Upload checkpoints
Browse files- ckpt_2500/config.json +20 -0
- ckpt_2500/model.safetensors +3 -0
- optimizer.pt +1 -1
ckpt_2500/config.json
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"GPT"
|
| 4 |
+
],
|
| 5 |
+
"batch_size": 4,
|
| 6 |
+
"context_len": 1024,
|
| 7 |
+
"d_model": 1024,
|
| 8 |
+
"device": "cuda",
|
| 9 |
+
"dtype": "float32",
|
| 10 |
+
"intermidiate_size": 4096,
|
| 11 |
+
"load_checkpoint": true,
|
| 12 |
+
"lr": 0.0006,
|
| 13 |
+
"model_type": "gpt_custom",
|
| 14 |
+
"n_epoch": 5,
|
| 15 |
+
"n_heads": 8,
|
| 16 |
+
"n_layers": 32,
|
| 17 |
+
"transformers_version": "4.56.1",
|
| 18 |
+
"vocab_size": 50304,
|
| 19 |
+
"weight_decay": 0.1
|
| 20 |
+
}
|
ckpt_2500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81407b20b1ac1795e17cb4505f3f2f7545bdada2548d85d1a0ea325b90497aac
|
| 3 |
+
size 2041186360
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4049086127
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f49f3bf47b5e59b2221daa9503a4da31f8bddeb2a6d082cd1ef2397d5c42643
|
| 3 |
size 4049086127
|