Hannibal046 commited on
Commit
9e44dbb
·
verified ·
1 Parent(s): 71cecfd

upload dflash_qwen3_8b_block7 config

Browse files
Files changed (1) hide show
  1. config.json +53 -0
config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3DSparkModel"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "block_size": 7,
8
+ "bos_token_id": 151643,
9
+ "dtype": "bfloat16",
10
+ "enable_confidence_head": false,
11
+ "eos_token_id": 151645,
12
+ "head_dim": 128,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 4096,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 12288,
17
+ "layer_types": [
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention"
23
+ ],
24
+ "markov_rank": 0,
25
+ "mask_token_id": 151669,
26
+ "max_position_embeddings": 40960,
27
+ "max_window_layers": 36,
28
+ "model_type": "qwen3",
29
+ "num_anchors": 512,
30
+ "num_attention_heads": 32,
31
+ "num_hidden_layers": 5,
32
+ "num_key_value_heads": 8,
33
+ "num_target_layers": 36,
34
+ "pad_token_id": null,
35
+ "rms_norm_eps": 1e-06,
36
+ "rope_parameters": {
37
+ "rope_theta": 1000000,
38
+ "rope_type": "default"
39
+ },
40
+ "sliding_window": null,
41
+ "target_layer_ids": [
42
+ 1,
43
+ 9,
44
+ 17,
45
+ 25,
46
+ 33
47
+ ],
48
+ "tie_word_embeddings": false,
49
+ "transformers_version": "5.10.2",
50
+ "use_cache": true,
51
+ "use_sliding_window": false,
52
+ "vocab_size": 151936
53
+ }