Text Generation
Transformers
PyTorch
Safetensors
English
stripedhyena
custom_code
EugeneLYC commited on
Commit
2ab6c7b
1 Parent(s): 230c4b6

add model safetensors

Browse files
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0f361101ba30a8ba049969b8e8eed344c87e337b05eabe3bbed74d8ae266b4d
3
+ size 9888607072
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f46597fe4a1b2a774a983af7c3adb3527797ba57cd0df1f5c03c5045c7623554
3
+ size 5403482216
model.safetensors.index.json ADDED
@@ -0,0 +1,362 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 15293097984
4
+ },
5
+ "weight_map": {
6
+ "backbone.blocks.0.filter.D": "model-00001-of-00002.safetensors",
7
+ "backbone.blocks.0.filter.poles": "model-00001-of-00002.safetensors",
8
+ "backbone.blocks.0.filter.residues": "model-00001-of-00002.safetensors",
9
+ "backbone.blocks.0.filter.short_filter_bias": "model-00001-of-00002.safetensors",
10
+ "backbone.blocks.0.filter.short_filter_weight": "model-00001-of-00002.safetensors",
11
+ "backbone.blocks.0.mlp.l1.weight": "model-00001-of-00002.safetensors",
12
+ "backbone.blocks.0.mlp.l2.weight": "model-00001-of-00002.safetensors",
13
+ "backbone.blocks.0.mlp.l3.weight": "model-00001-of-00002.safetensors",
14
+ "backbone.blocks.0.out_filter_dense.bias": "model-00001-of-00002.safetensors",
15
+ "backbone.blocks.0.out_filter_dense.weight": "model-00001-of-00002.safetensors",
16
+ "backbone.blocks.0.post_norm.scale": "model-00001-of-00002.safetensors",
17
+ "backbone.blocks.0.pre_norm.scale": "model-00001-of-00002.safetensors",
18
+ "backbone.blocks.0.projections.bias": "model-00001-of-00002.safetensors",
19
+ "backbone.blocks.0.projections.weight": "model-00001-of-00002.safetensors",
20
+ "backbone.blocks.1.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
21
+ "backbone.blocks.1.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
22
+ "backbone.blocks.1.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
23
+ "backbone.blocks.1.mlp.l1.weight": "model-00001-of-00002.safetensors",
24
+ "backbone.blocks.1.mlp.l2.weight": "model-00001-of-00002.safetensors",
25
+ "backbone.blocks.1.mlp.l3.weight": "model-00001-of-00002.safetensors",
26
+ "backbone.blocks.1.post_norm.scale": "model-00001-of-00002.safetensors",
27
+ "backbone.blocks.1.pre_norm.scale": "model-00001-of-00002.safetensors",
28
+ "backbone.blocks.10.filter.D": "model-00001-of-00002.safetensors",
29
+ "backbone.blocks.10.filter.poles": "model-00001-of-00002.safetensors",
30
+ "backbone.blocks.10.filter.residues": "model-00001-of-00002.safetensors",
31
+ "backbone.blocks.10.filter.short_filter_bias": "model-00001-of-00002.safetensors",
32
+ "backbone.blocks.10.filter.short_filter_weight": "model-00001-of-00002.safetensors",
33
+ "backbone.blocks.10.mlp.l1.weight": "model-00001-of-00002.safetensors",
34
+ "backbone.blocks.10.mlp.l2.weight": "model-00001-of-00002.safetensors",
35
+ "backbone.blocks.10.mlp.l3.weight": "model-00001-of-00002.safetensors",
36
+ "backbone.blocks.10.out_filter_dense.bias": "model-00001-of-00002.safetensors",
37
+ "backbone.blocks.10.out_filter_dense.weight": "model-00001-of-00002.safetensors",
38
+ "backbone.blocks.10.post_norm.scale": "model-00001-of-00002.safetensors",
39
+ "backbone.blocks.10.pre_norm.scale": "model-00001-of-00002.safetensors",
40
+ "backbone.blocks.10.projections.bias": "model-00001-of-00002.safetensors",
41
+ "backbone.blocks.10.projections.weight": "model-00001-of-00002.safetensors",
42
+ "backbone.blocks.11.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
43
+ "backbone.blocks.11.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
44
