Hugging Face's logo

umuthopeyildirim
/
fin-rwkv-1b5

Text Generation
Transformers
PyTorch
Safetensors
English
rwkv
finance
Inference Endpoints
Model card Files Community
1

Adding `safetensors` variant of this model

#1
by SFconvertbot - opened
base: refs/heads/main
←
from: refs/pr/1
Discussion Files changed
+457
-0
Files changed (5) hide show
  1. model-00001-of-00004.safetensors +3 -0
  2. model-00002-of-00004.safetensors +3 -0
  3. model-00003-of-00004.safetensors +3 -0
  4. model-00004-of-00004.safetensors +3 -0
  5. model.safetensors.index.json +445 -0
model-00001-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af75c914bf5135646f83103da87a311d80180b09c86e7c768deff75dbd724582
3
+ size 1989663816
model-00002-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c54a729630bb22ba0e2a1d4edb61de9a7051fb9941fa27ca4343945203527b8
3
+ size 1980557200
model-00003-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b7e0786a47be37e307a8e33edced3dda9bb76b63d3fb0ccb523720a99c48f928
3
+ size 1678383312
model-00004-of-00004.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f0ecd11be958a7c4749296a44a3cc4cd4a283b281165083d4bb7a2f6dc61bcec
3
+ size 411869304
model.safetensors.index.json ADDED
@@ -0,0 +1,445 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 6060425216
4
+ },
5
+ "weight_map": {
6
+ "head.weight": "model-00004-of-00004.safetensors",
7
+ "rwkv.blocks.0.attention.key.weight": "model-00001-of-00004.safetensors",
8
+ "rwkv.blocks.0.attention.output.weight": "model-00001-of-00004.safetensors",
9
+ "rwkv.blocks.0.attention.receptance.weight": "model-00001-of-00004.safetensors",
10
+ "rwkv.blocks.0.attention.time_decay": "model-00001-of-00004.safetensors",
11
+ "rwkv.blocks.0.attention.time_first": "model-00001-of-00004.safetensors",
12
+ "rwkv.blocks.0.attention.time_mix_key": "model-00001-of-00004.safetensors",
13
+ "rwkv.blocks.0.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
14
+ "rwkv.blocks.0.attention.time_mix_value": "model-00001-of-00004.safetensors",
15
+ "rwkv.blocks.0.attention.value.weight": "model-00001-of-00004.safetensors",
16
+ "rwkv.blocks.0.feed_forward.key.weight": "model-00001-of-00004.safetensors",
17
+ "rwkv.blocks.0.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
18
+ "rwkv.blocks.0.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
19
+ "rwkv.blocks.0.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
20
+ "rwkv.blocks.0.feed_forward.value.weight": "model-00001-of-00004.safetensors",
21
+ "rwkv.blocks.0.ln1.bias": "model-00001-of-00004.safetensors",
22
+ "rwkv.blocks.0.ln1.weight": "model-00001-of-00004.safetensors",
23
+ "rwkv.blocks.0.ln2.bias": "model-00001-of-00004.safetensors",
24
+ "rwkv.blocks.0.ln2.weight": "model-00001-of-00004.safetensors",
25
+ "rwkv.blocks.0.pre_ln.bias": "model-00001-of-00004.safetensors",
26
+ "rwkv.blocks.0.pre_ln.weight": "model-00001-of-00004.safetensors",
27
+ "rwkv.blocks.1.attention.key.weight": "model-00001-of-00004.safetensors",
28
+ "rwkv.blocks.1.attention.output.weight": "model-00001-of-00004.safetensors",
29
+ "rwkv.blocks.1.attention.receptance.weight": "model-00001-of-00004.safetensors",
30
+ "rwkv.blocks.1.attention.time_decay": "model-00001-of-00004.safetensors",
31
+ "rwkv.blocks.1.attention.time_first": "model-00001-of-00004.safetensors",
32
+ "rwkv.blocks.1.attention.time_mix_key": "model-00001-of-00004.safetensors",
33
+ "rwkv.blocks.1.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
34
+ "rwkv.blocks.1.attention.time_mix_value": "model-00001-of-00004.safetensors",
35
+ "rwkv.blocks.1.attention.value.weight": "model-00001-of-00004.safetensors",
36
+ "rwkv.blocks.1.feed_forward.key.weight": "model-00001-of-00004.safetensors",
37
+ "rwkv.blocks.1.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
38
+ "rwkv.blocks.1.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
39
+ "rwkv.blocks.1.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
40
+ "rwkv.blocks.1.feed_forward.value.weight": "model-00001-of-00004.safetensors",
41
+ "rwkv.blocks.1.ln1.bias": "model-00001-of-00004.safetensors",
42
+ "rwkv.blocks.1.ln1.weight": "model-00001-of-00004.safetensors",
43
+ "rwkv.blocks.1.ln2.bias": "model-00001-of-00004.safetensors",
44
+ "rwkv.blocks.1.ln2.weight": "model-00001-of-00004.safetensors",
45
+ "rwkv.blocks.10.attention.key.weight": "model-00002-of-00004.safetensors",
46
+ "rwkv.blocks.10.attention.output.weight": "model-00002-of-00004.safetensors",
47
