Linux User commited on
Commit
1520dab
·
1 Parent(s): d07a65c

add bf16 eagle

Browse files
MiniCPM4_1-8B-Eagle3-bf16/config.json ADDED
@@ -0,0 +1,162 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLMEagle3"
4
+ ],
5
+ "bos_token_id": 1,
6
+ "eos_token_id": [2,73440],
7
+ "hidden_act": "silu",
8
+ "hidden_size": 4096,
9
+ "initializer_range": 0.02,
10
+ "intermediate_size": 16384,
11
+ "max_position_embeddings": 65536,
12
+ "model_type": "llama",
13
+ "num_attention_heads": 32,
14
+ "num_key_value_heads": 2,
15
+ "num_hidden_layers": 1,
16
+ "pad_token_id": 2,
17
+ "rms_norm_eps": 1e-06,
18
+ "tie_word_embeddings": false,
19
+ "torch_dtype": "bfloat16",
20
+ "transformers_version": "4.28.1",
21
+ "use_cache": true,
22
+ "vocab_size": 73448,
23
+ "draft_vocab_size": 32000,
24
+ "rope_scaling": {
25
+ "factor": 1.0,
26
+ "long_factor": [
27
+ 0.9982316082870437,
28
+ 1.033048153422584,
29
+ 1.0749920956484724,
30
+ 1.1255096879436193,
31
+ 1.1863348602111476,
32
+ 1.259543828902579,
33
+ 1.3476188888731149,
34
+ 1.4535223827776373,
35
+ 1.5807816745852985,
36
+ 1.7335856049489526,
37
+ 1.9168922912975785,
38
+ 2.1365471404135326,
39
+ 2.3994084200118646,
40
+ 2.713475511863602,
41
+ 3.0880118452194134,
42
+ 3.533650295140154,
43
+ 4.062463396503134,
44
+ 4.687974098908333,
45
+ 5.425075306704039,
46
+ 6.289818967956352,
47
+ 7.29902962722721,
48
+ 8.469695779093664,
49
+ 9.81809877306655,
50
+ 11.358657902065282,
51
+ 13.102505860712087,
52
+ 15.055862949967128,
53
+ 17.218348131364184,
54
+ 19.581439255386453,
55
+ 22.127353314656723,
56
+ 24.828633849376587,
57
+ 27.6486820771775,
58
+ 30.54334096108829,
59
+ 33.46345345363812,
60
+ 36.358112337548896,
61
+ 39.17816056534983,
62
+ 41.879441100069684,
63
+ 44.425355159339965,
64
+ 46.78844628336223,
65
+ 48.95093146475928,
66
+ 50.90428855401433,
67
+ 52.648136512661125,
68
+ 54.18869564165987,
69
+ 55.537098635632745,
70
+ 56.7077647874992,
71
+ 57.71697544677006,
72
+ 58.58171910802236,
73
+ 59.31882031581807,
74
+ 59.94433101822328,
75
+ 60.47314411958625,
76
+ 60.918782569507,
77
+ 61.29331890286281,
78
+ 61.60738599471455,
79
+ 61.87024727431288,
80
+ 62.089902123428836,
81
+ 62.27320880977746,
82
+ 62.42601274014111,
83
+ 62.55327203194878,
84
+ 62.65917552585329,
85
+ 62.74725058582382,
86
+ 62.82045955451526,
87
+ 62.88128472678279,
88
+ 62.931802319077946,
89
+ 62.97374626130382,
90
+ 63.008562806439365
91
+ ],
92
+ "original_max_position_embeddings": 65536,
93
+ "rope_type": "longrope",
94
+ "short_factor": [
95
+ 0.9982316082870437,
96
+ 1.033048153422584,
97
+ 1.0749920956484724,
98
+ 1.1255096879436193,
99
+ 1.1863348602111476,
100
+ 1.259543828902579,
101
+ 1.3476188888731149,
102
+ 1.4535223827776373,
103
+ 1.5807816745852985,
104
+ 1.7335856049489526,
105
+ 1.9168922912975785,
106
+ 2.1365471404135326,
107
+ 2.3994084200118646,
108
+ 2.713475511863602,
109
+ 3.0880118452194134,
110
+ 3.533650295140154,
111
+ 4.062463396503134,
112
+ 4.687974098908333,
113
+ 5.425075306704039,
114
+ 6.289818967956352,
115
+ 7.29902962722721,
116
+ 8.469695779093664,
117
+ 9.81809877306655,
118
+ 11.358657902065282,
119
+ 13.102505860712087,
120
+ 15.055862949967128,
121
+ 17.218348131364184,
122
+ 19.581439255386453,
123
+ 22.127353314656723,
124
+ 24.828633849376587,
125
+ 27.6486820771775,
126
+ 30.54334096108829,
127
+ 33.46345345363812,
128
+ 36.358112337548896,
129
+ 39.17816056534983,
130
+ 41.879441100069684,
131
+ 44.425355159339965,
132
+ 46.78844628336223,
133
+ 48.95093146475928,
134
+ 50.90428855401433,
135
+ 52.648136512661125,
136
+ 54.18869564165987,
137
+ 55.537098635632745,
138
+ 56.7077647874992,
139
+ 57.71697544677006,
140
+ 58.58171910802236,
141
+ 59.31882031581807,
142
+ 59.94433101822328,
143
+ 60.47314411958625,
144
+ 60.918782569507,
145
+ 61.29331890286281,
146
+ 61.60738599471455,
147
+ 61.87024727431288,
148
+ 62.089902123428836,
149
+ 62.27320880977746,
150
+ 62.42601274014111,
151
+ 62.55327203194878,
152
+ 62.65917552585329,
153
+ 62.74725058582382,
154
+ 62.82045955451526,
155
+ 62.88128472678279,
156
+ 62.931802319077946,
157
+ 62.97374626130382,
158
+ 63.008562806439365
159
+ ],
160
+ "type": "longrope"
161
+ }
162
+ }
MiniCPM4_1-8B-Eagle3-bf16/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01ad40f59f37610d12f4b7402063f373fb28e6ad99e1ec7afbe33c08209e5e77
3
+ size 874879771