+ "backbone.blocks.11.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
45
+ "backbone.blocks.11.mlp.l1.weight": "model-00001-of-00002.safetensors",
46
+ "backbone.blocks.11.mlp.l2.weight": "model-00001-of-00002.safetensors",
47
+ "backbone.blocks.11.mlp.l3.weight": "model-00001-of-00002.safetensors",
48
+ "backbone.blocks.11.post_norm.scale": "model-00001-of-00002.safetensors",
49
+ "backbone.blocks.11.pre_norm.scale": "model-00001-of-00002.safetensors",
50
+ "backbone.blocks.12.filter.D": "model-00001-of-00002.safetensors",
51
+ "backbone.blocks.12.filter.poles": "model-00001-of-00002.safetensors",
52
+ "backbone.blocks.12.filter.residues": "model-00001-of-00002.safetensors",
53
+ "backbone.blocks.12.filter.short_filter_bias": "model-00001-of-00002.safetensors",
54
+ "backbone.blocks.12.filter.short_filter_weight": "model-00001-of-00002.safetensors",
55
+ "backbone.blocks.12.mlp.l1.weight": "model-00001-of-00002.safetensors",
56
+ "backbone.blocks.12.mlp.l2.weight": "model-00001-of-00002.safetensors",
57
+ "backbone.blocks.12.mlp.l3.weight": "model-00001-of-00002.safetensors",
58
+ "backbone.blocks.12.out_filter_dense.bias": "model-00001-of-00002.safetensors",
59
+ "backbone.blocks.12.out_filter_dense.weight": "model-00001-of-00002.safetensors",
60
+ "backbone.blocks.12.post_norm.scale": "model-00001-of-00002.safetensors",
61
+ "backbone.blocks.12.pre_norm.scale": "model-00001-of-00002.safetensors",
62
+ "backbone.blocks.12.projections.bias": "model-00001-of-00002.safetensors",
63
+ "backbone.blocks.12.projections.weight": "model-00001-of-00002.safetensors",
64
+ "backbone.blocks.13.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
65
+ "backbone.blocks.13.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
66
+ "backbone.blocks.13.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
67
+ "backbone.blocks.13.mlp.l1.weight": "model-00001-of-00002.safetensors",
68
+ "backbone.blocks.13.mlp.l2.weight": "model-00001-of-00002.safetensors",
69
+ "backbone.blocks.13.mlp.l3.weight": "model-00001-of-00002.safetensors",
70
+ "backbone.blocks.13.post_norm.scale": "model-00001-of-00002.safetensors",
71
+ "backbone.blocks.13.pre_norm.scale": "model-00001-of-00002.safetensors",
72
+ "backbone.blocks.14.filter.D": "model-00001-of-00002.safetensors",
73
+ "backbone.blocks.14.filter.poles": "model-00001-of-00002.safetensors",
74
+ "backbone.blocks.14.filter.residues": "model-00001-of-00002.safetensors",
75
+ "backbone.blocks.14.filter.short_filter_bias": "model-00001-of-00002.safetensors",
76
+ "backbone.blocks.14.filter.short_filter_weight": "model-00001-of-00002.safetensors",
77
+ "backbone.blocks.14.mlp.l1.weight": "model-00001-of-00002.safetensors",
78
+ "backbone.blocks.14.mlp.l2.weight": "model-00001-of-00002.safetensors",
79
+ "backbone.blocks.14.mlp.l3.weight": "model-00001-of-00002.safetensors",
80
+ "backbone.blocks.14.out_filter_dense.bias": "model-00001-of-00002.safetensors",
81
+ "backbone.blocks.14.out_filter_dense.weight": "model-00001-of-00002.safetensors",
82
+ "backbone.blocks.14.post_norm.scale": "model-00001-of-00002.safetensors",
83
+ "backbone.blocks.14.pre_norm.scale": "model-00001-of-00002.safetensors",
84
+ "backbone.blocks.14.projections.bias": "model-00001-of-00002.safetensors",
85
+ "backbone.blocks.14.projections.weight": "model-00001-of-00002.safetensors",
86
+ "backbone.blocks.15.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
87
+ "backbone.blocks.15.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
88
+ "backbone.blocks.15.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
89
+ "backbone.blocks.15.mlp.l1.weight": "model-00001-of-00002.safetensors",
90