+ "rwkv.blocks.10.attention.receptance.weight": "model-00002-of-00004.safetensors",
48
+ "rwkv.blocks.10.attention.time_decay": "model-00002-of-00004.safetensors",
49
+ "rwkv.blocks.10.attention.time_first": "model-00002-of-00004.safetensors",
50
+ "rwkv.blocks.10.attention.time_mix_key": "model-00002-of-00004.safetensors",
51
+ "rwkv.blocks.10.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
52
+ "rwkv.blocks.10.attention.time_mix_value": "model-00002-of-00004.safetensors",
53
+ "rwkv.blocks.10.attention.value.weight": "model-00002-of-00004.safetensors",
54
+ "rwkv.blocks.10.feed_forward.key.weight": "model-00002-of-00004.safetensors",
55
+ "rwkv.blocks.10.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
56
+ "rwkv.blocks.10.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
57
+ "rwkv.blocks.10.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
58
+ "rwkv.blocks.10.feed_forward.value.weight": "model-00002-of-00004.safetensors",
59
+ "rwkv.blocks.10.ln1.bias": "model-00002-of-00004.safetensors",
60
+ "rwkv.blocks.10.ln1.weight": "model-00002-of-00004.safetensors",
61
+ "rwkv.blocks.10.ln2.bias": "model-00002-of-00004.safetensors",
62
+ "rwkv.blocks.10.ln2.weight": "model-00002-of-00004.safetensors",
63
+ "rwkv.blocks.11.attention.key.weight": "model-00002-of-00004.safetensors",
64
+ "rwkv.blocks.11.attention.output.weight": "model-00002-of-00004.safetensors",
65
+ "rwkv.blocks.11.attention.receptance.weight": "model-00002-of-00004.safetensors",
66
+ "rwkv.blocks.11.attention.time_decay": "model-00002-of-00004.safetensors",
67
+ "rwkv.blocks.11.attention.time_first": "model-00002-of-00004.safetensors",
68
+ "rwkv.blocks.11.attention.time_mix_key": "model-00002-of-00004.safetensors",
69
+ "rwkv.blocks.11.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
70
+ "rwkv.blocks.11.attention.time_mix_value": "model-00002-of-00004.safetensors",
71
+ "rwkv.blocks.11.attention.value.weight": "model-00002-of-00004.safetensors",
72
+ "rwkv.blocks.11.feed_forward.key.weight": "model-00002-of-00004.safetensors",
73
+ "rwkv.blocks.11.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
74
+ "rwkv.blocks.11.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
75
+ "rwkv.blocks.11.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
76
+ "rwkv.blocks.11.feed_forward.value.weight": "model-00002-of-00004.safetensors",
77
+ "rwkv.blocks.11.ln1.bias": "model-00002-of-00004.safetensors",
78
+ "rwkv.blocks.11.ln1.weight": "model-00002-of-00004.safetensors",
79
+ "rwkv.blocks.11.ln2.bias": "model-00002-of-00004.safetensors",
80
+ "rwkv.blocks.11.ln2.weight": "model-00002-of-00004.safetensors",
81
+ "rwkv.blocks.12.attention.key.weight": "model-00002-of-00004.safetensors",
82
+ "rwkv.blocks.12.attention.output.weight": "model-00002-of-00004.safetensors",
83
+ "rwkv.blocks.12.attention.receptance.weight": "model-00002-of-00004.safetensors",
84
+ "rwkv.blocks.12.attention.time_decay": "model-00002-of-00004.safetensors",
85
+ "rwkv.blocks.12.attention.time_first": "model-00002-of-00004.safetensors",
86
+ "rwkv.blocks.12.attention.time_mix_key": "model-00002-of-00004.safetensors",
87
+ "rwkv.blocks.12.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
88
+ "rwkv.blocks.12.attention.time_mix_value": "model-00002-of-00004.safetensors",
89
+ "rwkv.blocks.12.attention.value.weight": "model-00002-of-00004.safetensors",
90
+ "rwkv.blocks.12.feed_forward.key.weight": "model-00002-of-00004.safetensors",
91
+ "rwkv.blocks.12.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
92
+ "rwkv.blocks.12.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
93
+ "rwkv.blocks.12.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
94
+ "rwkv.blocks.12.feed_forward.value.weight": "model-00002-of-00004.safetensors",
95
+ "rwkv.blocks.12.ln1.bias": "model-00002-of-00004.safetensors",
96
+ "rwkv.blocks.12.ln1.weight": "model-00002-of-00004.safetensors",
97
+ "rwkv.blocks.12.ln2.bias": "model-00002-of-00004.safetensors",
98
+ "rwkv.blocks.12.ln2.weight": "model-00002-of-00004.safetensors",
99
+ "rwkv.blocks.13.attention.key.weight": "model-00002-of-00004.safetensors",
100
+ "rwkv.blocks.13.attention.output.weight": "model-00002-of-00004.safetensors",
101
+ "rwkv.blocks.13.attention.receptance.weight": "model-00002-of-00004.safetensors",
102
+ "rwkv.blocks.13.attention.time_decay": "model-00002-of-00004.safetensors",
103
+ "rwkv.blocks.13.attention.time_first": "model-00002-of-00004.safetensors",
104
