Add bfloat16 weight, with 4.37M step trained
Browse files- config.json +2 -2
- model-00001-of-00007.safetensors → model-00001-of-00004.safetensors +2 -2
- model-00002-of-00007.safetensors → model-00002-of-00004.safetensors +2 -2
- model-00003-of-00007.safetensors → model-00003-of-00004.safetensors +2 -2
- model-00004-of-00007.safetensors → model-00004-of-00004.safetensors +2 -2
- model-00005-of-00007.safetensors +0 -3
- model-00006-of-00007.safetensors +0 -3
- model-00007-of-00007.safetensors +0 -3
- model.safetensors.index.json +439 -439
- pytorch_model-00001-of-00004.bin +3 -0
- pytorch_model-00001-of-00007.bin +0 -3
- pytorch_model-00002-of-00004.bin +3 -0
- pytorch_model-00002-of-00007.bin +0 -3
- pytorch_model-00003-of-00004.bin +3 -0
- pytorch_model-00003-of-00007.bin +0 -3
- pytorch_model-00004-of-00004.bin +3 -0
- pytorch_model-00004-of-00007.bin +0 -3
- pytorch_model-00005-of-00007.bin +0 -3
- pytorch_model-00006-of-00007.bin +0 -3
- pytorch_model-00007-of-00007.bin +0 -3
- pytorch_model.bin.index.json +439 -439
config.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
],
|
| 5 |
"attention_hidden_size": 2048,
|
| 6 |
"bos_token_id": 0,
|
| 7 |
-
"context_length":
|
| 8 |
"eos_token_id": 0,
|
| 9 |
"hidden_size": 2048,
|
| 10 |
"intermediate_size": 8192,
|
|
@@ -13,7 +13,7 @@
|
|
| 13 |
"num_hidden_layers": 24,
|
| 14 |
"rescale_every": 6,
|
| 15 |
"tie_word_embeddings": false,
|
| 16 |
-
"torch_dtype": "
|
| 17 |
"transformers_version": "4.30.0.dev0",
|
| 18 |
"use_cache": true,
|
| 19 |
"vocab_size": 52000
|
|
|
|
| 4 |
],
|
| 5 |
"attention_hidden_size": 2048,
|
| 6 |
"bos_token_id": 0,
|
| 7 |
+
"context_length": 1024,
|
| 8 |
"eos_token_id": 0,
|
| 9 |
"hidden_size": 2048,
|
| 10 |
"intermediate_size": 8192,
|
|
|
|
| 13 |
"num_hidden_layers": 24,
|
| 14 |
"rescale_every": 6,
|
| 15 |
"tie_word_embeddings": false,
|
| 16 |
+
"torch_dtype": "bfloat16",
|
| 17 |
"transformers_version": "4.30.0.dev0",
|
| 18 |
"use_cache": true,
|
| 19 |
"vocab_size": 52000
|
model-00001-of-00007.safetensors → model-00001-of-00004.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d75fb9d95860922589dcddc966b8fc6b6384e3d7bd400e5af1a9a58b775f053
|
| 3 |
+
size 993508320
|
model-00002-of-00007.safetensors → model-00002-of-00004.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64147d0c166b0d5c7509bf22bf18ad6d8fbb6c61d22d92c4c54e05e82766c488
|
| 3 |
+
size 998676480
|
model-00003-of-00007.safetensors → model-00003-of-00004.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58fe4d7ec793b5b115ef8bce4be123cb2b8a225b7ce52c75170663711983bb47
|
| 3 |
+
size 839198936
|
model-00004-of-00007.safetensors → model-00004-of-00004.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba98aab25de8765debd95803615a99f7194b0cb87c7fb1c5411650d2be8c173a
|
| 3 |
+
size 212992128
|
model-00005-of-00007.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:b19aa65336335092441eff2d6cebd4817aa5b01cbb19dd97f7ecbc4eb7e273c7
|
| 3 |
-
size 973463904
|
|
|
|
|
|
|
|
|
|
|
|
model-00006-of-00007.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:adf5cb621f40004551a01856021af47d6b16723f30ad0ecfe8427c878f7e1216
|
| 3 |
-
size 721713272
|
|
|
|
|
|
|
|
|
|
|
|
model-00007-of-00007.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:98e3392534465da28eeed210e9e615950277e08607e9d85829f33943220dfd6f
|
| 3 |
-
size 425984120
|
|
|
|
|
|
|
|
|
|
|
|
model.safetensors.index.json
CHANGED
|
@@ -1,445 +1,445 @@
|
|
| 1 |
{
|
| 2 |
"metadata": {
|
| 3 |
-
"total_size":
|
| 4 |
},
|
| 5 |
"weight_map": {
|
| 6 |
-
"head.weight": "model-
|
| 7 |
-
"rwkv.blocks.0.attention.key.weight": "model-00001-of-
|
| 8 |
-
"rwkv.blocks.0.attention.output.weight": "model-00001-of-
|
| 9 |
-
"rwkv.blocks.0.attention.receptance.weight": "model-00001-of-
|
| 10 |
-
"rwkv.blocks.0.attention.time_decay": "model-00001-of-
|
| 11 |
-
"rwkv.blocks.0.attention.time_first": "model-00001-of-
|
| 12 |
-
"rwkv.blocks.0.attention.time_mix_key": "model-00001-of-
|
| 13 |
-
"rwkv.blocks.0.attention.time_mix_receptance": "model-00001-of-
|
| 14 |
-
"rwkv.blocks.0.attention.time_mix_value": "model-00001-of-
|
| 15 |
-
"rwkv.blocks.0.attention.value.weight": "model-00001-of-
|
| 16 |
-
"rwkv.blocks.0.feed_forward.key.weight": "model-00001-of-
|
| 17 |
-
"rwkv.blocks.0.feed_forward.receptance.weight": "model-00001-of-
|
| 18 |
-
"rwkv.blocks.0.feed_forward.time_mix_key": "model-00001-of-
|
| 19 |
-
"rwkv.blocks.0.feed_forward.time_mix_receptance": "model-00001-of-
|
| 20 |
-
"rwkv.blocks.0.feed_forward.value.weight": "model-00001-of-
|
| 21 |
-
"rwkv.blocks.0.ln1.bias": "model-00001-of-
|
| 22 |
-
"rwkv.blocks.0.ln1.weight": "model-00001-of-
|
| 23 |
-
"rwkv.blocks.0.ln2.bias": "model-00001-of-
|
| 24 |
-
"rwkv.blocks.0.ln2.weight": "model-00001-of-
|
| 25 |
-
"rwkv.blocks.0.pre_ln.bias": "model-00001-of-
|
| 26 |
-
"rwkv.blocks.0.pre_ln.weight": "model-00001-of-
|
| 27 |
-
"rwkv.blocks.1.attention.key.weight": "model-00001-of-
|
| 28 |
-
"rwkv.blocks.1.attention.output.weight": "model-00001-of-
|
| 29 |
-
"rwkv.blocks.1.attention.receptance.weight": "model-00001-of-
|
| 30 |
-
"rwkv.blocks.1.attention.time_decay": "model-00001-of-
|
| 31 |
-
"rwkv.blocks.1.attention.time_first": "model-00001-of-
|
| 32 |
-
"rwkv.blocks.1.attention.time_mix_key": "model-00001-of-
|
| 33 |
-
"rwkv.blocks.1.attention.time_mix_receptance": "model-00001-of-
|
| 34 |
-
"rwkv.blocks.1.attention.time_mix_value": "model-00001-of-
|
| 35 |
-
"rwkv.blocks.1.attention.value.weight": "model-00001-of-
|
| 36 |
-
"rwkv.blocks.1.feed_forward.key.weight": "model-00001-of-
|
| 37 |
-
"rwkv.blocks.1.feed_forward.receptance.weight": "model-00001-of-
|
| 38 |
-
"rwkv.blocks.1.feed_forward.time_mix_key": "model-00001-of-
|
| 39 |
-
"rwkv.blocks.1.feed_forward.time_mix_receptance": "model-00001-of-
|
| 40 |
-
"rwkv.blocks.1.feed_forward.value.weight": "model-00001-of-
|
| 41 |
-
"rwkv.blocks.1.ln1.bias": "model-00001-of-
|
| 42 |
-
"rwkv.blocks.1.ln1.weight": "model-00001-of-
|
| 43 |
-
"rwkv.blocks.1.ln2.bias": "model-00001-of-
|
| 44 |
-
"rwkv.blocks.1.ln2.weight": "model-00001-of-
|
| 45 |
-
"rwkv.blocks.10.attention.key.weight": "model-
|
| 46 |
-
"rwkv.blocks.10.attention.output.weight": "model-
|
| 47 |
-
"rwkv.blocks.10.attention.receptance.weight": "model-
|
| 48 |
-
"rwkv.blocks.10.attention.time_decay": "model-
|
| 49 |
-
"rwkv.blocks.10.attention.time_first": "model-
|
| 50 |
-
"rwkv.blocks.10.attention.time_mix_key": "model-
|
| 51 |
-
"rwkv.blocks.10.attention.time_mix_receptance": "model-
|
| 52 |
-
"rwkv.blocks.10.attention.time_mix_value": "model-
|
| 53 |
-
"rwkv.blocks.10.attention.value.weight": "model-
|
| 54 |
-
"rwkv.blocks.10.feed_forward.key.weight": "model-
|
| 55 |
-
"rwkv.blocks.10.feed_forward.receptance.weight": "model-
|
| 56 |
-
"rwkv.blocks.10.feed_forward.time_mix_key": "model-
|
| 57 |
-
"rwkv.blocks.10.feed_forward.time_mix_receptance": "model-
|
| 58 |
-
"rwkv.blocks.10.feed_forward.value.weight": "model-
|
| 59 |
-
"rwkv.blocks.10.ln1.bias": "model-
|
| 60 |
-
"rwkv.blocks.10.ln1.weight": "model-
|
| 61 |
-
"rwkv.blocks.10.ln2.bias": "model-
|
| 62 |
-
"rwkv.blocks.10.ln2.weight": "model-
|
| 63 |
-
"rwkv.blocks.11.attention.key.weight": "model-
|
| 64 |
-
"rwkv.blocks.11.attention.output.weight": "model-
|
| 65 |
-
"rwkv.blocks.11.attention.receptance.weight": "model-
|
| 66 |
-
"rwkv.blocks.11.attention.time_decay": "model-
|
| 67 |
-
"rwkv.blocks.11.attention.time_first": "model-
|
| 68 |
-
"rwkv.blocks.11.attention.time_mix_key": "model-
|
| 69 |
-
"rwkv.blocks.11.attention.time_mix_receptance": "model-
|
| 70 |
-
"rwkv.blocks.11.attention.time_mix_value": "model-
|
| 71 |
-
"rwkv.blocks.11.attention.value.weight": "model-
|
| 72 |
-
"rwkv.blocks.11.feed_forward.key.weight": "model-
|
| 73 |
-
"rwkv.blocks.11.feed_forward.receptance.weight": "model-
|
| 74 |
-
"rwkv.blocks.11.feed_forward.time_mix_key": "model-
|
| 75 |
-
"rwkv.blocks.11.feed_forward.time_mix_receptance": "model-
|
| 76 |
-
"rwkv.blocks.11.feed_forward.value.weight": "model-
|
| 77 |
-
"rwkv.blocks.11.ln1.bias": "model-
|
| 78 |
-
"rwkv.blocks.11.ln1.weight": "model-
|
| 79 |
-
"rwkv.blocks.11.ln2.bias": "model-
|
| 80 |
-
"rwkv.blocks.11.ln2.weight": "model-
|
| 81 |
-
"rwkv.blocks.12.attention.key.weight": "model-
|
| 82 |
-
"rwkv.blocks.12.attention.output.weight": "model-
|
| 83 |
-
"rwkv.blocks.12.attention.receptance.weight": "model-
|
| 84 |
-
"rwkv.blocks.12.attention.time_decay": "model-
|
| 85 |
-
"rwkv.blocks.12.attention.time_first": "model-
|
| 86 |
-
"rwkv.blocks.12.attention.time_mix_key": "model-
|
| 87 |
-
"rwkv.blocks.12.attention.time_mix_receptance": "model-
|
| 88 |
-
"rwkv.blocks.12.attention.time_mix_value": "model-
|
| 89 |
-
"rwkv.blocks.12.attention.value.weight": "model-
|
| 90 |
-
"rwkv.blocks.12.feed_forward.key.weight": "model-
|
| 91 |
-
"rwkv.blocks.12.feed_forward.receptance.weight": "model-
|
| 92 |
-
"rwkv.blocks.12.feed_forward.time_mix_key": "model-
|
| 93 |
-
"rwkv.blocks.12.feed_forward.time_mix_receptance": "model-
|
| 94 |
-
"rwkv.blocks.12.feed_forward.value.weight": "model-
|
| 95 |
-
"rwkv.blocks.12.ln1.bias": "model-
|
| 96 |
-
"rwkv.blocks.12.ln1.weight": "model-
|
| 97 |
-
"rwkv.blocks.12.ln2.bias": "model-
|
| 98 |
-
"rwkv.blocks.12.ln2.weight": "model-
|
| 99 |
-
"rwkv.blocks.13.attention.key.weight": "model-
|
| 100 |
-
"rwkv.blocks.13.attention.output.weight": "model-
|
| 101 |
-
"rwkv.blocks.13.attention.receptance.weight": "model-
|
| 102 |
-
"rwkv.blocks.13.attention.time_decay": "model-
|
| 103 |
-
"rwkv.blocks.13.attention.time_first": "model-
|
| 104 |
-
"rwkv.blocks.13.attention.time_mix_key": "model-
|
| 105 |
-
"rwkv.blocks.13.attention.time_mix_receptance": "model-
|
| 106 |
-
"rwkv.blocks.13.attention.time_mix_value": "model-
|
| 107 |
-
"rwkv.blocks.13.attention.value.weight": "model-
|
| 108 |
-
"rwkv.blocks.13.feed_forward.key.weight": "model-
|
| 109 |
-
"rwkv.blocks.13.feed_forward.receptance.weight": "model-
|
| 110 |
-
"rwkv.blocks.13.feed_forward.time_mix_key": "model-
|
| 111 |
-
"rwkv.blocks.13.feed_forward.time_mix_receptance": "model-
|
| 112 |
-
"rwkv.blocks.13.feed_forward.value.weight": "model-
|
| 113 |
-
"rwkv.blocks.13.ln1.bias": "model-
|
| 114 |
-
"rwkv.blocks.13.ln1.weight": "model-
|
| 115 |
-
"rwkv.blocks.13.ln2.bias": "model-
|
| 116 |
-
"rwkv.blocks.13.ln2.weight": "model-
|
| 117 |
-
"rwkv.blocks.14.attention.key.weight": "model-
|
| 118 |
-
"rwkv.blocks.14.attention.output.weight": "model-
|
| 119 |
-
"rwkv.blocks.14.attention.receptance.weight": "model-
|
| 120 |
-
"rwkv.blocks.14.attention.time_decay": "model-
|
| 121 |
-
"rwkv.blocks.14.attention.time_first": "model-
|
| 122 |
-
"rwkv.blocks.14.attention.time_mix_key": "model-
|
| 123 |
-
"rwkv.blocks.14.attention.time_mix_receptance": "model-
|
| 124 |
-
"rwkv.blocks.14.attention.time_mix_value": "model-
|
| 125 |
-
"rwkv.blocks.14.attention.value.weight": "model-
|
| 126 |
-
"rwkv.blocks.14.feed_forward.key.weight": "model-
|
| 127 |
-
"rwkv.blocks.14.feed_forward.receptance.weight": "model-