+ "backbone.blocks.15.mlp.l2.weight": "model-00001-of-00002.safetensors",
91
+ "backbone.blocks.15.mlp.l3.weight": "model-00001-of-00002.safetensors",
92
+ "backbone.blocks.15.post_norm.scale": "model-00001-of-00002.safetensors",
93
+ "backbone.blocks.15.pre_norm.scale": "model-00001-of-00002.safetensors",
94
+ "backbone.blocks.16.filter.D": "model-00001-of-00002.safetensors",
95
+ "backbone.blocks.16.filter.poles": "model-00001-of-00002.safetensors",
96
+ "backbone.blocks.16.filter.residues": "model-00001-of-00002.safetensors",
97
+ "backbone.blocks.16.filter.short_filter_bias": "model-00001-of-00002.safetensors",
98
+ "backbone.blocks.16.filter.short_filter_weight": "model-00001-of-00002.safetensors",
99
+ "backbone.blocks.16.mlp.l1.weight": "model-00001-of-00002.safetensors",
100
+ "backbone.blocks.16.mlp.l2.weight": "model-00001-of-00002.safetensors",
101
+ "backbone.blocks.16.mlp.l3.weight": "model-00001-of-00002.safetensors",
102
+ "backbone.blocks.16.out_filter_dense.bias": "model-00001-of-00002.safetensors",
103
+ "backbone.blocks.16.out_filter_dense.weight": "model-00001-of-00002.safetensors",
104
+ "backbone.blocks.16.post_norm.scale": "model-00001-of-00002.safetensors",
105
+ "backbone.blocks.16.pre_norm.scale": "model-00001-of-00002.safetensors",
106
+ "backbone.blocks.16.projections.bias": "model-00001-of-00002.safetensors",
107
+ "backbone.blocks.16.projections.weight": "model-00001-of-00002.safetensors",
108
+ "backbone.blocks.17.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
109
+ "backbone.blocks.17.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
110
+ "backbone.blocks.17.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
111
+ "backbone.blocks.17.mlp.l1.weight": "model-00001-of-00002.safetensors",
112
+ "backbone.blocks.17.mlp.l2.weight": "model-00001-of-00002.safetensors",
113
+ "backbone.blocks.17.mlp.l3.weight": "model-00001-of-00002.safetensors",
114
+ "backbone.blocks.17.post_norm.scale": "model-00001-of-00002.safetensors",
115
+ "backbone.blocks.17.pre_norm.scale": "model-00001-of-00002.safetensors",
116
+ "backbone.blocks.18.filter.D": "model-00001-of-00002.safetensors",
117
+ "backbone.blocks.18.filter.poles": "model-00001-of-00002.safetensors",
118
+ "backbone.blocks.18.filter.residues": "model-00001-of-00002.safetensors",
119
+ "backbone.blocks.18.filter.short_filter_bias": "model-00001-of-00002.safetensors",
120
+ "backbone.blocks.18.filter.short_filter_weight": "model-00001-of-00002.safetensors",
121
+ "backbone.blocks.18.mlp.l1.weight": "model-00001-of-00002.safetensors",
122
+ "backbone.blocks.18.mlp.l2.weight": "model-00001-of-00002.safetensors",
123
+ "backbone.blocks.18.mlp.l3.weight": "model-00001-of-00002.safetensors",
124
+ "backbone.blocks.18.out_filter_dense.bias": "model-00001-of-00002.safetensors",
125
+ "backbone.blocks.18.out_filter_dense.weight": "model-00001-of-00002.safetensors",
126
+ "backbone.blocks.18.post_norm.scale": "model-00001-of-00002.safetensors",
127
+ "backbone.blocks.18.pre_norm.scale": "model-00001-of-00002.safetensors",
128
+ "backbone.blocks.18.projections.bias": "model-00001-of-00002.safetensors",
129
+ "backbone.blocks.18.projections.weight": "model-00001-of-00002.safetensors",
130
+ "backbone.blocks.19.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
131
+ "backbone.blocks.19.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
132
+ "backbone.blocks.19.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
133
+ "backbone.blocks.19.mlp.l1.weight": "model-00001-of-00002.safetensors",
134
+ "backbone.blocks.19.mlp.l2.weight": "model-00001-of-00002.safetensors",
135
