Upload NystromformerForMaskedLM
Browse files- config.json +28 -0
- pytorch_model.bin +3 -0
config.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "my_nystromformer-4096_untrained",
|
3 |
+
"architectures": [
|
4 |
+
"NystromformerForMaskedLM"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"bos_token_id": 0,
|
8 |
+
"conv_kernel_size": 65,
|
9 |
+
"eos_token_id": 2,
|
10 |
+
"hidden_act": "gelu_new",
|
11 |
+
"hidden_dropout_prob": 0.1,
|
12 |
+
"hidden_size": 768,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 3072,
|
15 |
+
"inv_coeff_init_option": false,
|
16 |
+
"layer_norm_eps": 1e-05,
|
17 |
+
"max_position_embeddings": 4096,
|
18 |
+
"model_type": "nystromformer",
|
19 |
+
"num_attention_heads": 12,
|
20 |
+
"num_hidden_layers": 12,
|
21 |
+
"num_landmarks": 64,
|
22 |
+
"pad_token_id": 1,
|
23 |
+
"segment_means_seq_len": 64,
|
24 |
+
"torch_dtype": "float32",
|
25 |
+
"transformers_version": "4.22.1",
|
26 |
+
"type_vocab_size": 2,
|
27 |
+
"vocab_size": 50265
|
28 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:413c681e7602dcb6f9ecb312ffc6571700605649fe12fbbd8fe0438eeb764cc7
|
3 |
+
size 509943187
|