|
| 128 |
-
"rwkv.blocks.14.feed_forward.time_mix_key": "model-
|
| 129 |
-
"rwkv.blocks.14.feed_forward.time_mix_receptance": "model-
|
| 130 |
-
"rwkv.blocks.14.feed_forward.value.weight": "model-
|
| 131 |
-
"rwkv.blocks.14.ln1.bias": "model-
|
| 132 |
-
"rwkv.blocks.14.ln1.weight": "model-
|
| 133 |
-
"rwkv.blocks.14.ln2.bias": "model-
|
| 134 |
-
"rwkv.blocks.14.ln2.weight": "model-
|
| 135 |
-
"rwkv.blocks.15.attention.key.weight": "model-
|
| 136 |
-
"rwkv.blocks.15.attention.output.weight": "model-
|
| 137 |
-
"rwkv.blocks.15.attention.receptance.weight": "model-
|
| 138 |
-
"rwkv.blocks.15.attention.time_decay": "model-
|
| 139 |
-
"rwkv.blocks.15.attention.time_first": "model-
|
| 140 |
-
"rwkv.blocks.15.attention.time_mix_key": "model-
|
| 141 |
-
"rwkv.blocks.15.attention.time_mix_receptance": "model-
|
| 142 |
-
"rwkv.blocks.15.attention.time_mix_value": "model-
|
| 143 |
-
"rwkv.blocks.15.attention.value.weight": "model-
|
| 144 |
-
"rwkv.blocks.15.feed_forward.key.weight": "model-
|
| 145 |
-
"rwkv.blocks.15.feed_forward.receptance.weight": "model-
|
| 146 |
-
"rwkv.blocks.15.feed_forward.time_mix_key": "model-
|
| 147 |
-
"rwkv.blocks.15.feed_forward.time_mix_receptance": "model-
|
| 148 |
-
"rwkv.blocks.15.feed_forward.value.weight": "model-
|
| 149 |
-
"rwkv.blocks.15.ln1.bias": "model-
|
| 150 |
-
"rwkv.blocks.15.ln1.weight": "model-
|
| 151 |
-
"rwkv.blocks.15.ln2.bias": "model-
|
| 152 |
-
"rwkv.blocks.15.ln2.weight": "model-
|
| 153 |
-
"rwkv.blocks.16.attention.key.weight": "model-
|
| 154 |
-
"rwkv.blocks.16.attention.output.weight": "model-
|
| 155 |
-
"rwkv.blocks.16.attention.receptance.weight": "model-
|
| 156 |
-
"rwkv.blocks.16.attention.time_decay": "model-
|
| 157 |
-
"rwkv.blocks.16.attention.time_first": "model-
|
| 158 |
-
"rwkv.blocks.16.attention.time_mix_key": "model-
|
| 159 |
-
"rwkv.blocks.16.attention.time_mix_receptance": "model-
|
| 160 |
-
"rwkv.blocks.16.attention.time_mix_value": "model-
|
| 161 |
-
"rwkv.blocks.16.attention.value.weight": "model-
|
| 162 |
-
"rwkv.blocks.16.feed_forward.key.weight": "model-
|
| 163 |
-
"rwkv.blocks.16.feed_forward.receptance.weight": "model-
|
| 164 |
-
"rwkv.blocks.16.feed_forward.time_mix_key": "model-
|
| 165 |
-
"rwkv.blocks.16.feed_forward.time_mix_receptance": "model-
|
| 166 |
-
"rwkv.blocks.16.feed_forward.value.weight": "model-
|
| 167 |
-
"rwkv.blocks.16.ln1.bias": "model-
|
| 168 |
-
"rwkv.blocks.16.ln1.weight": "model-
|
| 169 |
-
"rwkv.blocks.16.ln2.bias": "model-
|
| 170 |
-
"rwkv.blocks.16.ln2.weight": "model-
|
| 171 |
-
"rwkv.blocks.17.attention.key.weight": "model-
|
| 172 |
-
"rwkv.blocks.17.attention.output.weight": "model-
|
| 173 |
-
"rwkv.blocks.17.attention.receptance.weight": "model-
|
| 174 |
-
"rwkv.blocks.17.attention.time_decay": "model-
|
| 175 |
-
"rwkv.blocks.17.attention.time_first": "model-
|
| 176 |
-
"rwkv.blocks.17.attention.time_mix_key": "model-
|
| 177 |
-
"rwkv.blocks.17.attention.time_mix_receptance": "model-
|
| 178 |
-
"rwkv.blocks.17.attention.time_mix_value": "model-
|
| 179 |
-
"rwkv.blocks.17.attention.value.weight": "model-
|
| 180 |
-
"rwkv.blocks.17.feed_forward.key.weight": "model-
|
| 181 |
-
"rwkv.blocks.17.feed_forward.receptance.weight": "model-
|
| 182 |
-
"rwkv.blocks.17.feed_forward.time_mix_key": "model-
|
| 183 |
-
"rwkv.blocks.17.feed_forward.time_mix_receptance": "model-
|
| 184 |
-
"rwkv.blocks.17.feed_forward.value.weight": "model-
|
| 185 |
-
"rwkv.blocks.17.ln1.bias": "model-
|
| 186 |
-
"rwkv.blocks.17.ln1.weight": "model-
|
| 187 |
-
"rwkv.blocks.17.ln2.bias": "model-
|
| 188 |
-
"rwkv.blocks.17.ln2.weight": "model-
|
| 189 |
-
"rwkv.blocks.18.attention.key.weight": "model-
|
| 190 |
-
"rwkv.blocks.18.attention.output.weight": "model-
|
| 191 |
-
"rwkv.blocks.18.attention.receptance.weight": "model-
|
| 192 |
-
"rwkv.blocks.18.attention.time_decay": "model-
|
| 193 |
-
"rwkv.blocks.18.attention.time_first": "model-
|
| 194 |
-
"rwkv.blocks.18.attention.time_mix_key": "model-
|
| 195 |
-
"rwkv.blocks.18.attention.time_mix_receptance": "model-
|
| 196 |
-
"rwkv.blocks.18.attention.time_mix_value": "model-
|
| 197 |
-
"rwkv.blocks.18.attention.value.weight": "model-
|
| 198 |
-
"rwkv.blocks.18.feed_forward.key.weight": "model-
|
| 199 |
-
"rwkv.blocks.18.feed_forward.receptance.weight": "model-
|
| 200 |
-
"rwkv.blocks.18.feed_forward.time_mix_key": "model-
|
| 201 |
-
"rwkv.blocks.18.feed_forward.time_mix_receptance": "model-
|
| 202 |
-
"rwkv.blocks.18.feed_forward.value.weight": "model-
|
| 203 |
-
"rwkv.blocks.18.ln1.bias": "model-
|
| 204 |
-
"rwkv.blocks.18.ln1.weight": "model-
|
| 205 |
-
"rwkv.blocks.18.ln2.bias": "model-
|
| 206 |
-
"rwkv.blocks.18.ln2.weight": "model-
|
| 207 |
-
"rwkv.blocks.19.attention.key.weight": "model-
|
| 208 |
-
"rwkv.blocks.19.attention.output.weight": "model-
|
| 209 |
-
"rwkv.blocks.19.attention.receptance.weight": "model-
|
| 210 |
-
"rwkv.blocks.19.attention.time_decay": "model-
|
| 211 |
-
"rwkv.blocks.19.attention.time_first": "model-
|
| 212 |
-
"rwkv.blocks.19.attention.time_mix_key": "model-
|
| 213 |
-
"rwkv.blocks.19.attention.time_mix_receptance": "model-
|
| 214 |
-
"rwkv.blocks.19.attention.time_mix_value": "model-
|
| 215 |
-
"rwkv.blocks.19.attention.value.weight": "model-
|
| 216 |
-
"rwkv.blocks.19.feed_forward.key.weight": "model-
|
| 217 |
-
"rwkv.blocks.19.feed_forward.receptance.weight": "model-
|
| 218 |
-
"rwkv.blocks.19.feed_forward.time_mix_key": "model-
|
| 219 |
-
"rwkv.blocks.19.feed_forward.time_mix_receptance": "model-
|
| 220 |
-
"rwkv.blocks.19.feed_forward.value.weight": "model-
|
| 221 |
-
"rwkv.blocks.19.ln1.bias": "model-
|
| 222 |
-
"rwkv.blocks.19.ln1.weight": "model-
|
| 223 |
-
"rwkv.blocks.19.ln2.bias": "model-
|
| 224 |
-
"rwkv.blocks.19.ln2.weight": "model-
|
| 225 |
-
"rwkv.blocks.2.attention.key.weight": "model-00001-of-
|
| 226 |
-
"rwkv.blocks.2.attention.output.weight": "model-00001-of-
|
| 227 |
-
"rwkv.blocks.2.attention.receptance.weight": "model-00001-of-
|
| 228 |
-
"rwkv.blocks.2.attention.time_decay": "model-00001-of-
|
| 229 |
-
"rwkv.blocks.2.attention.time_first": "model-00001-of-
|
| 230 |
-
"rwkv.blocks.2.attention.time_mix_key": "model-00001-of-
|
| 231 |
-
"rwkv.blocks.2.attention.time_mix_receptance": "model-00001-of-
|
| 232 |
-
"rwkv.blocks.2.attention.time_mix_value": "model-00001-of-
|
| 233 |
-
"rwkv.blocks.2.attention.value.weight": "model-00001-of-
|
| 234 |
-
"rwkv.blocks.2.feed_forward.key.weight": "model-00001-of-
|
| 235 |
-
"rwkv.blocks.2.feed_forward.receptance.weight": "model-
|
| 236 |
-
"rwkv.blocks.2.feed_forward.time_mix_key": "model-00001-of-
|
| 237 |
-
"rwkv.blocks.2.feed_forward.time_mix_receptance": "model-00001-of-
|
| 238 |
-
"rwkv.blocks.2.feed_forward.value.weight": "model-
|
| 239 |
-
"rwkv.blocks.2.ln1.bias": "model-00001-of-
|
| 240 |
-
"rwkv.blocks.2.ln1.weight": "model-00001-of-
|
| 241 |
-
"rwkv.blocks.2.ln2.bias": "model-00001-of-
|
| 242 |
-
"rwkv.blocks.2.ln2.weight": "model-00001-of-
|
| 243 |
-
"rwkv.blocks.20.attention.key.weight": "model-
|
| 244 |
-
"rwkv.blocks.20.attention.output.weight": "model-
|
| 245 |
-
"rwkv.blocks.20.attention.receptance.weight": "model-
|
| 246 |
-
"rwkv.blocks.20.attention.time_decay": "model-
|
| 247 |
-
"rwkv.blocks.20.attention.time_first": "model-
|
| 248 |
-
"rwkv.blocks.20.attention.time_mix_key": "model-
|
| 249 |
-
"rwkv.blocks.20.attention.time_mix_receptance": "model-
|
| 250 |
-
"rwkv.blocks.20.attention.time_mix_value": "model-
|
| 251 |
-
"rwkv.blocks.20.attention.value.weight": "model-
|
| 252 |
-
"rwkv.blocks.20.feed_forward.key.weight": "model-
|
| 253 |
-
"rwkv.blocks.20.feed_forward.receptance.weight": "model-
|
| 254 |
-
"rwkv.blocks.20.feed_forward.time_mix_key": "model-
|
| 255 |
-
"rwkv.blocks.20.feed_forward.time_mix_receptance": "model-
|
| 256 |
-
"rwkv.blocks.20.feed_forward.value.weight": "model-
|
| 257 |
-
"rwkv.blocks.20.ln1.bias": "model-
|
| 258 |
-
"rwkv.blocks.20.ln1.weight": "model-
|
| 259 |
-
"rwkv.blocks.20.ln2.bias": "model-
|
| 260 |
-
"rwkv.blocks.20.ln2.weight": "model-
|
| 261 |
-
"rwkv.blocks.21.attention.key.weight": "model-
|
| 262 |
-
"rwkv.blocks.21.attention.output.weight": "model-
|
| 263 |
-
"rwkv.blocks.21.attention.receptance.weight": "model-
|
| 264 |
-
"rwkv.blocks.21.attention.time_decay": "model-
|
| 265 |
-
"rwkv.blocks.21.attention.time_first": "model-
|
| 266 |
-
"rwkv.blocks.21.attention.time_mix_key": "model-
|
| 267 |
-
"rwkv.blocks.21.attention.time_mix_receptance": "model-
|
| 268 |
-
"rwkv.blocks.21.attention.time_mix_value": "model-
|
| 269 |
-
"rwkv.blocks.21.attention.value.weight": "model-
|
| 270 |
-
"rwkv.blocks.21.feed_forward.key.weight": "model-
|
| 271 |
-
"rwkv.blocks.21.feed_forward.receptance.weight": "model-
|
| 272 |
-
"rwkv.blocks.21.feed_forward.time_mix_key": "model-
|
| 273 |
-
"rwkv.blocks.21.feed_forward.time_mix_receptance": "model-
|
| 274 |
-
"rwkv.blocks.21.feed_forward.value.weight": "model-
|
| 275 |
-
"rwkv.blocks.21.ln1.bias": "model-
|
| 276 |
-
"rwkv.blocks.21.ln1.weight": "model-
|
| 277 |
-
"rwkv.blocks.21.ln2.bias": "model-
|
| 278 |
-
"rwkv.blocks.21.ln2.weight": "model-
|
| 279 |
-
"rwkv.blocks.22.attention.key.weight": "model-
|
| 280 |
-
"rwkv.blocks.22.attention.output.weight": "model-
|
| 281 |
-
"rwkv.blocks.22.attention.receptance.weight": "model-
|
| 282 |
-
"rwkv.blocks.22.attention.time_decay": "model-
|
| 283 |
-
"rwkv.blocks.22.attention.time_first": "model-
|
| 284 |
-
"rwkv.blocks.22.attention.time_mix_key": "model-
|
| 285 |
-
"rwkv.blocks.22.attention.time_mix_receptance": "model-
|
| 286 |
-
"rwkv.blocks.22.attention.time_mix_value": "model-
|
| 287 |
-
"rwkv.blocks.22.attention.value.weight": "model-
|
| 288 |
-
"rwkv.blocks.22.feed_forward.key.weight": "model-
|
| 289 |
-
"rwkv.blocks.22.feed_forward.receptance.weight": "model-
|
| 290 |
-
"rwkv.blocks.22.feed_forward.time_mix_key": "model-
|
| 291 |
-
"rwkv.blocks.22.feed_forward.time_mix_receptance": "model-
|
| 292 |
-
"rwkv.blocks.22.feed_forward.value.weight": "model-
|
| 293 |
-
"rwkv.blocks.22.ln1.bias": "model-
|
| 294 |
-
"rwkv.blocks.22.ln1.weight": "model-
|
| 295 |
-
"rwkv.blocks.22.ln2.bias": "model-
|
| 296 |
-
"rwkv.blocks.22.ln2.weight": "model-
|
| 297 |
-
"rwkv.blocks.23.attention.key.weight": "model-
|
| 298 |
-
"rwkv.blocks.23.attention.output.weight": "model-
|
| 299 |
-
"rwkv.blocks.23.attention.receptance.weight": "model-
|
| 300 |
-
"rwkv.blocks.23.attention.time_decay": "model-
|
| 301 |
-
"rwkv.blocks.23.attention.time_first": "model-
|
| 302 |
-
"rwkv.blocks.23.attention.time_mix_key": "model-
|
| 303 |
-
"rwkv.blocks.23.attention.time_mix_receptance": "model-
|
| 304 |
-
"rwkv.blocks.23.attention.time_mix_value": "model-
|
| 305 |
-
"rwkv.blocks.23.attention.value.weight": "model-
|
| 306 |
-
"rwkv.blocks.23.feed_forward.key.weight": "model-
|
| 307 |
-
"rwkv.blocks.23.feed_forward.receptance.weight": "model-
|
| 308 |
-
"rwkv.blocks.23.feed_forward.time_mix_key": "model-
|
| 309 |
-
"rwkv.blocks.23.feed_forward.time_mix_receptance": "model-