+ "backbone.blocks.19.mlp.l3.weight": "model-00001-of-00002.safetensors",
136
+ "backbone.blocks.19.post_norm.scale": "model-00001-of-00002.safetensors",
137
+ "backbone.blocks.19.pre_norm.scale": "model-00001-of-00002.safetensors",
138
+ "backbone.blocks.2.filter.D": "model-00001-of-00002.safetensors",
139
+ "backbone.blocks.2.filter.poles": "model-00001-of-00002.safetensors",
140
+ "backbone.blocks.2.filter.residues": "model-00001-of-00002.safetensors",
141
+ "backbone.blocks.2.filter.short_filter_bias": "model-00001-of-00002.safetensors",
142
+ "backbone.blocks.2.filter.short_filter_weight": "model-00001-of-00002.safetensors",
143
+ "backbone.blocks.2.mlp.l1.weight": "model-00001-of-00002.safetensors",
144
+ "backbone.blocks.2.mlp.l2.weight": "model-00001-of-00002.safetensors",
145
+ "backbone.blocks.2.mlp.l3.weight": "model-00001-of-00002.safetensors",
146
+ "backbone.blocks.2.out_filter_dense.bias": "model-00001-of-00002.safetensors",
147
+ "backbone.blocks.2.out_filter_dense.weight": "model-00001-of-00002.safetensors",
148
+ "backbone.blocks.2.post_norm.scale": "model-00001-of-00002.safetensors",
149
+ "backbone.blocks.2.pre_norm.scale": "model-00001-of-00002.safetensors",
150
+ "backbone.blocks.2.projections.bias": "model-00001-of-00002.safetensors",
151
+ "backbone.blocks.2.projections.weight": "model-00001-of-00002.safetensors",
152
+ "backbone.blocks.20.filter.D": "model-00001-of-00002.safetensors",
153
+ "backbone.blocks.20.filter.poles": "model-00001-of-00002.safetensors",
154
+ "backbone.blocks.20.filter.residues": "model-00001-of-00002.safetensors",
155
+ "backbone.blocks.20.filter.short_filter_bias": "model-00001-of-00002.safetensors",
156
+ "backbone.blocks.20.filter.short_filter_weight": "model-00001-of-00002.safetensors",
157
+ "backbone.blocks.20.mlp.l1.weight": "model-00002-of-00002.safetensors",
158
+ "backbone.blocks.20.mlp.l2.weight": "model-00002-of-00002.safetensors",
159
+ "backbone.blocks.20.mlp.l3.weight": "model-00002-of-00002.safetensors",
160
+ "backbone.blocks.20.out_filter_dense.bias": "model-00001-of-00002.safetensors",
161
+ "backbone.blocks.20.out_filter_dense.weight": "model-00001-of-00002.safetensors",
162
+ "backbone.blocks.20.post_norm.scale": "model-00001-of-00002.safetensors",
163
+ "backbone.blocks.20.pre_norm.scale": "model-00001-of-00002.safetensors",
164
+ "backbone.blocks.20.projections.bias": "model-00001-of-00002.safetensors",
165
+ "backbone.blocks.20.projections.weight": "model-00001-of-00002.safetensors",
166
+ "backbone.blocks.21.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
167
+ "backbone.blocks.21.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
168
+ "backbone.blocks.21.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
169
+ "backbone.blocks.21.mlp.l1.weight": "model-00002-of-00002.safetensors",
170
+ "backbone.blocks.21.mlp.l2.weight": "model-00002-of-00002.safetensors",
171
+ "backbone.blocks.21.mlp.l3.weight": "model-00002-of-00002.safetensors",
172
+ "backbone.blocks.21.post_norm.scale": "model-00002-of-00002.safetensors",
173
+ "backbone.blocks.21.pre_norm.scale": "model-00002-of-00002.safetensors",
174
+ "backbone.blocks.22.filter.D": "model-00002-of-00002.safetensors",
175
+ "backbone.blocks.22.filter.poles": "model-00002-of-00002.safetensors",
176
+ "backbone.blocks.22.filter.residues": "model-00002-of-00002.safetensors",
177
+ "backbone.blocks.22.filter.short_filter_bias": "model-00002-of-00002.safetensors",
178
+ "backbone.blocks.22.filter.short_filter_weight": "model-00002-of-00002.safetensors",
179
+ "backbone.blocks.22.mlp.l1.weight": "model-00002-of-00002.safetensors",
180