+ "rwkv.blocks.13.attention.time_mix_key": "model-00002-of-00004.safetensors",
105
+ "rwkv.blocks.13.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
106
+ "rwkv.blocks.13.attention.time_mix_value": "model-00002-of-00004.safetensors",
107
+ "rwkv.blocks.13.attention.value.weight": "model-00002-of-00004.safetensors",
108
+ "rwkv.blocks.13.feed_forward.key.weight": "model-00002-of-00004.safetensors",
109
+ "rwkv.blocks.13.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
110
+ "rwkv.blocks.13.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
111
+ "rwkv.blocks.13.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
112
+ "rwkv.blocks.13.feed_forward.value.weight": "model-00002-of-00004.safetensors",
113
+ "rwkv.blocks.13.ln1.bias": "model-00002-of-00004.safetensors",
114
+ "rwkv.blocks.13.ln1.weight": "model-00002-of-00004.safetensors",
115
+ "rwkv.blocks.13.ln2.bias": "model-00002-of-00004.safetensors",
116
+ "rwkv.blocks.13.ln2.weight": "model-00002-of-00004.safetensors",
117
+ "rwkv.blocks.14.attention.key.weight": "model-00002-of-00004.safetensors",
118
+ "rwkv.blocks.14.attention.output.weight": "model-00002-of-00004.safetensors",
119
+ "rwkv.blocks.14.attention.receptance.weight": "model-00002-of-00004.safetensors",
120
+ "rwkv.blocks.14.attention.time_decay": "model-00002-of-00004.safetensors",
121
+ "rwkv.blocks.14.attention.time_first": "model-00002-of-00004.safetensors",
122
+ "rwkv.blocks.14.attention.time_mix_key": "model-00002-of-00004.safetensors",
123
+ "rwkv.blocks.14.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
124
+ "rwkv.blocks.14.attention.time_mix_value": "model-00002-of-00004.safetensors",
125
+ "rwkv.blocks.14.attention.value.weight": "model-00002-of-00004.safetensors",
126
+ "rwkv.blocks.14.feed_forward.key.weight": "model-00002-of-00004.safetensors",
127
+ "rwkv.blocks.14.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
128
+ "rwkv.blocks.14.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
129
+ "rwkv.blocks.14.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
130
+ "rwkv.blocks.14.feed_forward.value.weight": "model-00002-of-00004.safetensors",
131
+ "rwkv.blocks.14.ln1.bias": "model-00002-of-00004.safetensors",
132
+ "rwkv.blocks.14.ln1.weight": "model-00002-of-00004.safetensors",
133
+ "rwkv.blocks.14.ln2.bias": "model-00002-of-00004.safetensors",
134
+ "rwkv.blocks.14.ln2.weight": "model-00002-of-00004.safetensors",
135
+ "rwkv.blocks.15.attention.key.weight": "model-00002-of-00004.safetensors",
136
+ "rwkv.blocks.15.attention.output.weight": "model-00002-of-00004.safetensors",
137
+ "rwkv.blocks.15.attention.receptance.weight": "model-00002-of-00004.safetensors",
138
+ "rwkv.blocks.15.attention.time_decay": "model-00002-of-00004.safetensors",
139
+ "rwkv.blocks.15.attention.time_first": "model-00002-of-00004.safetensors",
140
+ "rwkv.blocks.15.attention.time_mix_key": "model-00002-of-00004.safetensors",
141
+ "rwkv.blocks.15.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
142
+ "rwkv.blocks.15.attention.time_mix_value": "model-00002-of-00004.safetensors",
143
+ "rwkv.blocks.15.attention.value.weight": "model-00002-of-00004.safetensors",
144
+ "rwkv.blocks.15.feed_forward.key.weight": "model-00002-of-00004.safetensors",
145
+ "rwkv.blocks.15.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
146
+ "rwkv.blocks.15.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
147
+ "rwkv.blocks.15.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
148
+ "rwkv.blocks.15.feed_forward.value.weight": "model-00002-of-00004.safetensors",
149
+ "rwkv.blocks.15.ln1.bias": "model-00002-of-00004.safetensors",
150
+ "rwkv.blocks.15.ln1.weight": "model-00002-of-00004.safetensors",
151
+ "rwkv.blocks.15.ln2.bias": "model-00002-of-00004.safetensors",
152
+ "rwkv.blocks.15.ln2.weight": "model-00002-of-00004.safetensors",
153
+ "rwkv.blocks.16.attention.key.weight": "model-00002-of-00004.safetensors",
154
+ "rwkv.blocks.16.attention.output.weight": "model-00002-of-00004.safetensors",
155
+ "rwkv.blocks.16.attention.receptance.weight": "model-00002-of-00004.safetensors",
156
+ "rwkv.blocks.16.attention.time_decay": "model-00002-of-00004.safetensors",
157
+ "rwkv.blocks.16.attention.time_first": "model-00002-of-00004.safetensors",
158
+ "rwkv.blocks.16.attention.time_mix_key": "model-00002-of-00004.safetensors",
159
+ "rwkv.blocks.16.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
160
+ "rwkv.blocks.16.attention.time_mix_value": "model-00002-of-00004.safetensors",
