Update config.json
Browse files- config.json +9 -4
config.json
CHANGED
@@ -3,7 +3,6 @@
|
|
3 |
"ReformerModelWithLMHead"
|
4 |
],
|
5 |
"attention_head_size": 64,
|
6 |
-
"attention_probs_dropout_prob": 0.1,
|
7 |
"attn_layers": [
|
8 |
"local",
|
9 |
"lsh",
|
@@ -24,15 +23,23 @@
|
|
24 |
],
|
25 |
"chunk_size_feed_forward": 0,
|
26 |
"chunk_size_lm_head": 0,
|
|
|
27 |
"feed_forward_size": 512,
|
|
|
28 |
"hidden_act": "relu",
|
29 |
-
"hidden_dropout_prob": 0.
|
30 |
"hidden_size": 256,
|
31 |
"initializer_range": 0.02,
|
32 |
"is_decoder": true,
|
33 |
"layer_norm_eps": 1e-12,
|
|
|
34 |
"local_attn_chunk_length": 64,
|
|
|
|
|
|
|
35 |
"lsh_attn_chunk_length": 64,
|
|
|
|
|
36 |
"max_position_embeddings": 524288,
|
37 |
"model_type": "reformer",
|
38 |
"num_attention_heads": 2,
|
@@ -40,8 +47,6 @@
|
|
40 |
64,
|
41 |
128
|
42 |
],
|
43 |
-
"num_chunks_after": 0,
|
44 |
-
"num_chunks_before": 1,
|
45 |
"num_hashes": 1,
|
46 |
"num_hidden_layers": 6,
|
47 |
"pad_token_id": 0,
|
|
|
3 |
"ReformerModelWithLMHead"
|
4 |
],
|
5 |
"attention_head_size": 64,
|
|
|
6 |
"attn_layers": [
|
7 |
"local",
|
8 |
"lsh",
|
|
|
23 |
],
|
24 |
"chunk_size_feed_forward": 0,
|
25 |
"chunk_size_lm_head": 0,
|
26 |
+
"eos_token_id": 2,
|
27 |
"feed_forward_size": 512,
|
28 |
+
"hash_seed": null,
|
29 |
"hidden_act": "relu",
|
30 |
+
"hidden_dropout_prob": 0.0,
|
31 |
"hidden_size": 256,
|
32 |
"initializer_range": 0.02,
|
33 |
"is_decoder": true,
|
34 |
"layer_norm_eps": 1e-12,
|
35 |
+
"local_attention_probs_dropout_prob": 0.0,
|
36 |
"local_attn_chunk_length": 64,
|
37 |
+
"local_num_chunks_after": 0,
|
38 |
+
"local_num_chunks_before": 1,
|
39 |
+
"lsh_attention_probs_dropout_prob": 0.0,
|
40 |
"lsh_attn_chunk_length": 64,
|
41 |
+
"lsh_num_chunks_after": 0,
|
42 |
+
"lsh_num_chunks_before": 1,
|
43 |
"max_position_embeddings": 524288,
|
44 |
"model_type": "reformer",
|
45 |
"num_attention_heads": 2,
|
|
|
47 |
64,
|
48 |
128
|
49 |
],
|
|
|
|
|
50 |
"num_hashes": 1,
|
51 |
"num_hidden_layers": 6,
|
52 |
"pad_token_id": 0,
|