+ "backbone.blocks.22.mlp.l2.weight": "model-00002-of-00002.safetensors",
181
+ "backbone.blocks.22.mlp.l3.weight": "model-00002-of-00002.safetensors",
182
+ "backbone.blocks.22.out_filter_dense.bias": "model-00002-of-00002.safetensors",
183
+ "backbone.blocks.22.out_filter_dense.weight": "model-00002-of-00002.safetensors",
184
+ "backbone.blocks.22.post_norm.scale": "model-00002-of-00002.safetensors",
185
+ "backbone.blocks.22.pre_norm.scale": "model-00002-of-00002.safetensors",
186
+ "backbone.blocks.22.projections.bias": "model-00002-of-00002.safetensors",
187
+ "backbone.blocks.22.projections.weight": "model-00002-of-00002.safetensors",
188
+ "backbone.blocks.23.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
189
+ "backbone.blocks.23.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
190
+ "backbone.blocks.23.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
191
+ "backbone.blocks.23.mlp.l1.weight": "model-00002-of-00002.safetensors",
192
+ "backbone.blocks.23.mlp.l2.weight": "model-00002-of-00002.safetensors",
193
+ "backbone.blocks.23.mlp.l3.weight": "model-00002-of-00002.safetensors",
194
+ "backbone.blocks.23.post_norm.scale": "model-00002-of-00002.safetensors",
195
+ "backbone.blocks.23.pre_norm.scale": "model-00002-of-00002.safetensors",
196
+ "backbone.blocks.24.filter.D": "model-00002-of-00002.safetensors",
197
+ "backbone.blocks.24.filter.poles": "model-00002-of-00002.safetensors",
198
+ "backbone.blocks.24.filter.residues": "model-00002-of-00002.safetensors",
199
+ "backbone.blocks.24.filter.short_filter_bias": "model-00002-of-00002.safetensors",
200
+ "backbone.blocks.24.filter.short_filter_weight": "model-00002-of-00002.safetensors",
201
+ "backbone.blocks.24.mlp.l1.weight": "model-00002-of-00002.safetensors",
202
+ "backbone.blocks.24.mlp.l2.weight": "model-00002-of-00002.safetensors",
203
+ "backbone.blocks.24.mlp.l3.weight": "model-00002-of-00002.safetensors",
204
+ "backbone.blocks.24.out_filter_dense.bias": "model-00002-of-00002.safetensors",
205
+ "backbone.blocks.24.out_filter_dense.weight": "model-00002-of-00002.safetensors",
206
+ "backbone.blocks.24.post_norm.scale": "model-00002-of-00002.safetensors",
207
+ "backbone.blocks.24.pre_norm.scale": "model-00002-of-00002.safetensors",
208
+ "backbone.blocks.24.projections.bias": "model-00002-of-00002.safetensors",
209
+ "backbone.blocks.24.projections.weight": "model-00002-of-00002.safetensors",
210
+ "backbone.blocks.25.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
211
+ "backbone.blocks.25.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
212
+ "backbone.blocks.25.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
213
+ "backbone.blocks.25.mlp.l1.weight": "model-00002-of-00002.safetensors",
214
+ "backbone.blocks.25.mlp.l2.weight": "model-00002-of-00002.safetensors",
215
+ "backbone.blocks.25.mlp.l3.weight": "model-00002-of-00002.safetensors",
216
+ "backbone.blocks.25.post_norm.scale": "model-00002-of-00002.safetensors",
217
+ "backbone.blocks.25.pre_norm.scale": "model-00002-of-00002.safetensors",
218
+ "backbone.blocks.26.filter.D": "model-00002-of-00002.safetensors",
219
+ "backbone.blocks.26.filter.poles": "model-00002-of-00002.safetensors",
220
+ "backbone.blocks.26.filter.residues": "model-00002-of-00002.safetensors",
221
+ "backbone.blocks.26.filter.short_filter_bias": "model-00002-of-00002.safetensors",
222
+ "backbone.blocks.26.filter.short_filter_weight": "model-00002-of-00002.safetensors",
223
+ "backbone.blocks.26.mlp.l1.weight": "model-00002-of-00002.safetensors",
224
+ "backbone.blocks.26.mlp.l2.weight": "model-00002-of-00002.safetensors",
225
