Commit
·
c6d90f3
1
Parent(s):
9e948ba
messing around with differing settings.
Browse files- config.json +3 -4
config.json
CHANGED
@@ -16,13 +16,11 @@
|
|
16 |
"AutoModelForSequenceClassification": "modeling_lsg_bert.LSGBertForSequenceClassification",
|
17 |
"AutoModelForTokenClassification": "modeling_lsg_bert.LSGBertForTokenClassification"
|
18 |
},
|
|
|
19 |
"block_size": 128,
|
|
|
20 |
"sparse_block_size": 128,
|
21 |
"sparsity_factor": 2,
|
22 |
-
"base_model_prefix": "lsg",
|
23 |
-
"sparsity_type": "norm",
|
24 |
-
"is_decoder": false,
|
25 |
-
"pool_with_global": true,
|
26 |
"num_global_tokens": 1,
|
27 |
"classifier_dropout": null,
|
28 |
"hidden_act": "gelu",
|
@@ -39,6 +37,7 @@
|
|
39 |
"num_hidden_layers": 24,
|
40 |
"pad_token_id": 0,
|
41 |
"position_embedding_type": "absolute",
|
|
|
42 |
"torch_dtype": "float32",
|
43 |
"transformers_version": "4.30.2",
|
44 |
"type_vocab_size": 2,
|
|
|
16 |
"AutoModelForSequenceClassification": "modeling_lsg_bert.LSGBertForSequenceClassification",
|
17 |
"AutoModelForTokenClassification": "modeling_lsg_bert.LSGBertForTokenClassification"
|
18 |
},
|
19 |
+
"base_model_prefix": "lsg",
|
20 |
"block_size": 128,
|
21 |
+
"pool_with_global": true,
|
22 |
"sparse_block_size": 128,
|
23 |
"sparsity_factor": 2,
|
|
|
|
|
|
|
|
|
24 |
"num_global_tokens": 1,
|
25 |
"classifier_dropout": null,
|
26 |
"hidden_act": "gelu",
|
|
|
37 |
"num_hidden_layers": 24,
|
38 |
"pad_token_id": 0,
|
39 |
"position_embedding_type": "absolute",
|
40 |
+
"sparsity_type": "norm",
|
41 |
"torch_dtype": "float32",
|
42 |
"transformers_version": "4.30.2",
|
43 |
"type_vocab_size": 2,
|