|
| 310 |
-
"rwkv.blocks.23.feed_forward.value.weight": "model-
|
| 311 |
-
"rwkv.blocks.23.ln1.bias": "model-
|
| 312 |
-
"rwkv.blocks.23.ln1.weight": "model-
|
| 313 |
-
"rwkv.blocks.23.ln2.bias": "model-
|
| 314 |
-
"rwkv.blocks.23.ln2.weight": "model-
|
| 315 |
-
"rwkv.blocks.3.attention.key.weight": "model-
|
| 316 |
-
"rwkv.blocks.3.attention.output.weight": "model-
|
| 317 |
-
"rwkv.blocks.3.attention.receptance.weight": "model-
|
| 318 |
-
"rwkv.blocks.3.attention.time_decay": "model-
|
| 319 |
-
"rwkv.blocks.3.attention.time_first": "model-
|
| 320 |
-
"rwkv.blocks.3.attention.time_mix_key": "model-
|
| 321 |
-
"rwkv.blocks.3.attention.time_mix_receptance": "model-
|
| 322 |
-
"rwkv.blocks.3.attention.time_mix_value": "model-
|
| 323 |
-
"rwkv.blocks.3.attention.value.weight": "model-
|
| 324 |
-
"rwkv.blocks.3.feed_forward.key.weight": "model-
|
| 325 |
-
"rwkv.blocks.3.feed_forward.receptance.weight": "model-
|
| 326 |
-
"rwkv.blocks.3.feed_forward.time_mix_key": "model-
|
| 327 |
-
"rwkv.blocks.3.feed_forward.time_mix_receptance": "model-
|
| 328 |
-
"rwkv.blocks.3.feed_forward.value.weight": "model-
|
| 329 |
-
"rwkv.blocks.3.ln1.bias": "model-
|
| 330 |
-
"rwkv.blocks.3.ln1.weight": "model-
|
| 331 |
-
"rwkv.blocks.3.ln2.bias": "model-
|
| 332 |
-
"rwkv.blocks.3.ln2.weight": "model-
|
| 333 |
-
"rwkv.blocks.4.attention.key.weight": "model-
|
| 334 |
-
"rwkv.blocks.4.attention.output.weight": "model-
|
| 335 |
-
"rwkv.blocks.4.attention.receptance.weight": "model-
|
| 336 |
-
"rwkv.blocks.4.attention.time_decay": "model-
|
| 337 |
-
"rwkv.blocks.4.attention.time_first": "model-
|
| 338 |
-
"rwkv.blocks.4.attention.time_mix_key": "model-
|
| 339 |
-
"rwkv.blocks.4.attention.time_mix_receptance": "model-
|
| 340 |
-
"rwkv.blocks.4.attention.time_mix_value": "model-
|
| 341 |
-
"rwkv.blocks.4.attention.value.weight": "model-
|
| 342 |
-
"rwkv.blocks.4.feed_forward.key.weight": "model-
|
| 343 |
-
"rwkv.blocks.4.feed_forward.receptance.weight": "model-
|
| 344 |
-
"rwkv.blocks.4.feed_forward.time_mix_key": "model-
|
| 345 |
-
"rwkv.blocks.4.feed_forward.time_mix_receptance": "model-
|
| 346 |
-
"rwkv.blocks.4.feed_forward.value.weight": "model-
|
| 347 |
-
"rwkv.blocks.4.ln1.bias": "model-
|
| 348 |
-
"rwkv.blocks.4.ln1.weight": "model-
|
| 349 |
-
"rwkv.blocks.4.ln2.bias": "model-
|
| 350 |
-
"rwkv.blocks.4.ln2.weight": "model-
|
| 351 |
-
"rwkv.blocks.5.attention.key.weight": "model-
|
| 352 |
-
"rwkv.blocks.5.attention.output.weight": "model-
|
| 353 |
-
"rwkv.blocks.5.attention.receptance.weight": "model-
|
| 354 |
-
"rwkv.blocks.5.attention.time_decay": "model-
|
| 355 |
-
"rwkv.blocks.5.attention.time_first": "model-
|
| 356 |
-
"rwkv.blocks.5.attention.time_mix_key": "model-
|
| 357 |
-
"rwkv.blocks.5.attention.time_mix_receptance": "model-
|
| 358 |
-
"rwkv.blocks.5.attention.time_mix_value": "model-
|
| 359 |
-
"rwkv.blocks.5.attention.value.weight": "model-
|
| 360 |
-
"rwkv.blocks.5.feed_forward.key.weight": "model-
|
| 361 |
-
"rwkv.blocks.5.feed_forward.receptance.weight": "model-
|
| 362 |
-
"rwkv.blocks.5.feed_forward.time_mix_key": "model-
|
| 363 |
-
"rwkv.blocks.5.feed_forward.time_mix_receptance": "model-
|
| 364 |
-
"rwkv.blocks.5.feed_forward.value.weight": "model-
|
| 365 |
-
"rwkv.blocks.5.ln1.bias": "model-
|
| 366 |
-
"rwkv.blocks.5.ln1.weight": "model-
|
| 367 |
-
"rwkv.blocks.5.ln2.bias": "model-
|
| 368 |
-
"rwkv.blocks.5.ln2.weight": "model-
|
| 369 |
-
"rwkv.blocks.6.attention.key.weight": "model-
|
| 370 |
-
"rwkv.blocks.6.attention.output.weight": "model-
|
| 371 |
-
"rwkv.blocks.6.attention.receptance.weight": "model-
|
| 372 |
-
"rwkv.blocks.6.attention.time_decay": "model-
|
| 373 |
-
"rwkv.blocks.6.attention.time_first": "model-
|
| 374 |
-
"rwkv.blocks.6.attention.time_mix_key": "model-
|
| 375 |
-
"rwkv.blocks.6.attention.time_mix_receptance": "model-
|
| 376 |
-
"rwkv.blocks.6.attention.time_mix_value": "model-
|
| 377 |
-
"rwkv.blocks.6.attention.value.weight": "model-
|
| 378 |
-
"rwkv.blocks.6.feed_forward.key.weight": "model-
|
| 379 |
-
"rwkv.blocks.6.feed_forward.receptance.weight": "model-
|
| 380 |
-
"rwkv.blocks.6.feed_forward.time_mix_key": "model-
|
| 381 |
-
"rwkv.blocks.6.feed_forward.time_mix_receptance": "model-
|
| 382 |
-
"rwkv.blocks.6.feed_forward.value.weight": "model-
|
| 383 |
-
"rwkv.blocks.6.ln1.bias": "model-
|
| 384 |
-
"rwkv.blocks.6.ln1.weight": "model-
|
| 385 |
-
"rwkv.blocks.6.ln2.bias": "model-
|
| 386 |
-
"rwkv.blocks.6.ln2.weight": "model-
|
| 387 |
-
"rwkv.blocks.7.attention.key.weight": "model-
|
| 388 |
-
"rwkv.blocks.7.attention.output.weight": "model-
|
| 389 |
-
"rwkv.blocks.7.attention.receptance.weight": "model-
|
| 390 |
-
"rwkv.blocks.7.attention.time_decay": "model-
|
| 391 |
-
"rwkv.blocks.7.attention.time_first": "model-
|
| 392 |
-
"rwkv.blocks.7.attention.time_mix_key": "model-
|
| 393 |
-
"rwkv.blocks.7.attention.time_mix_receptance": "model-
|
| 394 |
-
"rwkv.blocks.7.attention.time_mix_value": "model-
|
| 395 |
-
"rwkv.blocks.7.attention.value.weight": "model-
|
| 396 |
-
"rwkv.blocks.7.feed_forward.key.weight": "model-
|
| 397 |
-
"rwkv.blocks.7.feed_forward.receptance.weight": "model-
|
| 398 |
-
"rwkv.blocks.7.feed_forward.time_mix_key": "model-
|
| 399 |
-
"rwkv.blocks.7.feed_forward.time_mix_receptance": "model-
|
| 400 |
-
"rwkv.blocks.7.feed_forward.value.weight": "model-
|
| 401 |
-
"rwkv.blocks.7.ln1.bias": "model-
|
| 402 |
-
"rwkv.blocks.7.ln1.weight": "model-
|
| 403 |
-
"rwkv.blocks.7.ln2.bias": "model-
|
| 404 |
-
"rwkv.blocks.7.ln2.weight": "model-
|
| 405 |
-
"rwkv.blocks.8.attention.key.weight": "model-
|
| 406 |
-
"rwkv.blocks.8.attention.output.weight": "model-
|
| 407 |
-
"rwkv.blocks.8.attention.receptance.weight": "model-
|
| 408 |
-
"rwkv.blocks.8.attention.time_decay": "model-
|
| 409 |
-
"rwkv.blocks.8.attention.time_first": "model-
|
| 410 |
-
"rwkv.blocks.8.attention.time_mix_key": "model-
|
| 411 |
-
"rwkv.blocks.8.attention.time_mix_receptance": "model-
|
| 412 |
-
"rwkv.blocks.8.attention.time_mix_value": "model-
|
| 413 |
-
"rwkv.blocks.8.attention.value.weight": "model-
|
| 414 |
-
"rwkv.blocks.8.feed_forward.key.weight": "model-
|
| 415 |
-
"rwkv.blocks.8.feed_forward.receptance.weight": "model-
|
| 416 |
-
"rwkv.blocks.8.feed_forward.time_mix_key": "model-
|
| 417 |
-
"rwkv.blocks.8.feed_forward.time_mix_receptance": "model-
|
| 418 |
-
"rwkv.blocks.8.feed_forward.value.weight": "model-
|
| 419 |
-
"rwkv.blocks.8.ln1.bias": "model-
|
| 420 |
-
"rwkv.blocks.8.ln1.weight": "model-
|
| 421 |
-
"rwkv.blocks.8.ln2.bias": "model-
|
| 422 |
-
"rwkv.blocks.8.ln2.weight": "model-
|
| 423 |
-
"rwkv.blocks.9.attention.key.weight": "model-
|
| 424 |
-
"rwkv.blocks.9.attention.output.weight": "model-
|
| 425 |
-
"rwkv.blocks.9.attention.receptance.weight": "model-
|
| 426 |
-
"rwkv.blocks.9.attention.time_decay": "model-
|
| 427 |
-
"rwkv.blocks.9.attention.time_first": "model-
|
| 428 |
-
"rwkv.blocks.9.attention.time_mix_key": "model-
|
| 429 |
-
"rwkv.blocks.9.attention.time_mix_receptance": "model-
|
| 430 |
-
"rwkv.blocks.9.attention.time_mix_value": "model-
|
| 431 |
-
"rwkv.blocks.9.attention.value.weight": "model-
|
| 432 |
-
"rwkv.blocks.9.feed_forward.key.weight": "model-
|
| 433 |
-
"rwkv.blocks.9.feed_forward.receptance.weight": "model-
|
| 434 |
-
"rwkv.blocks.9.feed_forward.time_mix_key": "model-
|
| 435 |
-
"rwkv.blocks.9.feed_forward.time_mix_receptance": "model-
|
| 436 |
-
"rwkv.blocks.9.feed_forward.value.weight": "model-
|
| 437 |
-
"rwkv.blocks.9.ln1.bias": "model-
|
| 438 |
-
"rwkv.blocks.9.ln1.weight": "model-
|
| 439 |
-
"rwkv.blocks.9.ln2.bias": "model-
|
| 440 |
-
"rwkv.blocks.9.ln2.weight": "model-
|
| 441 |
-
"rwkv.embeddings.weight": "model-00001-of-
|
| 442 |
-
"rwkv.ln_out.bias": "model-
|
| 443 |
-
"rwkv.ln_out.weight": "model-
|
| 444 |
}
|
| 445 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"metadata": {
|
| 3 |
+
"total_size": 3044327424
|
| 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-00002-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 |
}
|
pytorch_model-00001-of-00004.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:03e970f7e8438f0fe5d313e676a0776aa6b6fc4ed164a1e8aa8c5be46130da5b
|
| 3 |
+
size 993539599
|
pytorch_model-00001-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d9824593ff175045fd9410158a8f587765db31a5893c8546f0b090767d91155c
|
| 3 |
-
size 996714221
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00002-of-00004.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a154d24496532d11ffbb2509e6c49cc9926d8d0621792bb18d0fe1234237fcdd
|
| 3 |
+
size 998713597
|
pytorch_model-00002-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:6c77bdb67063627181ed54ba5467abc578990fe9fe0d88d7cc2fe1f14495966f
|
| 3 |
-
size 990318153
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00003-of-00004.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec369f44ab282fdb329e1203a2a28315c815dbb0de4a98685e5dc8e689d0bcac
|
| 3 |
+
size 839228132
|
pytorch_model-00003-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:48be7452f587c9741b27ef6d49559ccce7a5f587a80408434dae87db407dcee3
|
| 3 |
-
size 990258571
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00004-of-00004.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ab3c128569d82aa3d712e818aacd06ad653f94a27f18dfdb83e4f0c6bfb78ecf
|
| 3 |
+
size 212992938
|
pytorch_model-00004-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:070c3594c14eaf9ab1bcf323bbc499e60208c7725ec1b5b25e63c86b3d1fd210
|
| 3 |
-
size 990318217
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00005-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:99585dba978c45c3e4828378f9d521bd04bb2d69e48d2949622fa4b9b8638f5d
|
| 3 |
-
size 973481081
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00006-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:6e3a421829b36606f823ab56eb80e9378e22f2c5172713382fd025a96bcdde08
|
| 3 |
-
size 721725905
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model-00007-of-00007.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:04b5da856bf56f981bc4d30072649ef0beee33d54d6003118971a47fd2dea2d4
|
| 3 |
-
size 425984938
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model.bin.index.json
CHANGED
|
@@ -1,445 +1,445 @@
|
|
| 1 |
{
|
| 2 |
"metadata": {
|
| 3 |
-
"total_size":
|
| 4 |
},
|
| 5 |
"weight_map": {
|
| 6 |
-
"head.weight": "pytorch_model-
|
| 7 |
-
"rwkv.blocks.0.attention.key.weight": "pytorch_model-00001-of-
|
| 8 |
-
"rwkv.blocks.0.attention.output.weight": "pytorch_model-00001-of-
|
| 9 |
-
"rwkv.blocks.0.attention.receptance.weight": "pytorch_model-00001-of-
|
| 10 |
-
"rwkv.blocks.0.attention.time_decay": "pytorch_model-00001-of-
|
| 11 |
-
"rwkv.blocks.0.attention.time_first": "pytorch_model-00001-of-
|
| 12 |
-
"rwkv.blocks.0.attention.time_mix_key": "pytorch_model-00001-of-
|
| 13 |
-
"rwkv.blocks.0.attention.time_mix_receptance": "pytorch_model-00001-of-
|
| 14 |
-
"rwkv.blocks.0.attention.time_mix_value": "pytorch_model-00001-of-
|