+ "backbone.blocks.26.mlp.l3.weight": "model-00002-of-00002.safetensors",
226
+ "backbone.blocks.26.out_filter_dense.bias": "model-00002-of-00002.safetensors",
227
+ "backbone.blocks.26.out_filter_dense.weight": "model-00002-of-00002.safetensors",
228
+ "backbone.blocks.26.post_norm.scale": "model-00002-of-00002.safetensors",
229
+ "backbone.blocks.26.pre_norm.scale": "model-00002-of-00002.safetensors",
230
+ "backbone.blocks.26.projections.bias": "model-00002-of-00002.safetensors",
231
+ "backbone.blocks.26.projections.weight": "model-00002-of-00002.safetensors",
232
+ "backbone.blocks.27.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
233
+ "backbone.blocks.27.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
234
+ "backbone.blocks.27.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
235
+ "backbone.blocks.27.mlp.l1.weight": "model-00002-of-00002.safetensors",
236
+ "backbone.blocks.27.mlp.l2.weight": "model-00002-of-00002.safetensors",
237
+ "backbone.blocks.27.mlp.l3.weight": "model-00002-of-00002.safetensors",
238
+ "backbone.blocks.27.post_norm.scale": "model-00002-of-00002.safetensors",
239
+ "backbone.blocks.27.pre_norm.scale": "model-00002-of-00002.safetensors",
240
+ "backbone.blocks.28.filter.D": "model-00002-of-00002.safetensors",
241
+ "backbone.blocks.28.filter.poles": "model-00002-of-00002.safetensors",
242
+ "backbone.blocks.28.filter.residues": "model-00002-of-00002.safetensors",
243
+ "backbone.blocks.28.filter.short_filter_bias": "model-00002-of-00002.safetensors",
244
+ "backbone.blocks.28.filter.short_filter_weight": "model-00002-of-00002.safetensors",
245
+ "backbone.blocks.28.mlp.l1.weight": "model-00002-of-00002.safetensors",
246
+ "backbone.blocks.28.mlp.l2.weight": "model-00002-of-00002.safetensors",
247
+ "backbone.blocks.28.mlp.l3.weight": "model-00002-of-00002.safetensors",
248
+ "backbone.blocks.28.out_filter_dense.bias": "model-00002-of-00002.safetensors",
249
+ "backbone.blocks.28.out_filter_dense.weight": "model-00002-of-00002.safetensors",
250
+ "backbone.blocks.28.post_norm.scale": "model-00002-of-00002.safetensors",
251
+ "backbone.blocks.28.pre_norm.scale": "model-00002-of-00002.safetensors",
252
+ "backbone.blocks.28.projections.bias": "model-00002-of-00002.safetensors",
253
+ "backbone.blocks.28.projections.weight": "model-00002-of-00002.safetensors",
254
+ "backbone.blocks.29.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
255
+ "backbone.blocks.29.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
256
+ "backbone.blocks.29.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
257
+ "backbone.blocks.29.mlp.l1.weight": "model-00002-of-00002.safetensors",
258
+ "backbone.blocks.29.mlp.l2.weight": "model-00002-of-00002.safetensors",
259
+ "backbone.blocks.29.mlp.l3.weight": "model-00002-of-00002.safetensors",
260
+ "backbone.blocks.29.post_norm.scale": "model-00002-of-00002.safetensors",
261
+ "backbone.blocks.29.pre_norm.scale": "model-00002-of-00002.safetensors",
262
+ "backbone.blocks.3.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
263
+ "backbone.blocks.3.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
264
+ "backbone.blocks.3.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
265
+ "backbone.blocks.3.mlp.l1.weight": "model-00001-of-00002.safetensors",
266
+ "backbone.blocks.3.mlp.l2.weight": "model-00001-of-00002.safetensors",
267
+ "backbone.blocks.3.mlp.l3.weight": "model-00001-of-00002.safetensors",
268
+ "backbone.blocks.3.post_norm.scale": "model-00001-of-00002.safetensors",
269
+ "backbone.blocks.3.pre_norm.scale": "model-00001-of-00002.safetensors",
270
+ "backbone.blocks.30.filter.D": "model-00002-of-00002.safetensors",
