Commit
·
a8dcf9a
1
Parent(s):
fa5bd70
Upload ScriptableLMForPreTraining
Browse files- config.json +63 -0
- model.safetensors +3 -0
config.json
ADDED
|
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arch": {
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ScriptableCrammedBERT"
|
| 5 |
+
],
|
| 6 |
+
"attention": {
|
| 7 |
+
"causal_attention": false,
|
| 8 |
+
"dropout_prob": 0.1,
|
| 9 |
+
"num_attention_heads": 12,
|
| 10 |
+
"qkv_bias": false,
|
| 11 |
+
"rotary_embedding": false,
|
| 12 |
+
"seq_op_in_fp32": false,
|
| 13 |
+
"sequence_op": "torch-softmax",
|
| 14 |
+
"skip_output_projection": false,
|
| 15 |
+
"type": "self-attention"
|
| 16 |
+
},
|
| 17 |
+
"classification_head": {
|
| 18 |
+
"classifier_dropout": 0.1,
|
| 19 |
+
"head_dim": 1024,
|
| 20 |
+
"include_ff_layer": true,
|
| 21 |
+
"nonlin": "Tanh",
|
| 22 |
+
"pooler": "avg"
|
| 23 |
+
},
|
| 24 |
+
"decoder_bias": false,
|
| 25 |
+
"embedding": {
|
| 26 |
+
"dropout_prob": 0.1,
|
| 27 |
+
"embedding_dim": 768,
|
| 28 |
+
"max_seq_length": 128,
|
| 29 |
+
"normalization": true,
|
| 30 |
+
"pad_token_id": 0,
|
| 31 |
+
"pos_embedding": "scaled-sinusoidal",
|
| 32 |
+
"stable_low_precision": false,
|
| 33 |
+
"vocab_size": 32768
|
| 34 |
+
},
|
| 35 |
+
"ffn_layer_frequency": 1,
|
| 36 |
+
"final_norm": true,
|
| 37 |
+
"hidden_dropout_prob": 0.1,
|
| 38 |
+
"hidden_size": 768,
|
| 39 |
+
"init": {
|
| 40 |
+
"std": 0.02,
|
| 41 |
+
"type": "normal"
|
| 42 |
+
},
|
| 43 |
+
"intermed_size": 3072,
|
| 44 |
+
"loss": "cross-entropy",
|
| 45 |
+
"nonlin": "GELUglu",
|
| 46 |
+
"norm": "LayerNorm",
|
| 47 |
+
"norm_eps": 1e-12,
|
| 48 |
+
"norm_scheme": "pre",
|
| 49 |
+
"num_labels": null,
|
| 50 |
+
"num_transformer_layers": 16,
|
| 51 |
+
"objective_layout": "MLM",
|
| 52 |
+
"skip_head_transform": true,
|
| 53 |
+
"sparse_prediction": 0.25,
|
| 54 |
+
"tie_weights": true,
|
| 55 |
+
"use_bias": false
|
| 56 |
+
},
|
| 57 |
+
"architectures": [
|
| 58 |
+
"ScriptableLMForPreTraining"
|
| 59 |
+
],
|
| 60 |
+
"model_type": "crammedBERT",
|
| 61 |
+
"torch_dtype": "float32",
|
| 62 |
+
"transformers_version": "4.27.1"
|
| 63 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:586fb0fd11667b20552754fbaa30236eb7a64ba6bcdbfdb5db0d3132f68fafe5
|
| 3 |
+
size 579039260
|