_projection_bias: false
|
quiet-thunder-789--checkpoint:v0/state.msgpack
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60b3368d8a3c6232867a43b252597f16dbec5de0264bf23168d938a461256a3d
|
| 3 |
+
size 4082621
|