271
+ "backbone.blocks.30.filter.poles": "model-00002-of-00002.safetensors",
272
+ "backbone.blocks.30.filter.residues": "model-00002-of-00002.safetensors",
273
+ "backbone.blocks.30.filter.short_filter_bias": "model-00002-of-00002.safetensors",
274
+ "backbone.blocks.30.filter.short_filter_weight": "model-00002-of-00002.safetensors",
275
+ "backbone.blocks.30.mlp.l1.weight": "model-00002-of-00002.safetensors",
276
+ "backbone.blocks.30.mlp.l2.weight": "model-00002-of-00002.safetensors",
277
+ "backbone.blocks.30.mlp.l3.weight": "model-00002-of-00002.safetensors",
278
+ "backbone.blocks.30.out_filter_dense.bias": "model-00002-of-00002.safetensors",
279
+ "backbone.blocks.30.out_filter_dense.weight": "model-00002-of-00002.safetensors",
280
+ "backbone.blocks.30.post_norm.scale": "model-00002-of-00002.safetensors",
281
+ "backbone.blocks.30.pre_norm.scale": "model-00002-of-00002.safetensors",
282
+ "backbone.blocks.30.projections.bias": "model-00002-of-00002.safetensors",
283
+ "backbone.blocks.30.projections.weight": "model-00002-of-00002.safetensors",
284
+ "backbone.blocks.31.inner_mha_cls.Wqkv.weight": "model-00002-of-00002.safetensors",
285
+ "backbone.blocks.31.inner_mha_cls.out_proj.weight": "model-00002-of-00002.safetensors",
286
+ "backbone.blocks.31.inner_mha_cls.rotary_emb.inv_freq": "model-00002-of-00002.safetensors",
287
+ "backbone.blocks.31.mlp.l1.weight": "model-00002-of-00002.safetensors",
288
+ "backbone.blocks.31.mlp.l2.weight": "model-00002-of-00002.safetensors",
289
+ "backbone.blocks.31.mlp.l3.weight": "model-00002-of-00002.safetensors",
290
+ "backbone.blocks.31.post_norm.scale": "model-00002-of-00002.safetensors",
291
+ "backbone.blocks.31.pre_norm.scale": "model-00002-of-00002.safetensors",
292
+ "backbone.blocks.4.filter.D": "model-00001-of-00002.safetensors",
293
+ "backbone.blocks.4.filter.poles": "model-00001-of-00002.safetensors",
294
+ "backbone.blocks.4.filter.residues": "model-00001-of-00002.safetensors",
295
+ "backbone.blocks.4.filter.short_filter_bias": "model-00001-of-00002.safetensors",
296
+ "backbone.blocks.4.filter.short_filter_weight": "model-00001-of-00002.safetensors",
297
+ "backbone.blocks.4.mlp.l1.weight": "model-00001-of-00002.safetensors",
298
+ "backbone.blocks.4.mlp.l2.weight": "model-00001-of-00002.safetensors",
299
+ "backbone.blocks.4.mlp.l3.weight": "model-00001-of-00002.safetensors",
300
+ "backbone.blocks.4.out_filter_dense.bias": "model-00001-of-00002.safetensors",
301
+ "backbone.blocks.4.out_filter_dense.weight": "model-00001-of-00002.safetensors",
302
+ "backbone.blocks.4.post_norm.scale": "model-00001-of-00002.safetensors",
303
+ "backbone.blocks.4.pre_norm.scale": "model-00001-of-00002.safetensors",
304
+ "backbone.blocks.4.projections.bias": "model-00001-of-00002.safetensors",
305
+ "backbone.blocks.4.projections.weight": "model-00001-of-00002.safetensors",
306
+ "backbone.blocks.5.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
307
+ "backbone.blocks.5.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
308
+ "backbone.blocks.5.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
309
+ "backbone.blocks.5.mlp.l1.weight": "model-00001-of-00002.safetensors",
310
+ "backbone.blocks.5.mlp.l2.weight": "model-00001-of-00002.safetensors",
311
+ "backbone.blocks.5.mlp.l3.weight": "model-00001-of-00002.safetensors",
312
+ "backbone.blocks.5.post_norm.scale": "model-00001-of-00002.safetensors",
313
+ "backbone.blocks.5.pre_norm.scale": "model-00001-of-00002.safetensors",
314
+ "backbone.blocks.6.filter.D": "model-00001-of-00002.safetensors",
315
+ "backbone.blocks.6.filter.poles": "model-00001-of-00002.safetensors",