| 15 |
-
"rwkv.blocks.0.attention.value.weight": "pytorch_model-00001-of-
|
| 16 |
-
"rwkv.blocks.0.feed_forward.key.weight": "pytorch_model-00001-of-
|
| 17 |
-
"rwkv.blocks.0.feed_forward.receptance.weight": "pytorch_model-00001-of-
|
| 18 |
-
"rwkv.blocks.0.feed_forward.time_mix_key": "pytorch_model-00001-of-
|
| 19 |
-
"rwkv.blocks.0.feed_forward.time_mix_receptance": "pytorch_model-00001-of-
|
| 20 |
-
"rwkv.blocks.0.feed_forward.value.weight": "pytorch_model-00001-of-
|
| 21 |
-
"rwkv.blocks.0.ln1.bias": "pytorch_model-00001-of-
|
| 22 |
-
"rwkv.blocks.0.ln1.weight": "pytorch_model-00001-of-
|
| 23 |
-
"rwkv.blocks.0.ln2.bias": "pytorch_model-00001-of-
|
| 24 |
-
"rwkv.blocks.0.ln2.weight": "pytorch_model-00001-of-
|
| 25 |
-
"rwkv.blocks.0.pre_ln.bias": "pytorch_model-00001-of-
|
| 26 |
-
"rwkv.blocks.0.pre_ln.weight": "pytorch_model-00001-of-
|
| 27 |
-
"rwkv.blocks.1.attention.key.weight": "pytorch_model-00001-of-
|
| 28 |
-
"rwkv.blocks.1.attention.output.weight": "pytorch_model-00001-of-
|
| 29 |
-
"rwkv.blocks.1.attention.receptance.weight": "pytorch_model-00001-of-
|
| 30 |
-
"rwkv.blocks.1.attention.time_decay": "pytorch_model-00001-of-
|
| 31 |
-
"rwkv.blocks.1.attention.time_first": "pytorch_model-00001-of-
|
| 32 |
-
"rwkv.blocks.1.attention.time_mix_key": "pytorch_model-00001-of-
|
| 33 |
-
"rwkv.blocks.1.attention.time_mix_receptance": "pytorch_model-00001-of-
|
| 34 |
-
"rwkv.blocks.1.attention.time_mix_value": "pytorch_model-00001-of-
|
| 35 |
-
"rwkv.blocks.1.attention.value.weight": "pytorch_model-00001-of-
|
| 36 |
-
"rwkv.blocks.1.feed_forward.key.weight": "pytorch_model-00001-of-
|
| 37 |
-
"rwkv.blocks.1.feed_forward.receptance.weight": "pytorch_model-00001-of-
|
| 38 |
-
"rwkv.blocks.1.feed_forward.time_mix_key": "pytorch_model-00001-of-
|
| 39 |
-
"rwkv.blocks.1.feed_forward.time_mix_receptance": "pytorch_model-00001-of-
|
| 40 |
-
"rwkv.blocks.1.feed_forward.value.weight": "pytorch_model-00001-of-
|
| 41 |
-
"rwkv.blocks.1.ln1.bias": "pytorch_model-00001-of-
|
| 42 |
-
"rwkv.blocks.1.ln1.weight": "pytorch_model-00001-of-
|
| 43 |
-
"rwkv.blocks.1.ln2.bias": "pytorch_model-00001-of-
|
| 44 |
-
"rwkv.blocks.1.ln2.weight": "pytorch_model-00001-of-
|
| 45 |
-
"rwkv.blocks.10.attention.key.weight": "pytorch_model-
|
| 46 |
-
"rwkv.blocks.10.attention.output.weight": "pytorch_model-
|
| 47 |
-
"rwkv.blocks.10.attention.receptance.weight": "pytorch_model-
|
| 48 |
-
"rwkv.blocks.10.attention.time_decay": "pytorch_model-
|
| 49 |
-
"rwkv.blocks.10.attention.time_first": "pytorch_model-
|
| 50 |
-
"rwkv.blocks.10.attention.time_mix_key": "pytorch_model-
|
| 51 |
-
"rwkv.blocks.10.attention.time_mix_receptance": "pytorch_model-
|
| 52 |
-
"rwkv.blocks.10.attention.time_mix_value": "pytorch_model-
|
| 53 |
-
"rwkv.blocks.10.attention.value.weight": "pytorch_model-
|
| 54 |
-
"rwkv.blocks.10.feed_forward.key.weight": "pytorch_model-
|
| 55 |
-
"rwkv.blocks.10.feed_forward.receptance.weight": "pytorch_model-
|
| 56 |
-
"rwkv.blocks.10.feed_forward.time_mix_key": "pytorch_model-
|
| 57 |
-
"rwkv.blocks.10.feed_forward.time_mix_receptance": "pytorch_model-
|
| 58 |
-
"rwkv.blocks.10.feed_forward.value.weight": "pytorch_model-
|
| 59 |
-
"rwkv.blocks.10.ln1.bias": "pytorch_model-
|
| 60 |
-
"rwkv.blocks.10.ln1.weight": "pytorch_model-
|
| 61 |
-
"rwkv.blocks.10.ln2.bias": "pytorch_model-
|
| 62 |
-
"rwkv.blocks.10.ln2.weight": "pytorch_model-
|
| 63 |
-
"rwkv.blocks.11.attention.key.weight": "pytorch_model-
|
| 64 |
-
"rwkv.blocks.11.attention.output.weight": "pytorch_model-
|
| 65 |
-
"rwkv.blocks.11.attention.receptance.weight": "pytorch_model-
|
| 66 |
-
"rwkv.blocks.11.attention.time_decay": "pytorch_model-
|
| 67 |
-
"rwkv.blocks.11.attention.time_first": "pytorch_model-
|
| 68 |
-
"rwkv.blocks.11.attention.time_mix_key": "pytorch_model-
|
| 69 |
-
"rwkv.blocks.11.attention.time_mix_receptance": "pytorch_model-
|
| 70 |
-
"rwkv.blocks.11.attention.time_mix_value": "pytorch_model-
|
| 71 |
-
"rwkv.blocks.11.attention.value.weight": "pytorch_model-
|
| 72 |
-
"rwkv.blocks.11.feed_forward.key.weight": "pytorch_model-
|
| 73 |
-
"rwkv.blocks.11.feed_forward.receptance.weight": "pytorch_model-
|
| 74 |
-
"rwkv.blocks.11.feed_forward.time_mix_key": "pytorch_model-
|
| 75 |
-
"rwkv.blocks.11.feed_forward.time_mix_receptance": "pytorch_model-
|
| 76 |
-
"rwkv.blocks.11.feed_forward.value.weight": "pytorch_model-
|
| 77 |
-
"rwkv.blocks.11.ln1.bias": "pytorch_model-
|
| 78 |
-
"rwkv.blocks.11.ln1.weight": "pytorch_model-
|
| 79 |
-
"rwkv.blocks.11.ln2.bias": "pytorch_model-
|
| 80 |
-
"rwkv.blocks.11.ln2.weight": "pytorch_model-
|
| 81 |
-
"rwkv.blocks.12.attention.key.weight": "pytorch_model-
|
| 82 |
-
"rwkv.blocks.12.attention.output.weight": "pytorch_model-
|
| 83 |
-
"rwkv.blocks.12.attention.receptance.weight": "pytorch_model-
|
| 84 |
-
"rwkv.blocks.12.attention.time_decay": "pytorch_model-
|
| 85 |
-
"rwkv.blocks.12.attention.time_first": "pytorch_model-
|
| 86 |
-
"rwkv.blocks.12.attention.time_mix_key": "pytorch_model-
|
| 87 |
-
"rwkv.blocks.12.attention.time_mix_receptance": "pytorch_model-
|
| 88 |
-
"rwkv.blocks.12.attention.time_mix_value": "pytorch_model-
|
| 89 |
-
"rwkv.blocks.12.attention.value.weight": "pytorch_model-
|
| 90 |
-
"rwkv.blocks.12.feed_forward.key.weight": "pytorch_model-
|
| 91 |
-
"rwkv.blocks.12.feed_forward.receptance.weight": "pytorch_model-
|
| 92 |
-
"rwkv.blocks.12.feed_forward.time_mix_key": "pytorch_model-
|
| 93 |
-
"rwkv.blocks.12.feed_forward.time_mix_receptance": "pytorch_model-
|
| 94 |
-
"rwkv.blocks.12.feed_forward.value.weight": "pytorch_model-
|
| 95 |
-
"rwkv.blocks.12.ln1.bias": "pytorch_model-
|
| 96 |
-
"rwkv.blocks.12.ln1.weight": "pytorch_model-
|
| 97 |
-
"rwkv.blocks.12.ln2.bias": "pytorch_model-
|
| 98 |
-
"rwkv.blocks.12.ln2.weight": "pytorch_model-
|
| 99 |
-
"rwkv.blocks.13.attention.key.weight": "pytorch_model-
|
| 100 |
-
"rwkv.blocks.13.attention.output.weight": "pytorch_model-
|
| 101 |
-
"rwkv.blocks.13.attention.receptance.weight": "pytorch_model-
|
| 102 |
-
"rwkv.blocks.13.attention.time_decay": "pytorch_model-
|
| 103 |
-
"rwkv.blocks.13.attention.time_first": "pytorch_model-
|
| 104 |
-
"rwkv.blocks.13.attention.time_mix_key": "pytorch_model-
|
| 105 |
-
"rwkv.blocks.13.attention.time_mix_receptance": "pytorch_model-
|
| 106 |
-
"rwkv.blocks.13.attention.time_mix_value": "pytorch_model-
|
| 107 |
-
"rwkv.blocks.13.attention.value.weight": "pytorch_model-
|
| 108 |
-
"rwkv.blocks.13.feed_forward.key.weight": "pytorch_model-
|
| 109 |
-
"rwkv.blocks.13.feed_forward.receptance.weight": "pytorch_model-
|
| 110 |
-
"rwkv.blocks.13.feed_forward.time_mix_key": "pytorch_model-
|
| 111 |
-
"rwkv.blocks.13.feed_forward.time_mix_receptance": "pytorch_model-
|
| 112 |
-
"rwkv.blocks.13.feed_forward.value.weight": "pytorch_model-
|
| 113 |
-
"rwkv.blocks.13.ln1.bias": "pytorch_model-
|
| 114 |
-
"rwkv.blocks.13.ln1.weight": "pytorch_model-
|
| 115 |
-
"rwkv.blocks.13.ln2.bias": "pytorch_model-
|
| 116 |
-
"rwkv.blocks.13.ln2.weight": "pytorch_model-
|
| 117 |
-
"rwkv.blocks.14.attention.key.weight": "pytorch_model-
|
| 118 |
-
"rwkv.blocks.14.attention.output.weight": "pytorch_model-
|
| 119 |
-
"rwkv.blocks.14.attention.receptance.weight": "pytorch_model-
|
| 120 |
-
"rwkv.blocks.14.attention.time_decay": "pytorch_model-
|
| 121 |
-
"rwkv.blocks.14.attention.time_first": "pytorch_model-
|
| 122 |
-
"rwkv.blocks.14.attention.time_mix_key": "pytorch_model-
|
| 123 |
-
"rwkv.blocks.14.attention.time_mix_receptance": "pytorch_model-
|
| 124 |
-
"rwkv.blocks.14.attention.time_mix_value": "pytorch_model-
|
| 125 |
-
"rwkv.blocks.14.attention.value.weight": "pytorch_model-
|
| 126 |
-
"rwkv.blocks.14.feed_forward.key.weight": "pytorch_model-
|
| 127 |
-
"rwkv.blocks.14.feed_forward.receptance.weight": "pytorch_model-
|
| 128 |
-
"rwkv.blocks.14.feed_forward.time_mix_key": "pytorch_model-
|
| 129 |
-
"rwkv.blocks.14.feed_forward.time_mix_receptance": "pytorch_model-
|
| 130 |
-
"rwkv.blocks.14.feed_forward.value.weight": "pytorch_model-
|
| 131 |
-
"rwkv.blocks.14.ln1.bias": "pytorch_model-
|
| 132 |
-
"rwkv.blocks.14.ln1.weight": "pytorch_model-
|
| 133 |
-
"rwkv.blocks.14.ln2.bias": "pytorch_model-
|
| 134 |
-
"rwkv.blocks.14.ln2.weight": "pytorch_model-
|
| 135 |
-
"rwkv.blocks.15.attention.key.weight": "pytorch_model-
|
| 136 |
-
"rwkv.blocks.15.attention.output.weight": "pytorch_model-
|
| 137 |
-
"rwkv.blocks.15.attention.receptance.weight": "pytorch_model-
|
| 138 |
-
"rwkv.blocks.15.attention.time_decay": "pytorch_model-
|
| 139 |
-
"rwkv.blocks.15.attention.time_first": "pytorch_model-
|
| 140 |
-
"rwkv.blocks.15.attention.time_mix_key": "pytorch_model-
|
| 141 |
-
"rwkv.blocks.15.attention.time_mix_receptance": "pytorch_model-
|
| 142 |
-
"rwkv.blocks.15.attention.time_mix_value": "pytorch_model-
|
| 143 |
-
"rwkv.blocks.15.attention.value.weight": "pytorch_model-
|
| 144 |
-
"rwkv.blocks.15.feed_forward.key.weight": "pytorch_model-
|
| 145 |
-
"rwkv.blocks.15.feed_forward.receptance.weight": "pytorch_model-
|
| 146 |
-
"rwkv.blocks.15.feed_forward.time_mix_key": "pytorch_model-
|
| 147 |
-
"rwkv.blocks.15.feed_forward.time_mix_receptance": "pytorch_model-
|
| 148 |
-
"rwkv.blocks.15.feed_forward.value.weight": "pytorch_model-
|
| 149 |
-
"rwkv.blocks.15.ln1.bias": "pytorch_model-
|
| 150 |
-
"rwkv.blocks.15.ln1.weight": "pytorch_model-
|
| 151 |
-
"rwkv.blocks.15.ln2.bias": "pytorch_model-
|
| 152 |
-
"rwkv.blocks.15.ln2.weight": "pytorch_model-
|
| 153 |
-
"rwkv.blocks.16.attention.key.weight": "pytorch_model-
|
| 154 |
-
"rwkv.blocks.16.attention.output.weight": "pytorch_model-
|
| 155 |
-
"rwkv.blocks.16.attention.receptance.weight": "pytorch_model-
|
| 156 |
-
"rwkv.blocks.16.attention.time_decay": "pytorch_model-
|
| 157 |
-
"rwkv.blocks.16.attention.time_first": "pytorch_model-
|
| 158 |
-
"rwkv.blocks.16.attention.time_mix_key": "pytorch_model-
|
| 159 |
-
"rwkv.blocks.16.attention.time_mix_receptance": "pytorch_model-
|
| 160 |
-
"rwkv.blocks.16.attention.time_mix_value": "pytorch_model-
|
| 161 |
-
"rwkv.blocks.16.attention.value.weight": "pytorch_model-
|
| 162 |
-
"rwkv.blocks.16.feed_forward.key.weight": "pytorch_model-
|
| 163 |
-
"rwkv.blocks.16.feed_forward.receptance.weight": "pytorch_model-
|
| 164 |
-
"rwkv.blocks.16.feed_forward.time_mix_key": "pytorch_model-
|
| 165 |
-
"rwkv.blocks.16.feed_forward.time_mix_receptance": "pytorch_model-
|
| 166 |
-
"rwkv.blocks.16.feed_forward.value.weight": "pytorch_model-
|
| 167 |
-
"rwkv.blocks.16.ln1.bias": "pytorch_model-
|
| 168 |
-
"rwkv.blocks.16.ln1.weight": "pytorch_model-
|
| 169 |
-
"rwkv.blocks.16.ln2.bias": "pytorch_model-
|
| 170 |
-
"rwkv.blocks.16.ln2.weight": "pytorch_model-
|
| 171 |
-
"rwkv.blocks.17.attention.key.weight": "pytorch_model-
|
| 172 |
-
"rwkv.blocks.17.attention.output.weight": "pytorch_model-
|
| 173 |
-
"rwkv.blocks.17.attention.receptance.weight": "pytorch_model-
|
| 174 |
-
"rwkv.blocks.17.attention.time_decay": "pytorch_model-
|
| 175 |
-
"rwkv.blocks.17.attention.time_first": "pytorch_model-