161
+ "rwkv.blocks.16.attention.value.weight": "model-00002-of-00004.safetensors",
162
+ "rwkv.blocks.16.feed_forward.key.weight": "model-00003-of-00004.safetensors",
163
+ "rwkv.blocks.16.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
164
+ "rwkv.blocks.16.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
165
+ "rwkv.blocks.16.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
166
+ "rwkv.blocks.16.feed_forward.value.weight": "model-00003-of-00004.safetensors",
167
+ "rwkv.blocks.16.ln1.bias": "model-00002-of-00004.safetensors",
168
+ "rwkv.blocks.16.ln1.weight": "model-00002-of-00004.safetensors",
169
+ "rwkv.blocks.16.ln2.bias": "model-00002-of-00004.safetensors",
170
+ "rwkv.blocks.16.ln2.weight": "model-00002-of-00004.safetensors",
171
+ "rwkv.blocks.17.attention.key.weight": "model-00003-of-00004.safetensors",
172
+ "rwkv.blocks.17.attention.output.weight": "model-00003-of-00004.safetensors",
173
+ "rwkv.blocks.17.attention.receptance.weight": "model-00003-of-00004.safetensors",
174
+ "rwkv.blocks.17.attention.time_decay": "model-00003-of-00004.safetensors",
175
+ "rwkv.blocks.17.attention.time_first": "model-00003-of-00004.safetensors",
176
+ "rwkv.blocks.17.attention.time_mix_key": "model-00003-of-00004.safetensors",
177
+ "rwkv.blocks.17.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
178
+ "rwkv.blocks.17.attention.time_mix_value": "model-00003-of-00004.safetensors",
179
+ "rwkv.blocks.17.attention.value.weight": "model-00003-of-00004.safetensors",
180
+ "rwkv.blocks.17.feed_forward.key.weight": "model-00003-of-00004.safetensors",
181
+ "rwkv.blocks.17.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
182
+ "rwkv.blocks.17.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
183
+ "rwkv.blocks.17.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
184
+ "rwkv.blocks.17.feed_forward.value.weight": "model-00003-of-00004.safetensors",
185
+ "rwkv.blocks.17.ln1.bias": "model-00003-of-00004.safetensors",
186
+ "rwkv.blocks.17.ln1.weight": "model-00003-of-00004.safetensors",
187
+ "rwkv.blocks.17.ln2.bias": "model-00003-of-00004.safetensors",
188
+ "rwkv.blocks.17.ln2.weight": "model-00003-of-00004.safetensors",
189
+ "rwkv.blocks.18.attention.key.weight": "model-00003-of-00004.safetensors",
190
+ "rwkv.blocks.18.attention.output.weight": "model-00003-of-00004.safetensors",
191
+ "rwkv.blocks.18.attention.receptance.weight": "model-00003-of-00004.safetensors",
192
+ "rwkv.blocks.18.attention.time_decay": "model-00003-of-00004.safetensors",
193
+ "rwkv.blocks.18.attention.time_first": "model-00003-of-00004.safetensors",
194
+ "rwkv.blocks.18.attention.time_mix_key": "model-00003-of-00004.safetensors",
195
+ "rwkv.blocks.18.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
196
+ "rwkv.blocks.18.attention.time_mix_value": "model-00003-of-00004.safetensors",
197
+ "rwkv.blocks.18.attention.value.weight": "model-00003-of-00004.safetensors",
198
+ "rwkv.blocks.18.feed_forward.key.weight": "model-00003-of-00004.safetensors",
199
+ "rwkv.blocks.18.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
200
+ "rwkv.blocks.18.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
201
+ "rwkv.blocks.18.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
202
+ "rwkv.blocks.18.feed_forward.value.weight": "model-00003-of-00004.safetensors",
203
+ "rwkv.blocks.18.ln1.bias": "model-00003-of-00004.safetensors",
204
+ "rwkv.blocks.18.ln1.weight": "model-00003-of-00004.safetensors",
205
+ "rwkv.blocks.18.ln2.bias": "model-00003-of-00004.safetensors",
206
+ "rwkv.blocks.18.ln2.weight": "model-00003-of-00004.safetensors",
207
+ "rwkv.blocks.19.attention.key.weight": "model-00003-of-00004.safetensors",
208
+ "rwkv.blocks.19.attention.output.weight": "model-00003-of-00004.safetensors",
209
+ "rwkv.blocks.19.attention.receptance.weight": "model-00003-of-00004.safetensors",
210
+ "rwkv.blocks.19.attention.time_decay": "model-00003-of-00004.safetensors",
211
+ "rwkv.blocks.19.attention.time_first": "model-00003-of-00004.safetensors",
212
+ "rwkv.blocks.19.attention.time_mix_key": "model-00003-of-00004.safetensors",
213
+ "rwkv.blocks.19.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
214
+ "rwkv.blocks.19.attention.time_mix_value": "model-00003-of-00004.safetensors",
215
+ "rwkv.blocks.19.attention.value.weight": "model-00003-of-00004.safetensors",
216
+ "rwkv.blocks.19.feed_forward.key.weight": "model-00003-of-00004.safetensors",
217