316
+ "backbone.blocks.6.filter.residues": "model-00001-of-00002.safetensors",
317
+ "backbone.blocks.6.filter.short_filter_bias": "model-00001-of-00002.safetensors",
318
+ "backbone.blocks.6.filter.short_filter_weight": "model-00001-of-00002.safetensors",
319
+ "backbone.blocks.6.mlp.l1.weight": "model-00001-of-00002.safetensors",
320
+ "backbone.blocks.6.mlp.l2.weight": "model-00001-of-00002.safetensors",
321
+ "backbone.blocks.6.mlp.l3.weight": "model-00001-of-00002.safetensors",
322
+ "backbone.blocks.6.out_filter_dense.bias": "model-00001-of-00002.safetensors",
323
+ "backbone.blocks.6.out_filter_dense.weight": "model-00001-of-00002.safetensors",
324
+ "backbone.blocks.6.post_norm.scale": "model-00001-of-00002.safetensors",
325
+ "backbone.blocks.6.pre_norm.scale": "model-00001-of-00002.safetensors",
326
+ "backbone.blocks.6.projections.bias": "model-00001-of-00002.safetensors",
327
+ "backbone.blocks.6.projections.weight": "model-00001-of-00002.safetensors",
328
+ "backbone.blocks.7.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
329
+ "backbone.blocks.7.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
330
+ "backbone.blocks.7.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
331
+ "backbone.blocks.7.mlp.l1.weight": "model-00001-of-00002.safetensors",
332
+ "backbone.blocks.7.mlp.l2.weight": "model-00001-of-00002.safetensors",
333
+ "backbone.blocks.7.mlp.l3.weight": "model-00001-of-00002.safetensors",
334
+ "backbone.blocks.7.post_norm.scale": "model-00001-of-00002.safetensors",
335
+ "backbone.blocks.7.pre_norm.scale": "model-00001-of-00002.safetensors",
336
+ "backbone.blocks.8.filter.D": "model-00001-of-00002.safetensors",
337
+ "backbone.blocks.8.filter.poles": "model-00001-of-00002.safetensors",
338
+ "backbone.blocks.8.filter.residues": "model-00001-of-00002.safetensors",
339
+ "backbone.blocks.8.filter.short_filter_bias": "model-00001-of-00002.safetensors",
340
+ "backbone.blocks.8.filter.short_filter_weight": "model-00001-of-00002.safetensors",
341
+ "backbone.blocks.8.mlp.l1.weight": "model-00001-of-00002.safetensors",
342
+ "backbone.blocks.8.mlp.l2.weight": "model-00001-of-00002.safetensors",
343
+ "backbone.blocks.8.mlp.l3.weight": "model-00001-of-00002.safetensors",
344
+ "backbone.blocks.8.out_filter_dense.bias": "model-00001-of-00002.safetensors",
345
+ "backbone.blocks.8.out_filter_dense.weight": "model-00001-of-00002.safetensors",
346
+ "backbone.blocks.8.post_norm.scale": "model-00001-of-00002.safetensors",
347
+ "backbone.blocks.8.pre_norm.scale": "model-00001-of-00002.safetensors",
348
+ "backbone.blocks.8.projections.bias": "model-00001-of-00002.safetensors",
349
+ "backbone.blocks.8.projections.weight": "model-00001-of-00002.safetensors",
350
+ "backbone.blocks.9.inner_mha_cls.Wqkv.weight": "model-00001-of-00002.safetensors",
351
+ "backbone.blocks.9.inner_mha_cls.out_proj.weight": "model-00001-of-00002.safetensors",
352
+ "backbone.blocks.9.inner_mha_cls.rotary_emb.inv_freq": "model-00001-of-00002.safetensors",
353
+ "backbone.blocks.9.mlp.l1.weight": "model-00001-of-00002.safetensors",
354
+ "backbone.blocks.9.mlp.l2.weight": "model-00001-of-00002.safetensors",
355
+ "backbone.blocks.9.mlp.l3.weight": "model-00001-of-00002.safetensors",
356
+ "backbone.blocks.9.post_norm.scale": "model-00001-of-00002.safetensors",
357
+ "backbone.blocks.9.pre_norm.scale": "model-00001-of-00002.safetensors",
358
+ "backbone.embedding_layer.weight": "model-00001-of-00002.safetensors",
359
+ "backbone.norm.scale": "model-00001-of-00002.safetensors",
360
+ "backbone.unembed.weight": "model-00001-of-00002.safetensors"
361
+ }
362
+ }