|
| 176 |
-
"rwkv.blocks.17.attention.time_mix_key": "pytorch_model-
|
| 177 |
-
"rwkv.blocks.17.attention.time_mix_receptance": "pytorch_model-
|
| 178 |
-
"rwkv.blocks.17.attention.time_mix_value": "pytorch_model-
|
| 179 |
-
"rwkv.blocks.17.attention.value.weight": "pytorch_model-
|
| 180 |
-
"rwkv.blocks.17.feed_forward.key.weight": "pytorch_model-
|
| 181 |
-
"rwkv.blocks.17.feed_forward.receptance.weight": "pytorch_model-
|
| 182 |
-
"rwkv.blocks.17.feed_forward.time_mix_key": "pytorch_model-
|
| 183 |
-
"rwkv.blocks.17.feed_forward.time_mix_receptance": "pytorch_model-
|
| 184 |
-
"rwkv.blocks.17.feed_forward.value.weight": "pytorch_model-
|
| 185 |
-
"rwkv.blocks.17.ln1.bias": "pytorch_model-
|
| 186 |
-
"rwkv.blocks.17.ln1.weight": "pytorch_model-
|
| 187 |
-
"rwkv.blocks.17.ln2.bias": "pytorch_model-
|
| 188 |
-
"rwkv.blocks.17.ln2.weight": "pytorch_model-
|
| 189 |
-
"rwkv.blocks.18.attention.key.weight": "pytorch_model-
|
| 190 |
-
"rwkv.blocks.18.attention.output.weight": "pytorch_model-
|
| 191 |
-
"rwkv.blocks.18.attention.receptance.weight": "pytorch_model-
|
| 192 |
-
"rwkv.blocks.18.attention.time_decay": "pytorch_model-
|
| 193 |
-
"rwkv.blocks.18.attention.time_first": "pytorch_model-
|
| 194 |
-
"rwkv.blocks.18.attention.time_mix_key": "pytorch_model-
|
| 195 |
-
"rwkv.blocks.18.attention.time_mix_receptance": "pytorch_model-
|
| 196 |
-
"rwkv.blocks.18.attention.time_mix_value": "pytorch_model-
|
| 197 |
-
"rwkv.blocks.18.attention.value.weight": "pytorch_model-
|
| 198 |
-
"rwkv.blocks.18.feed_forward.key.weight": "pytorch_model-
|
| 199 |
-
"rwkv.blocks.18.feed_forward.receptance.weight": "pytorch_model-
|
| 200 |
-
"rwkv.blocks.18.feed_forward.time_mix_key": "pytorch_model-
|
| 201 |
-
"rwkv.blocks.18.feed_forward.time_mix_receptance": "pytorch_model-
|
| 202 |
-
"rwkv.blocks.18.feed_forward.value.weight": "pytorch_model-
|
| 203 |
-
"rwkv.blocks.18.ln1.bias": "pytorch_model-
|
| 204 |
-
"rwkv.blocks.18.ln1.weight": "pytorch_model-
|
| 205 |
-
"rwkv.blocks.18.ln2.bias": "pytorch_model-
|
| 206 |
-
"rwkv.blocks.18.ln2.weight": "pytorch_model-
|
| 207 |
-
"rwkv.blocks.19.attention.key.weight": "pytorch_model-
|
| 208 |
-
"rwkv.blocks.19.attention.output.weight": "pytorch_model-
|
| 209 |
-
"rwkv.blocks.19.attention.receptance.weight": "pytorch_model-
|
| 210 |
-
"rwkv.blocks.19.attention.time_decay": "pytorch_model-
|
| 211 |
-
"rwkv.blocks.19.attention.time_first": "pytorch_model-
|
| 212 |
-
"rwkv.blocks.19.attention.time_mix_key": "pytorch_model-
|
| 213 |
-
"rwkv.blocks.19.attention.time_mix_receptance": "pytorch_model-
|
| 214 |
-
"rwkv.blocks.19.attention.time_mix_value": "pytorch_model-
|
| 215 |
-
"rwkv.blocks.19.attention.value.weight": "pytorch_model-
|
| 216 |
-
"rwkv.blocks.19.feed_forward.key.weight": "pytorch_model-
|
| 217 |
-
"rwkv.blocks.19.feed_forward.receptance.weight": "pytorch_model-
|
| 218 |
-
"rwkv.blocks.19.feed_forward.time_mix_key": "pytorch_model-
|
| 219 |
-
"rwkv.blocks.19.feed_forward.time_mix_receptance": "pytorch_model-
|
| 220 |
-
"rwkv.blocks.19.feed_forward.value.weight": "pytorch_model-
|
| 221 |
-
"rwkv.blocks.19.ln1.bias": "pytorch_model-
|
| 222 |
-
"rwkv.blocks.19.ln1.weight": "pytorch_model-
|
| 223 |
-
"rwkv.blocks.19.ln2.bias": "pytorch_model-
|
| 224 |
-
"rwkv.blocks.19.ln2.weight": "pytorch_model-
|
| 225 |
-
"rwkv.blocks.2.attention.key.weight": "pytorch_model-00001-of-
|
| 226 |
-
"rwkv.blocks.2.attention.output.weight": "pytorch_model-00001-of-
|
| 227 |
-
"rwkv.blocks.2.attention.receptance.weight": "pytorch_model-00001-of-
|
| 228 |
-
"rwkv.blocks.2.attention.time_decay": "pytorch_model-00001-of-
|
| 229 |
-
"rwkv.blocks.2.attention.time_first": "pytorch_model-00001-of-
|
| 230 |
-
"rwkv.blocks.2.attention.time_mix_key": "pytorch_model-00001-of-
|
| 231 |
-
"rwkv.blocks.2.attention.time_mix_receptance": "pytorch_model-00001-of-
|
| 232 |
-
"rwkv.blocks.2.attention.time_mix_value": "pytorch_model-00001-of-
|
| 233 |
-
"rwkv.blocks.2.attention.value.weight": "pytorch_model-00001-of-
|
| 234 |
-
"rwkv.blocks.2.feed_forward.key.weight": "pytorch_model-00001-of-
|
| 235 |
-
"rwkv.blocks.2.feed_forward.receptance.weight": "pytorch_model-
|
| 236 |
-
"rwkv.blocks.2.feed_forward.time_mix_key": "pytorch_model-00001-of-
|
| 237 |
-
"rwkv.blocks.2.feed_forward.time_mix_receptance": "pytorch_model-00001-of-
|
| 238 |
-
"rwkv.blocks.2.feed_forward.value.weight": "pytorch_model-
|
| 239 |
-
"rwkv.blocks.2.ln1.bias": "pytorch_model-00001-of-
|
| 240 |
-
"rwkv.blocks.2.ln1.weight": "pytorch_model-00001-of-
|
| 241 |
-
"rwkv.blocks.2.ln2.bias": "pytorch_model-00001-of-
|
| 242 |
-
"rwkv.blocks.2.ln2.weight": "pytorch_model-00001-of-
|
| 243 |
-
"rwkv.blocks.20.attention.key.weight": "pytorch_model-
|
| 244 |
-
"rwkv.blocks.20.attention.output.weight": "pytorch_model-
|
| 245 |
-
"rwkv.blocks.20.attention.receptance.weight": "pytorch_model-
|
| 246 |
-
"rwkv.blocks.20.attention.time_decay": "pytorch_model-
|
| 247 |
-
"rwkv.blocks.20.attention.time_first": "pytorch_model-
|
| 248 |
-
"rwkv.blocks.20.attention.time_mix_key": "pytorch_model-
|
| 249 |
-
"rwkv.blocks.20.attention.time_mix_receptance": "pytorch_model-
|
| 250 |
-
"rwkv.blocks.20.attention.time_mix_value": "pytorch_model-
|
| 251 |
-
"rwkv.blocks.20.attention.value.weight": "pytorch_model-
|
| 252 |
-
"rwkv.blocks.20.feed_forward.key.weight": "pytorch_model-
|
| 253 |
-
"rwkv.blocks.20.feed_forward.receptance.weight": "pytorch_model-
|
| 254 |
-
"rwkv.blocks.20.feed_forward.time_mix_key": "pytorch_model-
|
| 255 |
-
"rwkv.blocks.20.feed_forward.time_mix_receptance": "pytorch_model-
|
| 256 |
-
"rwkv.blocks.20.feed_forward.value.weight": "pytorch_model-
|
| 257 |
-
"rwkv.blocks.20.ln1.bias": "pytorch_model-
|
| 258 |
-
"rwkv.blocks.20.ln1.weight": "pytorch_model-
|
| 259 |
-
"rwkv.blocks.20.ln2.bias": "pytorch_model-
|
| 260 |
-
"rwkv.blocks.20.ln2.weight": "pytorch_model-
|
| 261 |
-
"rwkv.blocks.21.attention.key.weight": "pytorch_model-
|
| 262 |
-
"rwkv.blocks.21.attention.output.weight": "pytorch_model-
|
| 263 |
-
"rwkv.blocks.21.attention.receptance.weight": "pytorch_model-
|
| 264 |
-
"rwkv.blocks.21.attention.time_decay": "pytorch_model-
|
| 265 |
-
"rwkv.blocks.21.attention.time_first": "pytorch_model-
|
| 266 |
-
"rwkv.blocks.21.attention.time_mix_key": "pytorch_model-
|
| 267 |
-
"rwkv.blocks.21.attention.time_mix_receptance": "pytorch_model-
|
| 268 |
-
"rwkv.blocks.21.attention.time_mix_value": "pytorch_model-
|
| 269 |
-
"rwkv.blocks.21.attention.value.weight": "pytorch_model-
|
| 270 |
-
"rwkv.blocks.21.feed_forward.key.weight": "pytorch_model-
|
| 271 |
-
"rwkv.blocks.21.feed_forward.receptance.weight": "pytorch_model-
|
| 272 |
-
"rwkv.blocks.21.feed_forward.time_mix_key": "pytorch_model-
|
| 273 |
-
"rwkv.blocks.21.feed_forward.time_mix_receptance": "pytorch_model-
|
| 274 |
-
"rwkv.blocks.21.feed_forward.value.weight": "pytorch_model-
|
| 275 |
-
"rwkv.blocks.21.ln1.bias": "pytorch_model-
|
| 276 |
-
"rwkv.blocks.21.ln1.weight": "pytorch_model-
|
| 277 |
-
"rwkv.blocks.21.ln2.bias": "pytorch_model-
|
| 278 |
-
"rwkv.blocks.21.ln2.weight": "pytorch_model-
|
| 279 |
-
"rwkv.blocks.22.attention.key.weight": "pytorch_model-
|
| 280 |
-
"rwkv.blocks.22.attention.output.weight": "pytorch_model-
|
| 281 |
-
"rwkv.blocks.22.attention.receptance.weight": "pytorch_model-
|
| 282 |
-
"rwkv.blocks.22.attention.time_decay": "pytorch_model-
|
| 283 |
-
"rwkv.blocks.22.attention.time_first": "pytorch_model-
|
| 284 |
-
"rwkv.blocks.22.attention.time_mix_key": "pytorch_model-
|
| 285 |
-
"rwkv.blocks.22.attention.time_mix_receptance": "pytorch_model-
|
| 286 |
-
"rwkv.blocks.22.attention.time_mix_value": "pytorch_model-
|
| 287 |
-
"rwkv.blocks.22.attention.value.weight": "pytorch_model-
|
| 288 |
-
"rwkv.blocks.22.feed_forward.key.weight": "pytorch_model-
|
| 289 |
-
"rwkv.blocks.22.feed_forward.receptance.weight": "pytorch_model-
|
| 290 |
-
"rwkv.blocks.22.feed_forward.time_mix_key": "pytorch_model-
|
| 291 |
-
"rwkv.blocks.22.feed_forward.time_mix_receptance": "pytorch_model-
|
| 292 |
-
"rwkv.blocks.22.feed_forward.value.weight": "pytorch_model-
|
| 293 |
-
"rwkv.blocks.22.ln1.bias": "pytorch_model-
|
| 294 |
-
"rwkv.blocks.22.ln1.weight": "pytorch_model-
|
| 295 |
-
"rwkv.blocks.22.ln2.bias": "pytorch_model-
|
| 296 |
-
"rwkv.blocks.22.ln2.weight": "pytorch_model-
|
| 297 |
-
"rwkv.blocks.23.attention.key.weight": "pytorch_model-
|
| 298 |
-
"rwkv.blocks.23.attention.output.weight": "pytorch_model-
|
| 299 |
-
"rwkv.blocks.23.attention.receptance.weight": "pytorch_model-
|
| 300 |
-
"rwkv.blocks.23.attention.time_decay": "pytorch_model-
|
| 301 |
-
"rwkv.blocks.23.attention.time_first": "pytorch_model-
|
| 302 |
-
"rwkv.blocks.23.attention.time_mix_key": "pytorch_model-
|
| 303 |
-
"rwkv.blocks.23.attention.time_mix_receptance": "pytorch_model-
|
| 304 |
-
"rwkv.blocks.23.attention.time_mix_value": "pytorch_model-
|
| 305 |
-
"rwkv.blocks.23.attention.value.weight": "pytorch_model-
|
| 306 |
-
"rwkv.blocks.23.feed_forward.key.weight": "pytorch_model-
|
| 307 |
-
"rwkv.blocks.23.feed_forward.receptance.weight": "pytorch_model-
|
| 308 |
-
"rwkv.blocks.23.feed_forward.time_mix_key": "pytorch_model-
|
| 309 |
-
"rwkv.blocks.23.feed_forward.time_mix_receptance": "pytorch_model-
|
| 310 |
-
"rwkv.blocks.23.feed_forward.value.weight": "pytorch_model-
|
| 311 |
-
"rwkv.blocks.23.ln1.bias": "pytorch_model-
|
| 312 |
-
"rwkv.blocks.23.ln1.weight": "pytorch_model-
|
| 313 |
-
"rwkv.blocks.23.ln2.bias": "pytorch_model-
|
| 314 |
-
"rwkv.blocks.23.ln2.weight": "pytorch_model-
|
| 315 |
-
"rwkv.blocks.3.attention.key.weight": "pytorch_model-
|
| 316 |
-
"rwkv.blocks.3.attention.output.weight": "pytorch_model-
|
| 317 |
-
"rwkv.blocks.3.attention.receptance.weight": "pytorch_model-
|
| 318 |
-
"rwkv.blocks.3.attention.time_decay": "pytorch_model-
|
| 319 |
-
"rwkv.blocks.3.attention.time_first": "pytorch_model-
|
| 320 |
-
"rwkv.blocks.3.attention.time_mix_key": "pytorch_model-
|
| 321 |
-
"rwkv.blocks.3.attention.time_mix_receptance": "pytorch_model-
|
| 322 |
-
"rwkv.blocks.3.attention.time_mix_value": "pytorch_model-
|
| 323 |
-
"rwkv.blocks.3.attention.value.weight": "pytorch_model-
|
| 324 |
-
"rwkv.blocks.3.feed_forward.key.weight": "pytorch_model-
|
| 325 |
-
"rwkv.blocks.3.feed_forward.receptance.weight": "pytorch_model-
|
| 326 |
-
"rwkv.blocks.3.feed_forward.time_mix_key": "pytorch_model-
|
| 327 |
-
"rwkv.blocks.3.feed_forward.time_mix_receptance": "pytorch_model-
|
| 328 |
-
"rwkv.blocks.3.feed_forward.value.weight": "pytorch_model-
|
| 329 |
-
"rwkv.blocks.3.ln1.bias": "pytorch_model-
|
| 330 |
-
"rwkv.blocks.3.ln1.weight": "pytorch_model-
|
| 331 |
-
"rwkv.blocks.3.ln2.bias": "pytorch_model-
|
| 332 |
-
"rwkv.blocks.3.ln2.weight": "pytorch_model-
|
| 333 |
-
"rwkv.blocks.4.attention.key.weight": "pytorch_model-
|
| 334 |
-
"rwkv.blocks.4.attention.output.weight": "pytorch_model-
|
| 335 |
-
"rwkv.blocks.4.attention.receptance.weight": "pytorch_model-
|
| 336 |
-
