Upload 2 files
Browse files- .gitattributes +1 -0
- babyGPT_152M +3 -0
- babyGPT_152M_config +24 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
babyGPT_152M filter=lfs diff=lfs merge=lfs -text
|
babyGPT_152M
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c0d3981c27a435802c336c1edaa052f94a432bf55c92d006ec70978a94b6ec23
|
| 3 |
+
size 678700810
|
babyGPT_152M_config
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
batch_size: 6
|
| 2 |
+
ckpt_config: config_tmp
|
| 3 |
+
ckpt_dir: model/checkpoints
|
| 4 |
+
ckpt_model: model_tmp
|
| 5 |
+
current_shard: 43
|
| 6 |
+
device: cuda
|
| 7 |
+
dropout_rate: 0.2
|
| 8 |
+
dtype: long
|
| 9 |
+
eval_inter: 200
|
| 10 |
+
eval_iter: 10
|
| 11 |
+
log_file: log.txt
|
| 12 |
+
log_inter: 20
|
| 13 |
+
lr: 0.0006
|
| 14 |
+
max_iter: 100000
|
| 15 |
+
max_loss: 3.0736479312181473
|
| 16 |
+
n_batches: 16
|
| 17 |
+
n_blocks: 16
|
| 18 |
+
n_embd: 768
|
| 19 |
+
n_head: 8
|
| 20 |
+
seq_len: 1024
|
| 21 |
+
tokenizer_type: gpt2
|
| 22 |
+
training_duration: 225724.18966698647
|
| 23 |
+
training_step: 48827
|
| 24 |
+
vocab_size: 50304
|