+ "rwkv.blocks.19.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
218
+ "rwkv.blocks.19.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
219
+ "rwkv.blocks.19.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
220
+ "rwkv.blocks.19.feed_forward.value.weight": "model-00003-of-00004.safetensors",
221
+ "rwkv.blocks.19.ln1.bias": "model-00003-of-00004.safetensors",
222
+ "rwkv.blocks.19.ln1.weight": "model-00003-of-00004.safetensors",
223
+ "rwkv.blocks.19.ln2.bias": "model-00003-of-00004.safetensors",
224
+ "rwkv.blocks.19.ln2.weight": "model-00003-of-00004.safetensors",
225
+ "rwkv.blocks.2.attention.key.weight": "model-00001-of-00004.safetensors",
226
+ "rwkv.blocks.2.attention.output.weight": "model-00001-of-00004.safetensors",
227
+ "rwkv.blocks.2.attention.receptance.weight": "model-00001-of-00004.safetensors",
228
+ "rwkv.blocks.2.attention.time_decay": "model-00001-of-00004.safetensors",
229
+ "rwkv.blocks.2.attention.time_first": "model-00001-of-00004.safetensors",
230
+ "rwkv.blocks.2.attention.time_mix_key": "model-00001-of-00004.safetensors",
231
+ "rwkv.blocks.2.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
232
+ "rwkv.blocks.2.attention.time_mix_value": "model-00001-of-00004.safetensors",
233
+ "rwkv.blocks.2.attention.value.weight": "model-00001-of-00004.safetensors",
234
+ "rwkv.blocks.2.feed_forward.key.weight": "model-00001-of-00004.safetensors",
235
+ "rwkv.blocks.2.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
236
+ "rwkv.blocks.2.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
237
+ "rwkv.blocks.2.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
238
+ "rwkv.blocks.2.feed_forward.value.weight": "model-00001-of-00004.safetensors",
239
+ "rwkv.blocks.2.ln1.bias": "model-00001-of-00004.safetensors",
240
+ "rwkv.blocks.2.ln1.weight": "model-00001-of-00004.safetensors",
241
+ "rwkv.blocks.2.ln2.bias": "model-00001-of-00004.safetensors",
242
+ "rwkv.blocks.2.ln2.weight": "model-00001-of-00004.safetensors",
243
+ "rwkv.blocks.20.attention.key.weight": "model-00003-of-00004.safetensors",
244
+ "rwkv.blocks.20.attention.output.weight": "model-00003-of-00004.safetensors",
245
+ "rwkv.blocks.20.attention.receptance.weight": "model-00003-of-00004.safetensors",
246
+ "rwkv.blocks.20.attention.time_decay": "model-00003-of-00004.safetensors",
247
+ "rwkv.blocks.20.attention.time_first": "model-00003-of-00004.safetensors",
248
+ "rwkv.blocks.20.attention.time_mix_key": "model-00003-of-00004.safetensors",
249
+ "rwkv.blocks.20.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
250
+ "rwkv.blocks.20.attention.time_mix_value": "model-00003-of-00004.safetensors",
251
+ "rwkv.blocks.20.attention.value.weight": "model-00003-of-00004.safetensors",
252
+ "rwkv.blocks.20.feed_forward.key.weight": "model-00003-of-00004.safetensors",
253
+ "rwkv.blocks.20.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
254
+ "rwkv.blocks.20.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
255
+ "rwkv.blocks.20.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
256
+ "rwkv.blocks.20.feed_forward.value.weight": "model-00003-of-00004.safetensors",
257
+ "rwkv.blocks.20.ln1.bias": "model-00003-of-00004.safetensors",
258
+ "rwkv.blocks.20.ln1.weight": "model-00003-of-00004.safetensors",
259
+ "rwkv.blocks.20.ln2.bias": "model-00003-of-00004.safetensors",
260
+ "rwkv.blocks.20.ln2.weight": "model-00003-of-00004.safetensors",
261
+ "rwkv.blocks.21.attention.key.weight": "model-00003-of-00004.safetensors",
262
+ "rwkv.blocks.21.attention.output.weight": "model-00003-of-00004.safetensors",
263
+ "rwkv.blocks.21.attention.receptance.weight": "model-00003-of-00004.safetensors",
264
+ "rwkv.blocks.21.attention.time_decay": "model-00003-of-00004.safetensors",
265
+ "rwkv.blocks.21.attention.time_first": "model-00003-of-00004.safetensors",
266
+ "rwkv.blocks.21.attention.time_mix_key": "model-00003-of-00004.safetensors",
267
+ "rwkv.blocks.21.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
268
+ "rwkv.blocks.21.attention.time_mix_value": "model-00003-of-00004.safetensors",
269
+ "rwkv.blocks.21.attention.value.weight": "model-00003-of-00004.safetensors",
270
+ "rwkv.blocks.21.feed_forward.key.weight": "model-00003-of-00004.safetensors",
271
+ "rwkv.blocks.21.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
272
+ "rwkv.blocks.21.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
273
+ "rwkv.blocks.21.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
274
