Commit
·
99a290e
1
Parent(s):
9501448
Upload ScriptableLMForPreTraining
Browse files- config.json +131 -0
- model.safetensors +3 -0
config.json
ADDED
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"arch": {
|
3 |
+
"architectures": [
|
4 |
+
"ScriptableCrammedBERT"
|
5 |
+
],
|
6 |
+
"attention": {
|
7 |
+
"causal_attention": false,
|
8 |
+
"dropout_prob": 0.1,
|
9 |
+
"kernel_stride_by_layer": [
|
10 |
+
[
|
11 |
+
1,
|
12 |
+
1
|
13 |
+
],
|
14 |
+
[
|
15 |
+
3,
|
16 |
+
2
|
17 |
+
],
|
18 |
+
[
|
19 |
+
3,
|
20 |
+
2
|
21 |
+
],
|
22 |
+
[
|
23 |
+
3,
|
24 |
+
2
|
25 |
+
],
|
26 |
+
[
|
27 |
+
3,
|
28 |
+
2
|
29 |
+
],
|
30 |
+
[
|
31 |
+
3,
|
32 |
+
2
|
33 |
+
],
|
34 |
+
[
|
35 |
+
3,
|
36 |
+
2
|
37 |
+
],
|
38 |
+
[
|
39 |
+
1,
|
40 |
+
1
|
41 |
+
],
|
42 |
+
[
|
43 |
+
1,
|
44 |
+
0.5
|
45 |
+
],
|
46 |
+
[
|
47 |
+
3,
|
48 |
+
0.5
|
49 |
+
],
|
50 |
+
[
|
51 |
+
3,
|
52 |
+
0.5
|
53 |
+
],
|
54 |
+
[
|
55 |
+
3,
|
56 |
+
0.5
|
57 |
+
],
|
58 |
+
[
|
59 |
+
3,
|
60 |
+
0.5
|
61 |
+
],
|
62 |
+
[
|
63 |
+
3,
|
64 |
+
0.5
|
65 |
+
],
|
66 |
+
[
|
67 |
+
3,
|
68 |
+
1
|
69 |
+
],
|
70 |
+
[
|
71 |
+
1,
|
72 |
+
1
|
73 |
+
]
|
74 |
+
],
|
75 |
+
"num_attention_heads": 12,
|
76 |
+
"qkv_bias": false,
|
77 |
+
"rotary_embedding": false,
|
78 |
+
"seq_op_in_fp32": false,
|
79 |
+
"sequence_op": "torch-softmax",
|
80 |
+
"skip_output_projection": false,
|
81 |
+
"type": "self-attention"
|
82 |
+
},
|
83 |
+
"classification_head": {
|
84 |
+
"classifier_dropout": 0.1,
|
85 |
+
"head_dim": 1024,
|
86 |
+
"include_ff_layer": true,
|
87 |
+
"nonlin": "Tanh",
|
88 |
+
"pooler": "avg"
|
89 |
+
},
|
90 |
+
"decoder_bias": false,
|
91 |
+
"embedding": {
|
92 |
+
"dropout_prob": 0.1,
|
93 |
+
"embedding_dim": 768,
|
94 |
+
"max_seq_length": 128,
|
95 |
+
"normalization": true,
|
96 |
+
"pad_token_id": 0,
|
97 |
+
"pos_embedding": "scaled-sinusoidal",
|
98 |
+
"stable_low_precision": false,
|
99 |
+
"vocab_size": 32768
|
100 |
+
},
|
101 |
+
"ffn_layer_frequency": 1,
|
102 |
+
"final_norm": true,
|
103 |
+
"hidden_dropout_prob": 0.1,
|
104 |
+
"hidden_size": 768,
|
105 |
+
"init": {
|
106 |
+
"std": 0.02,
|
107 |
+
"type": "normal"
|
108 |
+
},
|
109 |
+
"intermed_size": 3072,
|
110 |
+
"loss": "cross-entropy",
|
111 |
+
"nonlin": "GELUglu",
|
112 |
+
"norm": "LayerNorm",
|
113 |
+
"norm_eps": 1e-12,
|
114 |
+
"norm_scheme": "pre",
|
115 |
+
"num_labels": null,
|
116 |
+
"num_transformer_layers": 16,
|
117 |
+
"objective_layout": "MLM",
|
118 |
+
"residual_add": "interp_avg",
|
119 |
+
"skip_head_transform": true,
|
120 |
+
"sparse_prediction": 0.25,
|
121 |
+
"tie_weights": true,
|
122 |
+
"unet_add": true,
|
123 |
+
"use_bias": false
|
124 |
+
},
|
125 |
+
"architectures": [
|
126 |
+
"ScriptableLMForPreTraining"
|
127 |
+
],
|
128 |
+
"model_type": "crammedBERT",
|
129 |
+
"torch_dtype": "float32",
|
130 |
+
"transformers_version": "4.27.1"
|
131 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0079a2e4875cd3a6e8c14daaf3139fbb4850c97dc2fb5d88726315c1a3f20124
|
3 |
+
size 635664276
|