"rwkv.blocks.4.attention.time_decay": "pytorch_model-
|
| 337 |
-
"rwkv.blocks.4.attention.time_first": "pytorch_model-
|
| 338 |
-
"rwkv.blocks.4.attention.time_mix_key": "pytorch_model-
|
| 339 |
-
"rwkv.blocks.4.attention.time_mix_receptance": "pytorch_model-
|
| 340 |
-
"rwkv.blocks.4.attention.time_mix_value": "pytorch_model-
|
| 341 |
-
"rwkv.blocks.4.attention.value.weight": "pytorch_model-
|
| 342 |
-
"rwkv.blocks.4.feed_forward.key.weight": "pytorch_model-
|
| 343 |
-
"rwkv.blocks.4.feed_forward.receptance.weight": "pytorch_model-
|
| 344 |
-
"rwkv.blocks.4.feed_forward.time_mix_key": "pytorch_model-
|
| 345 |
-
"rwkv.blocks.4.feed_forward.time_mix_receptance": "pytorch_model-
|
| 346 |
-
"rwkv.blocks.4.feed_forward.value.weight": "pytorch_model-
|
| 347 |
-
"rwkv.blocks.4.ln1.bias": "pytorch_model-
|
| 348 |
-
"rwkv.blocks.4.ln1.weight": "pytorch_model-
|
| 349 |
-
"rwkv.blocks.4.ln2.bias": "pytorch_model-
|
| 350 |
-
"rwkv.blocks.4.ln2.weight": "pytorch_model-
|
| 351 |
-
"rwkv.blocks.5.attention.key.weight": "pytorch_model-
|
| 352 |
-
"rwkv.blocks.5.attention.output.weight": "pytorch_model-
|
| 353 |
-
"rwkv.blocks.5.attention.receptance.weight": "pytorch_model-
|
| 354 |
-
"rwkv.blocks.5.attention.time_decay": "pytorch_model-
|
| 355 |
-
"rwkv.blocks.5.attention.time_first": "pytorch_model-
|
| 356 |
-
"rwkv.blocks.5.attention.time_mix_key": "pytorch_model-
|
| 357 |
-
"rwkv.blocks.5.attention.time_mix_receptance": "pytorch_model-
|
| 358 |
-
"rwkv.blocks.5.attention.time_mix_value": "pytorch_model-
|
| 359 |
-
"rwkv.blocks.5.attention.value.weight": "pytorch_model-
|
| 360 |
-
"rwkv.blocks.5.feed_forward.key.weight": "pytorch_model-
|
| 361 |
-
"rwkv.blocks.5.feed_forward.receptance.weight": "pytorch_model-
|
| 362 |
-
"rwkv.blocks.5.feed_forward.time_mix_key": "pytorch_model-
|
| 363 |
-
"rwkv.blocks.5.feed_forward.time_mix_receptance": "pytorch_model-
|
| 364 |
-
"rwkv.blocks.5.feed_forward.value.weight": "pytorch_model-
|
| 365 |
-
"rwkv.blocks.5.ln1.bias": "pytorch_model-
|
| 366 |
-
"rwkv.blocks.5.ln1.weight": "pytorch_model-
|
| 367 |
-
"rwkv.blocks.5.ln2.bias": "pytorch_model-
|
| 368 |
-
"rwkv.blocks.5.ln2.weight": "pytorch_model-
|
| 369 |
-
"rwkv.blocks.6.attention.key.weight": "pytorch_model-
|
| 370 |
-
"rwkv.blocks.6.attention.output.weight": "pytorch_model-
|
| 371 |
-
"rwkv.blocks.6.attention.receptance.weight": "pytorch_model-
|
| 372 |
-
"rwkv.blocks.6.attention.time_decay": "pytorch_model-
|
| 373 |
-
"rwkv.blocks.6.attention.time_first": "pytorch_model-
|
| 374 |
-
"rwkv.blocks.6.attention.time_mix_key": "pytorch_model-
|
| 375 |
-
"rwkv.blocks.6.attention.time_mix_receptance": "pytorch_model-
|
| 376 |
-
"rwkv.blocks.6.attention.time_mix_value": "pytorch_model-
|
| 377 |
-
"rwkv.blocks.6.attention.value.weight": "pytorch_model-
|
| 378 |
-
"rwkv.blocks.6.feed_forward.key.weight": "pytorch_model-
|
| 379 |
-
"rwkv.blocks.6.feed_forward.receptance.weight": "pytorch_model-
|
| 380 |
-
"rwkv.blocks.6.feed_forward.time_mix_key": "pytorch_model-
|
| 381 |
-
"rwkv.blocks.6.feed_forward.time_mix_receptance": "pytorch_model-
|
| 382 |
-
"rwkv.blocks.6.feed_forward.value.weight": "pytorch_model-
|
| 383 |
-
"rwkv.blocks.6.ln1.bias": "pytorch_model-
|
| 384 |
-
"rwkv.blocks.6.ln1.weight": "pytorch_model-
|
| 385 |
-
"rwkv.blocks.6.ln2.bias": "pytorch_model-
|
| 386 |
-
"rwkv.blocks.6.ln2.weight": "pytorch_model-
|
| 387 |
-
"rwkv.blocks.7.attention.key.weight": "pytorch_model-
|
| 388 |
-
"rwkv.blocks.7.attention.output.weight": "pytorch_model-
|
| 389 |
-
"rwkv.blocks.7.attention.receptance.weight": "pytorch_model-
|
| 390 |
-
"rwkv.blocks.7.attention.time_decay": "pytorch_model-
|
| 391 |
-
"rwkv.blocks.7.attention.time_first": "pytorch_model-
|
| 392 |
-
"rwkv.blocks.7.attention.time_mix_key": "pytorch_model-
|
| 393 |
-
"rwkv.blocks.7.attention.time_mix_receptance": "pytorch_model-
|
| 394 |
-
"rwkv.blocks.7.attention.time_mix_value": "pytorch_model-
|
| 395 |
-
"rwkv.blocks.7.attention.value.weight": "pytorch_model-
|
| 396 |
-
"rwkv.blocks.7.feed_forward.key.weight": "pytorch_model-
|
| 397 |
-
"rwkv.blocks.7.feed_forward.receptance.weight": "pytorch_model-
|
| 398 |
-
"rwkv.blocks.7.feed_forward.time_mix_key": "pytorch_model-
|
| 399 |
-
"rwkv.blocks.7.feed_forward.time_mix_receptance": "pytorch_model-
|
| 400 |
-
"rwkv.blocks.7.feed_forward.value.weight": "pytorch_model-
|
| 401 |
-
"rwkv.blocks.7.ln1.bias": "pytorch_model-
|
| 402 |
-
"rwkv.blocks.7.ln1.weight": "pytorch_model-
|
| 403 |
-
"rwkv.blocks.7.ln2.bias": "pytorch_model-
|
| 404 |
-
"rwkv.blocks.7.ln2.weight": "pytorch_model-
|
| 405 |
-
"rwkv.blocks.8.attention.key.weight": "pytorch_model-
|
| 406 |
-
"rwkv.blocks.8.attention.output.weight": "pytorch_model-
|
| 407 |
-
"rwkv.blocks.8.attention.receptance.weight": "pytorch_model-
|
| 408 |
-
"rwkv.blocks.8.attention.time_decay": "pytorch_model-
|
| 409 |
-
"rwkv.blocks.8.attention.time_first": "pytorch_model-
|
| 410 |
-
"rwkv.blocks.8.attention.time_mix_key": "pytorch_model-
|
| 411 |
-
"rwkv.blocks.8.attention.time_mix_receptance": "pytorch_model-
|
| 412 |
-
"rwkv.blocks.8.attention.time_mix_value": "pytorch_model-
|
| 413 |
-
"rwkv.blocks.8.attention.value.weight": "pytorch_model-
|
| 414 |
-
"rwkv.blocks.8.feed_forward.key.weight": "pytorch_model-
|
| 415 |
-
"rwkv.blocks.8.feed_forward.receptance.weight": "pytorch_model-
|
| 416 |
-
"rwkv.blocks.8.feed_forward.time_mix_key": "pytorch_model-
|
| 417 |
-
"rwkv.blocks.8.feed_forward.time_mix_receptance": "pytorch_model-
|
| 418 |
-
"rwkv.blocks.8.feed_forward.value.weight": "pytorch_model-
|
| 419 |
-
"rwkv.blocks.8.ln1.bias": "pytorch_model-
|
| 420 |
-
"rwkv.blocks.8.ln1.weight": "pytorch_model-
|
| 421 |
-
"rwkv.blocks.8.ln2.bias": "pytorch_model-
|
| 422 |
-
"rwkv.blocks.8.ln2.weight": "pytorch_model-
|
| 423 |
-
"rwkv.blocks.9.attention.key.weight": "pytorch_model-
|
| 424 |
-
"rwkv.blocks.9.attention.output.weight": "pytorch_model-
|
| 425 |
-
"rwkv.blocks.9.attention.receptance.weight": "pytorch_model-
|
| 426 |
-
"rwkv.blocks.9.attention.time_decay": "pytorch_model-
|
| 427 |
-
"rwkv.blocks.9.attention.time_first": "pytorch_model-
|
| 428 |
-
"rwkv.blocks.9.attention.time_mix_key": "pytorch_model-
|
| 429 |
-
"rwkv.blocks.9.attention.time_mix_receptance": "pytorch_model-
|
| 430 |
-
"rwkv.blocks.9.attention.time_mix_value": "pytorch_model-
|
| 431 |
-
"rwkv.blocks.9.attention.value.weight": "pytorch_model-
|
| 432 |
-
"rwkv.blocks.9.feed_forward.key.weight": "pytorch_model-
|
| 433 |
-
"rwkv.blocks.9.feed_forward.receptance.weight": "pytorch_model-
|
| 434 |
-
"rwkv.blocks.9.feed_forward.time_mix_key": "pytorch_model-
|
| 435 |
-
"rwkv.blocks.9.feed_forward.time_mix_receptance": "pytorch_model-
|
| 436 |
-
"rwkv.blocks.9.feed_forward.value.weight": "pytorch_model-
|
| 437 |
-
"rwkv.blocks.9.ln1.bias": "pytorch_model-
|
| 438 |
-
"rwkv.blocks.9.ln1.weight": "pytorch_model-
|
| 439 |
-
"rwkv.blocks.9.ln2.bias": "pytorch_model-
|
| 440 |
-
"rwkv.blocks.9.ln2.weight": "pytorch_model-
|
| 441 |
-
"rwkv.embeddings.weight": "pytorch_model-00001-of-
|
| 442 |
-
"rwkv.ln_out.bias": "pytorch_model-
|
| 443 |
-
"rwkv.ln_out.weight": "pytorch_model-
|
| 444 |
}
|
| 445 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"metadata": {
|
| 3 |
+
"total_size": 3044327424
|
| 4 |
},
|
| 5 |
"weight_map": {
|
| 6 |
+
"head.weight": "pytorch_model-00004-of-00004.bin",
|
| 7 |
+
"rwkv.blocks.0.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 8 |
+
"rwkv.blocks.0.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 9 |
+
"rwkv.blocks.0.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 10 |
+
"rwkv.blocks.0.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 11 |
+
"rwkv.blocks.0.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 12 |
+
"rwkv.blocks.0.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 13 |
+
"rwkv.blocks.0.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 14 |
+
"rwkv.blocks.0.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 15 |
+
"rwkv.blocks.0.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 16 |
+
"rwkv.blocks.0.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 17 |
+
"rwkv.blocks.0.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 18 |
+
"rwkv.blocks.0.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 19 |
+
"rwkv.blocks.0.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 20 |
+
"rwkv.blocks.0.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 21 |
+
"rwkv.blocks.0.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 22 |
+
"rwkv.blocks.0.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 23 |
+
"rwkv.blocks.0.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 24 |
+
"rwkv.blocks.0.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 25 |
+
"rwkv.blocks.0.pre_ln.bias": "pytorch_model-00001-of-00004.bin",
|
| 26 |
+
"rwkv.blocks.0.pre_ln.weight": "pytorch_model-00001-of-00004.bin",
|
| 27 |
+
"rwkv.blocks.1.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 28 |
+
"rwkv.blocks.1.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 29 |
+
"rwkv.blocks.1.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 30 |
+
"rwkv.blocks.1.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 31 |
+
"rwkv.blocks.1.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 32 |
+
"rwkv.blocks.1.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 33 |
+
"rwkv.blocks.1.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 34 |
+
"rwkv.blocks.1.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 35 |
+
"rwkv.blocks.1.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 36 |
+
"rwkv.blocks.1.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 37 |
+
"rwkv.blocks.1.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 38 |
+
"rwkv.blocks.1.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 39 |
+
"rwkv.blocks.1.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 40 |
+
"rwkv.blocks.1.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 41 |
+
"rwkv.blocks.1.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 42 |
+
"rwkv.blocks.1.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 43 |
+
"rwkv.blocks.1.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 44 |
+
"rwkv.blocks.1.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 45 |
+
"rwkv.blocks.10.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 46 |
+
"rwkv.blocks.10.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 47 |
+
"rwkv.blocks.10.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 48 |
+
"rwkv.blocks.10.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 49 |
+
"rwkv.blocks.10.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 50 |
+
"rwkv.blocks.10.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 51 |
+
"rwkv.blocks.10.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 52 |
+
"rwkv.blocks.10.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 53 |
+
"rwkv.blocks.10.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 54 |
+
"rwkv.blocks.10.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 55 |