+ "rwkv.blocks.21.feed_forward.value.weight": "model-00003-of-00004.safetensors",
275
+ "rwkv.blocks.21.ln1.bias": "model-00003-of-00004.safetensors",
276
+ "rwkv.blocks.21.ln1.weight": "model-00003-of-00004.safetensors",
277
+ "rwkv.blocks.21.ln2.bias": "model-00003-of-00004.safetensors",
278
+ "rwkv.blocks.21.ln2.weight": "model-00003-of-00004.safetensors",
279
+ "rwkv.blocks.22.attention.key.weight": "model-00003-of-00004.safetensors",
280
+ "rwkv.blocks.22.attention.output.weight": "model-00003-of-00004.safetensors",
281
+ "rwkv.blocks.22.attention.receptance.weight": "model-00003-of-00004.safetensors",
282
+ "rwkv.blocks.22.attention.time_decay": "model-00003-of-00004.safetensors",
283
+ "rwkv.blocks.22.attention.time_first": "model-00003-of-00004.safetensors",
284
+ "rwkv.blocks.22.attention.time_mix_key": "model-00003-of-00004.safetensors",
285
+ "rwkv.blocks.22.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
286
+ "rwkv.blocks.22.attention.time_mix_value": "model-00003-of-00004.safetensors",
287
+ "rwkv.blocks.22.attention.value.weight": "model-00003-of-00004.safetensors",
288
+ "rwkv.blocks.22.feed_forward.key.weight": "model-00003-of-00004.safetensors",
289
+ "rwkv.blocks.22.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
290
+ "rwkv.blocks.22.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
291
+ "rwkv.blocks.22.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
292
+ "rwkv.blocks.22.feed_forward.value.weight": "model-00003-of-00004.safetensors",
293
+ "rwkv.blocks.22.ln1.bias": "model-00003-of-00004.safetensors",
294
+ "rwkv.blocks.22.ln1.weight": "model-00003-of-00004.safetensors",
295
+ "rwkv.blocks.22.ln2.bias": "model-00003-of-00004.safetensors",
296
+ "rwkv.blocks.22.ln2.weight": "model-00003-of-00004.safetensors",
297
+ "rwkv.blocks.23.attention.key.weight": "model-00003-of-00004.safetensors",
298
+ "rwkv.blocks.23.attention.output.weight": "model-00003-of-00004.safetensors",
299
+ "rwkv.blocks.23.attention.receptance.weight": "model-00003-of-00004.safetensors",
300
+ "rwkv.blocks.23.attention.time_decay": "model-00003-of-00004.safetensors",
301
+ "rwkv.blocks.23.attention.time_first": "model-00003-of-00004.safetensors",
302
+ "rwkv.blocks.23.attention.time_mix_key": "model-00003-of-00004.safetensors",
303
+ "rwkv.blocks.23.attention.time_mix_receptance": "model-00003-of-00004.safetensors",
304
+ "rwkv.blocks.23.attention.time_mix_value": "model-00003-of-00004.safetensors",
305
+ "rwkv.blocks.23.attention.value.weight": "model-00003-of-00004.safetensors",
306
+ "rwkv.blocks.23.feed_forward.key.weight": "model-00003-of-00004.safetensors",
307
+ "rwkv.blocks.23.feed_forward.receptance.weight": "model-00003-of-00004.safetensors",
308
+ "rwkv.blocks.23.feed_forward.time_mix_key": "model-00003-of-00004.safetensors",
309
+ "rwkv.blocks.23.feed_forward.time_mix_receptance": "model-00003-of-00004.safetensors",
310
+ "rwkv.blocks.23.feed_forward.value.weight": "model-00003-of-00004.safetensors",
311
+ "rwkv.blocks.23.ln1.bias": "model-00003-of-00004.safetensors",
312
+ "rwkv.blocks.23.ln1.weight": "model-00003-of-00004.safetensors",
313
+ "rwkv.blocks.23.ln2.bias": "model-00003-of-00004.safetensors",
314
+ "rwkv.blocks.23.ln2.weight": "model-00003-of-00004.safetensors",
315
+ "rwkv.blocks.3.attention.key.weight": "model-00001-of-00004.safetensors",
316
+ "rwkv.blocks.3.attention.output.weight": "model-00001-of-00004.safetensors",
317
+ "rwkv.blocks.3.attention.receptance.weight": "model-00001-of-00004.safetensors",
318
+ "rwkv.blocks.3.attention.time_decay": "model-00001-of-00004.safetensors",
319
+ "rwkv.blocks.3.attention.time_first": "model-00001-of-00004.safetensors",
320
+ "rwkv.blocks.3.attention.time_mix_key": "model-00001-of-00004.safetensors",
321
+ "rwkv.blocks.3.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
322
+ "rwkv.blocks.3.attention.time_mix_value": "model-00001-of-00004.safetensors",
323
+ "rwkv.blocks.3.attention.value.weight": "model-00001-of-00004.safetensors",
324
+ "rwkv.blocks.3.feed_forward.key.weight": "model-00001-of-00004.safetensors",
325
+ "rwkv.blocks.3.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
326
+ "rwkv.blocks.3.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
327
+ "rwkv.blocks.3.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
328
+ "rwkv.blocks.3.feed_forward.value.weight": "model-00001-of-00004.safetensors",
329
+ "rwkv.blocks.3.ln1.bias": "model-00001-of-00004.safetensors",
330
