update ckpt
Browse files- config.json +5 -0
- pytorch_model.bin +1 -1
config.json
CHANGED
|
@@ -1,10 +1,12 @@
|
|
| 1 |
{
|
|
|
|
| 2 |
"activation_dropout": 0.0,
|
| 3 |
"activation_function": "gelu",
|
| 4 |
"architectures": [
|
| 5 |
"XGLMForCausalLM"
|
| 6 |
],
|
| 7 |
"attention_dropout": 0.1,
|
|
|
|
| 8 |
"bos_token_id": 0,
|
| 9 |
"classifier_dropout": 0.0,
|
| 10 |
"d_model": 1024,
|
|
@@ -16,9 +18,12 @@
|
|
| 16 |
"dropout": 0.1,
|
| 17 |
"encoder_layerdrop": 0.0,
|
| 18 |
"eos_token_id": 2,
|
|
|
|
| 19 |
"init_std": 0.02,
|
|
|
|
| 20 |
"max_position_embeddings": 2048,
|
| 21 |
"model_type": "xglm",
|
|
|
|
| 22 |
"pad_token_id": 1,
|
| 23 |
"scale_embedding": true,
|
| 24 |
"torch_dtype": "float32",
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "xglm-564M/",
|
| 3 |
"activation_dropout": 0.0,
|
| 4 |
"activation_function": "gelu",
|
| 5 |
"architectures": [
|
| 6 |
"XGLMForCausalLM"
|
| 7 |
],
|
| 8 |
"attention_dropout": 0.1,
|
| 9 |
+
"attention_heads": 16,
|
| 10 |
"bos_token_id": 0,
|
| 11 |
"classifier_dropout": 0.0,
|
| 12 |
"d_model": 1024,
|
|
|
|
| 18 |
"dropout": 0.1,
|
| 19 |
"encoder_layerdrop": 0.0,
|
| 20 |
"eos_token_id": 2,
|
| 21 |
+
"ffn_dim": 4096,
|
| 22 |
"init_std": 0.02,
|
| 23 |
+
"layerdrop": 0.0,
|
| 24 |
"max_position_embeddings": 2048,
|
| 25 |
"model_type": "xglm",
|
| 26 |
+
"num_layers": 24,
|
| 27 |
"pad_token_id": 1,
|
| 28 |
"scale_embedding": true,
|
| 29 |
"torch_dtype": "float32",
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2266383929
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33de0db4f71eed508d93035d14a6fd4d5b61c4bdfbb72f6c3d01f2ffdc7bff7c
|
| 3 |
size 2266383929
|