+
"rwkv.blocks.10.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 56 |
+
"rwkv.blocks.10.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 57 |
+
"rwkv.blocks.10.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 58 |
+
"rwkv.blocks.10.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 59 |
+
"rwkv.blocks.10.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 60 |
+
"rwkv.blocks.10.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 61 |
+
"rwkv.blocks.10.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 62 |
+
"rwkv.blocks.10.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 63 |
+
"rwkv.blocks.11.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 64 |
+
"rwkv.blocks.11.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 65 |
+
"rwkv.blocks.11.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 66 |
+
"rwkv.blocks.11.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 67 |
+
"rwkv.blocks.11.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 68 |
+
"rwkv.blocks.11.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 69 |
+
"rwkv.blocks.11.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 70 |
+
"rwkv.blocks.11.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 71 |
+
"rwkv.blocks.11.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 72 |
+
"rwkv.blocks.11.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 73 |
+
"rwkv.blocks.11.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 74 |
+
"rwkv.blocks.11.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 75 |
+
"rwkv.blocks.11.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 76 |
+
"rwkv.blocks.11.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 77 |
+
"rwkv.blocks.11.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 78 |
+
"rwkv.blocks.11.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 79 |
+
"rwkv.blocks.11.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 80 |
+
"rwkv.blocks.11.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 81 |
+
"rwkv.blocks.12.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 82 |
+
"rwkv.blocks.12.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 83 |
+
"rwkv.blocks.12.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 84 |
+
"rwkv.blocks.12.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 85 |
+
"rwkv.blocks.12.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 86 |
+
"rwkv.blocks.12.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 87 |
+
"rwkv.blocks.12.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 88 |
+
"rwkv.blocks.12.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 89 |
+
"rwkv.blocks.12.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 90 |
+
"rwkv.blocks.12.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 91 |
+
"rwkv.blocks.12.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 92 |
+
"rwkv.blocks.12.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 93 |
+
"rwkv.blocks.12.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 94 |
+
"rwkv.blocks.12.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 95 |
+
"rwkv.blocks.12.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 96 |
+
"rwkv.blocks.12.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 97 |
+
"rwkv.blocks.12.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 98 |
+
"rwkv.blocks.12.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 99 |
+
"rwkv.blocks.13.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 100 |
+
"rwkv.blocks.13.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 101 |
+
"rwkv.blocks.13.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 102 |
+
"rwkv.blocks.13.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 103 |
+
"rwkv.blocks.13.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 104 |
+
"rwkv.blocks.13.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 105 |
+
"rwkv.blocks.13.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 106 |
+
"rwkv.blocks.13.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 107 |
+
"rwkv.blocks.13.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 108 |
+
"rwkv.blocks.13.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 109 |
+
"rwkv.blocks.13.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 110 |
+
"rwkv.blocks.13.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 111 |
+
"rwkv.blocks.13.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 112 |
+
"rwkv.blocks.13.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 113 |
+
"rwkv.blocks.13.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 114 |
+
"rwkv.blocks.13.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 115 |
+
"rwkv.blocks.13.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 116 |
+
"rwkv.blocks.13.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 117 |
+
"rwkv.blocks.14.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 118 |
+
"rwkv.blocks.14.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 119 |
+
"rwkv.blocks.14.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 120 |
+
"rwkv.blocks.14.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 121 |
+
"rwkv.blocks.14.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 122 |
+
"rwkv.blocks.14.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 123 |
+
"rwkv.blocks.14.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 124 |
+
"rwkv.blocks.14.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 125 |
+
"rwkv.blocks.14.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 126 |
+
"rwkv.blocks.14.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 127 |
+
"rwkv.blocks.14.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 128 |
+
"rwkv.blocks.14.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 129 |
+
"rwkv.blocks.14.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 130 |
+
"rwkv.blocks.14.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 131 |
+
"rwkv.blocks.14.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 132 |
+
"rwkv.blocks.14.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 133 |
+
"rwkv.blocks.14.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 134 |
+
"rwkv.blocks.14.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 135 |
+
"rwkv.blocks.15.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 136 |
+
"rwkv.blocks.15.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 137 |
+
"rwkv.blocks.15.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 138 |
+
"rwkv.blocks.15.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 139 |
+
"rwkv.blocks.15.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 140 |
+
"rwkv.blocks.15.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 141 |
+
"rwkv.blocks.15.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 142 |
+
"rwkv.blocks.15.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 143 |
+
"rwkv.blocks.15.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 144 |
+
"rwkv.blocks.15.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 145 |
+
"rwkv.blocks.15.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 146 |
+
"rwkv.blocks.15.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 147 |
+
"rwkv.blocks.15.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 148 |
+
"rwkv.blocks.15.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 149 |
+
"rwkv.blocks.15.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 150 |
+
"rwkv.blocks.15.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 151 |
+
"rwkv.blocks.15.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 152 |
+
"rwkv.blocks.15.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 153 |
+
"rwkv.blocks.16.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 154 |
+
"rwkv.blocks.16.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 155 |
+
"rwkv.blocks.16.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 156 |
+
"rwkv.blocks.16.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 157 |
+
"rwkv.blocks.16.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 158 |
+
"rwkv.blocks.16.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 159 |
+
"rwkv.blocks.16.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 160 |
+
"rwkv.blocks.16.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 161 |
+
"rwkv.blocks.16.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 162 |
+
"rwkv.blocks.16.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 163 |
+
"rwkv.blocks.16.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 164 |
+
"rwkv.blocks.16.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 165 |
+
"rwkv.blocks.16.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 166 |
+
"rwkv.blocks.16.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 167 |
+
"rwkv.blocks.16.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 168 |
+
"rwkv.blocks.16.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 169 |
+
"rwkv.blocks.16.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 170 |
+
"rwkv.blocks.16.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 171 |
+
"rwkv.blocks.17.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 172 |
+
"rwkv.blocks.17.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 173 |
+
"rwkv.blocks.17.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 174 |
+
"rwkv.blocks.17.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 175 |
+
"rwkv.blocks.17.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 176 |
+
"rwkv.blocks.17.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 177 |
+
"rwkv.blocks.17.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 178 |
+
"rwkv.blocks.17.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 179 |
+
"rwkv.blocks.17.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 180 |
+
"rwkv.blocks.17.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 181 |
+
"rwkv.blocks.17.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 182 |
+
"rwkv.blocks.17.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 183 |
+
"rwkv.blocks.17.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 184 |
+
"rwkv.blocks.17.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 185 |
+
"rwkv.blocks.17.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 186 |
+
"rwkv.blocks.17.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 187 |
+
"rwkv.blocks.17.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 188 |
+
"rwkv.blocks.17.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 189 |
+
"rwkv.blocks.18.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 190 |
+
"rwkv.blocks.18.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 191 |
+
"rwkv.blocks.18.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 192 |
+
"rwkv.blocks.18.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 193 |
+
"rwkv.blocks.18.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 194 |
+
"rwkv.blocks.18.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 195 |
+
"rwkv.blocks.18.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 196 |
+
"rwkv.blocks.18.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 197 |
+
