BrightBlueCheese
commited on
Commit
•
07ef7c9
1
Parent(s):
807b72c
pretrained-model
Browse files
ChemLlama_Medium_30m_vloss_val_loss=0.029_ep_epoch=04.ckpt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8796322ae9ff04cb8eebe5543a9182e734c98373b6b45145525addb14395481a
|
| 3 |
+
size 353459359
|
hparams.yaml
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
attention_dropout: 0.0
|
| 2 |
+
bos_token_id: 12
|
| 3 |
+
eos_token_id: 13
|
| 4 |
+
hidden_size: 768
|
| 5 |
+
intermediate_size: 768
|
| 6 |
+
learning_rate: 0.0001
|
| 7 |
+
max_epochs: 7
|
| 8 |
+
max_position_embeddings: 512
|
| 9 |
+
num_attention_heads: 8
|
| 10 |
+
num_hidden_layers: 7
|
| 11 |
+
num_labels: 105
|
| 12 |
+
pad_token_id: 0
|
| 13 |
+
steps_per_epoch: 375000
|
| 14 |
+
vocab_size: 591
|
| 15 |
+
warmup_epochs: 1
|