Update config.json
Browse files- config.json +10 -0
config.json
CHANGED
@@ -16,6 +16,10 @@
|
|
16 |
"intermediate_size": 14336,
|
17 |
"max_position_embeddings": 32768,
|
18 |
"max_sequence_length": 131072,
|
|
|
|
|
|
|
|
|
19 |
"model_type": "mistral",
|
20 |
"num_attention_heads": 32,
|
21 |
"num_hidden_layers": 32,
|
@@ -34,5 +38,11 @@
|
|
34 |
"transformers_version": "4.38.2",
|
35 |
"unsloth_version": "2024.3",
|
36 |
"use_cache": true,
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
"vocab_size": 32002
|
38 |
}
|
|
|
16 |
"intermediate_size": 14336,
|
17 |
"max_position_embeddings": 32768,
|
18 |
"max_sequence_length": 131072,
|
19 |
+
"max_thoughts": 10,
|
20 |
+
"merged_lm_and_talk_heads": false,
|
21 |
+
"merged_lm_and_think_heads": true,
|
22 |
+
"merged_talk_heads": true,
|
23 |
"model_type": "mistral",
|
24 |
"num_attention_heads": 32,
|
25 |
"num_hidden_layers": 32,
|
|
|
38 |
"transformers_version": "4.38.2",
|
39 |
"unsloth_version": "2024.3",
|
40 |
"use_cache": true,
|
41 |
+
"use_complex_talk_head": true,
|
42 |
+
"use_complex_think_head": false,
|
43 |
+
"use_concat_talk_head": true,
|
44 |
+
"use_shallow_talk": false,
|
45 |
+
"use_shallow_think": true,
|
46 |
+
"use_weighted_talk_head": true,
|
47 |
"vocab_size": 32002
|
48 |
}
|