+ "rwkv.blocks.3.ln1.weight": "model-00001-of-00004.safetensors",
331
+ "rwkv.blocks.3.ln2.bias": "model-00001-of-00004.safetensors",
332
+ "rwkv.blocks.3.ln2.weight": "model-00001-of-00004.safetensors",
333
+ "rwkv.blocks.4.attention.key.weight": "model-00001-of-00004.safetensors",
334
+ "rwkv.blocks.4.attention.output.weight": "model-00001-of-00004.safetensors",
335
+ "rwkv.blocks.4.attention.receptance.weight": "model-00001-of-00004.safetensors",
336
+ "rwkv.blocks.4.attention.time_decay": "model-00001-of-00004.safetensors",
337
+ "rwkv.blocks.4.attention.time_first": "model-00001-of-00004.safetensors",
338
+ "rwkv.blocks.4.attention.time_mix_key": "model-00001-of-00004.safetensors",
339
+ "rwkv.blocks.4.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
340
+ "rwkv.blocks.4.attention.time_mix_value": "model-00001-of-00004.safetensors",
341
+ "rwkv.blocks.4.attention.value.weight": "model-00001-of-00004.safetensors",
342
+ "rwkv.blocks.4.feed_forward.key.weight": "model-00001-of-00004.safetensors",
343
+ "rwkv.blocks.4.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
344
+ "rwkv.blocks.4.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
345
+ "rwkv.blocks.4.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
346
+ "rwkv.blocks.4.feed_forward.value.weight": "model-00001-of-00004.safetensors",
347
+ "rwkv.blocks.4.ln1.bias": "model-00001-of-00004.safetensors",
348
+ "rwkv.blocks.4.ln1.weight": "model-00001-of-00004.safetensors",
349
+ "rwkv.blocks.4.ln2.bias": "model-00001-of-00004.safetensors",
350
+ "rwkv.blocks.4.ln2.weight": "model-00001-of-00004.safetensors",
351
+ "rwkv.blocks.5.attention.key.weight": "model-00001-of-00004.safetensors",
352
+ "rwkv.blocks.5.attention.output.weight": "model-00001-of-00004.safetensors",
353
+ "rwkv.blocks.5.attention.receptance.weight": "model-00001-of-00004.safetensors",
354
+ "rwkv.blocks.5.attention.time_decay": "model-00001-of-00004.safetensors",
355
+ "rwkv.blocks.5.attention.time_first": "model-00001-of-00004.safetensors",
356
+ "rwkv.blocks.5.attention.time_mix_key": "model-00001-of-00004.safetensors",
357
+ "rwkv.blocks.5.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
358
+ "rwkv.blocks.5.attention.time_mix_value": "model-00001-of-00004.safetensors",
359
+ "rwkv.blocks.5.attention.value.weight": "model-00001-of-00004.safetensors",
360
+ "rwkv.blocks.5.feed_forward.key.weight": "model-00001-of-00004.safetensors",
361
+ "rwkv.blocks.5.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
362
+ "rwkv.blocks.5.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
363
+ "rwkv.blocks.5.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
364
+ "rwkv.blocks.5.feed_forward.value.weight": "model-00001-of-00004.safetensors",
365
+ "rwkv.blocks.5.ln1.bias": "model-00001-of-00004.safetensors",
366
+ "rwkv.blocks.5.ln1.weight": "model-00001-of-00004.safetensors",
367
+ "rwkv.blocks.5.ln2.bias": "model-00001-of-00004.safetensors",
368
+ "rwkv.blocks.5.ln2.weight": "model-00001-of-00004.safetensors",
369
+ "rwkv.blocks.6.attention.key.weight": "model-00001-of-00004.safetensors",
370
+ "rwkv.blocks.6.attention.output.weight": "model-00001-of-00004.safetensors",
371
+ "rwkv.blocks.6.attention.receptance.weight": "model-00001-of-00004.safetensors",
372
+ "rwkv.blocks.6.attention.time_decay": "model-00001-of-00004.safetensors",
373
+ "rwkv.blocks.6.attention.time_first": "model-00001-of-00004.safetensors",
374
+ "rwkv.blocks.6.attention.time_mix_key": "model-00001-of-00004.safetensors",
375
+ "rwkv.blocks.6.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
376
+ "rwkv.blocks.6.attention.time_mix_value": "model-00001-of-00004.safetensors",
377
+ "rwkv.blocks.6.attention.value.weight": "model-00001-of-00004.safetensors",
378
+ "rwkv.blocks.6.feed_forward.key.weight": "model-00001-of-00004.safetensors",
379
+ "rwkv.blocks.6.feed_forward.receptance.weight": "model-00001-of-00004.safetensors",
380
+ "rwkv.blocks.6.feed_forward.time_mix_key": "model-00001-of-00004.safetensors",
381
+ "rwkv.blocks.6.feed_forward.time_mix_receptance": "model-00001-of-00004.safetensors",
382
+ "rwkv.blocks.6.feed_forward.value.weight": "model-00001-of-00004.safetensors",
383
+ "rwkv.blocks.6.ln1.bias": "model-00001-of-00004.safetensors",
384
+ "rwkv.blocks.6.ln1.weight": "model-00001-of-00004.safetensors",
385
+ "rwkv.blocks.6.ln2.bias": "model-00001-of-00004.safetensors",
386
+ "rwkv.blocks.6.ln2.weight": "model-00001-of-00004.safetensors",
387
