mgoin commited on
Commit
7f7543b
·
verified ·
1 Parent(s): 9f418e9

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +83 -0
config.json ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "DSparkDraftModel"
4
+ ],
5
+ "auto_map": {
6
+ "": "config.DSparkSpeculatorConfig"
7
+ },
8
+ "aux_hidden_state_layer_ids": [
9
+ 8,
10
+ 23,
11
+ 39,
12
+ 55,
13
+ 70
14
+ ],
15
+ "block_size": 8,
16
+ "confidence_head_with_markov": true,
17
+ "draft_vocab_size": 154880,
18
+ "dtype": "bfloat16",
19
+ "enable_confidence_head": true,
20
+ "markov_head_type": "vanilla",
21
+ "markov_rank": 256,
22
+ "mask_token_id": 154856,
23
+ "max_anchors": 1024,
24
+ "sliding_window_non_causal": false,
25
+ "speculators_config": {
26
+ "algorithm": "dspark",
27
+ "default_proposal_method": "greedy",
28
+ "proposal_methods": [
29
+ {
30
+ "accept_tolerance": 0.0,
31
+ "proposal_type": "greedy",
32
+ "speculative_tokens": 7,
33
+ "verifier_accept_k": 1
34
+ }
35
+ ],
36
+ "verifier": {
37
+ "architectures": [
38
+ "GlmMoeDsaForCausalLM"
39
+ ],
40
+ "name_or_path": "zai-org/GLM-5.2-FP8"
41
+ }
42
+ },
43
+ "speculators_model_type": "dspark",
44
+ "speculators_version": "0.5.0.dev38",
45
+ "target_hidden_size": null,
46
+ "tie_word_embeddings": false,
47
+ "transformer_layer_config": {
48
+ "attention_bias": false,
49
+ "attention_dropout": 0.0,
50
+ "bos_token_id": null,
51
+ "eos_token_id": null,
52
+ "head_dim": 64,
53
+ "hidden_act": "silu",
54
+ "hidden_size": 6144,
55
+ "initializer_range": 0.02,
56
+ "intermediate_size": 12288,
57
+ "layer_types": [
58
+ "full_attention",
59
+ "full_attention",
60
+ "full_attention",
61
+ "full_attention",
62
+ "full_attention"
63
+ ],
64
+ "max_position_embeddings": 1048576,
65
+ "max_window_layers": 28,
66
+ "model_type": "qwen3",
67
+ "num_attention_heads": 64,
68
+ "num_hidden_layers": 5,
69
+ "num_key_value_heads": 64,
70
+ "pad_token_id": null,
71
+ "rms_norm_eps": 1e-05,
72
+ "rope_parameters": {
73
+ "rope_theta": 8000000,
74
+ "rope_type": "default"
75
+ },
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": false,
78
+ "use_cache": true,
79
+ "use_sliding_window": false,
80
+ "vocab_size": 154880
81
+ },
82
+ "transformers_version": "5.12.1"
83
+ }