+ "rwkv.blocks.7.attention.key.weight": "model-00001-of-00004.safetensors",
388
+ "rwkv.blocks.7.attention.output.weight": "model-00002-of-00004.safetensors",
389
+ "rwkv.blocks.7.attention.receptance.weight": "model-00001-of-00004.safetensors",
390
+ "rwkv.blocks.7.attention.time_decay": "model-00001-of-00004.safetensors",
391
+ "rwkv.blocks.7.attention.time_first": "model-00001-of-00004.safetensors",
392
+ "rwkv.blocks.7.attention.time_mix_key": "model-00001-of-00004.safetensors",
393
+ "rwkv.blocks.7.attention.time_mix_receptance": "model-00001-of-00004.safetensors",
394
+ "rwkv.blocks.7.attention.time_mix_value": "model-00001-of-00004.safetensors",
395
+ "rwkv.blocks.7.attention.value.weight": "model-00001-of-00004.safetensors",
396
+ "rwkv.blocks.7.feed_forward.key.weight": "model-00002-of-00004.safetensors",
397
+ "rwkv.blocks.7.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
398
+ "rwkv.blocks.7.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
399
+ "rwkv.blocks.7.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
400
+ "rwkv.blocks.7.feed_forward.value.weight": "model-00002-of-00004.safetensors",
401
+ "rwkv.blocks.7.ln1.bias": "model-00001-of-00004.safetensors",
402
+ "rwkv.blocks.7.ln1.weight": "model-00001-of-00004.safetensors",
403
+ "rwkv.blocks.7.ln2.bias": "model-00001-of-00004.safetensors",
404
+ "rwkv.blocks.7.ln2.weight": "model-00001-of-00004.safetensors",
405
+ "rwkv.blocks.8.attention.key.weight": "model-00002-of-00004.safetensors",
406
+ "rwkv.blocks.8.attention.output.weight": "model-00002-of-00004.safetensors",
407
+ "rwkv.blocks.8.attention.receptance.weight": "model-00002-of-00004.safetensors",
408
+ "rwkv.blocks.8.attention.time_decay": "model-00002-of-00004.safetensors",
409
+ "rwkv.blocks.8.attention.time_first": "model-00002-of-00004.safetensors",
410
+ "rwkv.blocks.8.attention.time_mix_key": "model-00002-of-00004.safetensors",
411
+ "rwkv.blocks.8.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
412
+ "rwkv.blocks.8.attention.time_mix_value": "model-00002-of-00004.safetensors",
413
+ "rwkv.blocks.8.attention.value.weight": "model-00002-of-00004.safetensors",
414
+ "rwkv.blocks.8.feed_forward.key.weight": "model-00002-of-00004.safetensors",
415
+ "rwkv.blocks.8.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
416
+ "rwkv.blocks.8.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
417
+ "rwkv.blocks.8.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
418
+ "rwkv.blocks.8.feed_forward.value.weight": "model-00002-of-00004.safetensors",
419
+ "rwkv.blocks.8.ln1.bias": "model-00002-of-00004.safetensors",
420
+ "rwkv.blocks.8.ln1.weight": "model-00002-of-00004.safetensors",
421
+ "rwkv.blocks.8.ln2.bias": "model-00002-of-00004.safetensors",
422
+ "rwkv.blocks.8.ln2.weight": "model-00002-of-00004.safetensors",
423
+ "rwkv.blocks.9.attention.key.weight": "model-00002-of-00004.safetensors",
424
+ "rwkv.blocks.9.attention.output.weight": "model-00002-of-00004.safetensors",
425
+ "rwkv.blocks.9.attention.receptance.weight": "model-00002-of-00004.safetensors",
426
+ "rwkv.blocks.9.attention.time_decay": "model-00002-of-00004.safetensors",
427
+ "rwkv.blocks.9.attention.time_first": "model-00002-of-00004.safetensors",
428
+ "rwkv.blocks.9.attention.time_mix_key": "model-00002-of-00004.safetensors",
429
+ "rwkv.blocks.9.attention.time_mix_receptance": "model-00002-of-00004.safetensors",
430
+ "rwkv.blocks.9.attention.time_mix_value": "model-00002-of-00004.safetensors",
431
+ "rwkv.blocks.9.attention.value.weight": "model-00002-of-00004.safetensors",
432
+ "rwkv.blocks.9.feed_forward.key.weight": "model-00002-of-00004.safetensors",
433
+ "rwkv.blocks.9.feed_forward.receptance.weight": "model-00002-of-00004.safetensors",
434
+ "rwkv.blocks.9.feed_forward.time_mix_key": "model-00002-of-00004.safetensors",
435
+ "rwkv.blocks.9.feed_forward.time_mix_receptance": "model-00002-of-00004.safetensors",
436
+ "rwkv.blocks.9.feed_forward.value.weight": "model-00002-of-00004.safetensors",
437
+ "rwkv.blocks.9.ln1.bias": "model-00002-of-00004.safetensors",
438
+ "rwkv.blocks.9.ln1.weight": "model-00002-of-00004.safetensors",
439
+ "rwkv.blocks.9.ln2.bias": "model-00002-of-00004.safetensors",
440
+ "rwkv.blocks.9.ln2.weight": "model-00002-of-00004.safetensors",
441
+ "rwkv.embeddings.weight": "model-00001-of-00004.safetensors",
442
+ "rwkv.ln_out.bias": "model-00003-of-00004.safetensors",
443
+ "rwkv.ln_out.weight": "model-00003-of-00004.safetensors"
444
+ }
445
+ }