"rwkv.blocks.18.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 198 |
+
"rwkv.blocks.18.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 199 |
+
"rwkv.blocks.18.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 200 |
+
"rwkv.blocks.18.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 201 |
+
"rwkv.blocks.18.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 202 |
+
"rwkv.blocks.18.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 203 |
+
"rwkv.blocks.18.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 204 |
+
"rwkv.blocks.18.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 205 |
+
"rwkv.blocks.18.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 206 |
+
"rwkv.blocks.18.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 207 |
+
"rwkv.blocks.19.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 208 |
+
"rwkv.blocks.19.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 209 |
+
"rwkv.blocks.19.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 210 |
+
"rwkv.blocks.19.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 211 |
+
"rwkv.blocks.19.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 212 |
+
"rwkv.blocks.19.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 213 |
+
"rwkv.blocks.19.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 214 |
+
"rwkv.blocks.19.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 215 |
+
"rwkv.blocks.19.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 216 |
+
"rwkv.blocks.19.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 217 |
+
"rwkv.blocks.19.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 218 |
+
"rwkv.blocks.19.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 219 |
+
"rwkv.blocks.19.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 220 |
+
"rwkv.blocks.19.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 221 |
+
"rwkv.blocks.19.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 222 |
+
"rwkv.blocks.19.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 223 |
+
"rwkv.blocks.19.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 224 |
+
"rwkv.blocks.19.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 225 |
+
"rwkv.blocks.2.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 226 |
+
"rwkv.blocks.2.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 227 |
+
"rwkv.blocks.2.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 228 |
+
"rwkv.blocks.2.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 229 |
+
"rwkv.blocks.2.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 230 |
+
"rwkv.blocks.2.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 231 |
+
"rwkv.blocks.2.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 232 |
+
"rwkv.blocks.2.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 233 |
+
"rwkv.blocks.2.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 234 |
+
"rwkv.blocks.2.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 235 |
+
"rwkv.blocks.2.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 236 |
+
"rwkv.blocks.2.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 237 |
+
"rwkv.blocks.2.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 238 |
+
"rwkv.blocks.2.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 239 |
+
"rwkv.blocks.2.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 240 |
+
"rwkv.blocks.2.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 241 |
+
"rwkv.blocks.2.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 242 |
+
"rwkv.blocks.2.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 243 |
+
"rwkv.blocks.20.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 244 |
+
"rwkv.blocks.20.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 245 |
+
"rwkv.blocks.20.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 246 |
+
"rwkv.blocks.20.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 247 |
+
"rwkv.blocks.20.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 248 |
+
"rwkv.blocks.20.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 249 |
+
"rwkv.blocks.20.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 250 |
+
"rwkv.blocks.20.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 251 |
+
"rwkv.blocks.20.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 252 |
+
"rwkv.blocks.20.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 253 |
+
"rwkv.blocks.20.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 254 |
+
"rwkv.blocks.20.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 255 |
+
"rwkv.blocks.20.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 256 |
+
"rwkv.blocks.20.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 257 |
+
"rwkv.blocks.20.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 258 |
+
"rwkv.blocks.20.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 259 |
+
"rwkv.blocks.20.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 260 |
+
"rwkv.blocks.20.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 261 |
+
"rwkv.blocks.21.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 262 |
+
"rwkv.blocks.21.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 263 |
+
"rwkv.blocks.21.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 264 |
+
"rwkv.blocks.21.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 265 |
+
"rwkv.blocks.21.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 266 |
+
"rwkv.blocks.21.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 267 |
+
"rwkv.blocks.21.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 268 |
+
"rwkv.blocks.21.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 269 |
+
"rwkv.blocks.21.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 270 |
+
"rwkv.blocks.21.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 271 |
+
"rwkv.blocks.21.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 272 |
+
"rwkv.blocks.21.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 273 |
+
"rwkv.blocks.21.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 274 |
+
"rwkv.blocks.21.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 275 |
+
"rwkv.blocks.21.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 276 |
+
"rwkv.blocks.21.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 277 |
+
"rwkv.blocks.21.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 278 |
+
"rwkv.blocks.21.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 279 |
+
"rwkv.blocks.22.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 280 |
+
"rwkv.blocks.22.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 281 |
+
"rwkv.blocks.22.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 282 |
+
"rwkv.blocks.22.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 283 |
+
"rwkv.blocks.22.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 284 |
+
"rwkv.blocks.22.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 285 |
+
"rwkv.blocks.22.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 286 |
+
"rwkv.blocks.22.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 287 |
+
"rwkv.blocks.22.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 288 |
+
"rwkv.blocks.22.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 289 |
+
"rwkv.blocks.22.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 290 |
+
"rwkv.blocks.22.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 291 |
+
"rwkv.blocks.22.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 292 |
+
"rwkv.blocks.22.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 293 |
+
"rwkv.blocks.22.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 294 |
+
"rwkv.blocks.22.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 295 |
+
"rwkv.blocks.22.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 296 |
+
"rwkv.blocks.22.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 297 |
+
"rwkv.blocks.23.attention.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 298 |
+
"rwkv.blocks.23.attention.output.weight": "pytorch_model-00003-of-00004.bin",
|
| 299 |
+
"rwkv.blocks.23.attention.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 300 |
+
"rwkv.blocks.23.attention.time_decay": "pytorch_model-00003-of-00004.bin",
|
| 301 |
+
"rwkv.blocks.23.attention.time_first": "pytorch_model-00003-of-00004.bin",
|
| 302 |
+
"rwkv.blocks.23.attention.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 303 |
+
"rwkv.blocks.23.attention.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 304 |
+
"rwkv.blocks.23.attention.time_mix_value": "pytorch_model-00003-of-00004.bin",
|
| 305 |
+
"rwkv.blocks.23.attention.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 306 |
+
"rwkv.blocks.23.feed_forward.key.weight": "pytorch_model-00003-of-00004.bin",
|
| 307 |
+
"rwkv.blocks.23.feed_forward.receptance.weight": "pytorch_model-00003-of-00004.bin",
|
| 308 |
+
"rwkv.blocks.23.feed_forward.time_mix_key": "pytorch_model-00003-of-00004.bin",
|
| 309 |
+
"rwkv.blocks.23.feed_forward.time_mix_receptance": "pytorch_model-00003-of-00004.bin",
|
| 310 |
+
"rwkv.blocks.23.feed_forward.value.weight": "pytorch_model-00003-of-00004.bin",
|
| 311 |
+
"rwkv.blocks.23.ln1.bias": "pytorch_model-00003-of-00004.bin",
|
| 312 |
+
"rwkv.blocks.23.ln1.weight": "pytorch_model-00003-of-00004.bin",
|
| 313 |
+
"rwkv.blocks.23.ln2.bias": "pytorch_model-00003-of-00004.bin",
|
| 314 |
+
"rwkv.blocks.23.ln2.weight": "pytorch_model-00003-of-00004.bin",
|
| 315 |
+
"rwkv.blocks.3.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 316 |
+
"rwkv.blocks.3.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 317 |
+
"rwkv.blocks.3.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 318 |
+
"rwkv.blocks.3.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 319 |
+
"rwkv.blocks.3.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 320 |
+
"rwkv.blocks.3.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 321 |
+
"rwkv.blocks.3.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 322 |
+
"rwkv.blocks.3.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 323 |
+
"rwkv.blocks.3.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 324 |
+
"rwkv.blocks.3.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 325 |
+
"rwkv.blocks.3.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 326 |
+
"rwkv.blocks.3.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 327 |
+
"rwkv.blocks.3.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 328 |
+
"rwkv.blocks.3.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 329 |
+
"rwkv.blocks.3.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 330 |
+
"rwkv.blocks.3.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 331 |
+
"rwkv.blocks.3.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 332 |
+
"rwkv.blocks.3.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 333 |
+
"rwkv.blocks.4.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 334 |
+
"rwkv.blocks.4.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 335 |
+
"rwkv.blocks.4.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 336 |
+
"rwkv.blocks.4.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 337 |
+
"rwkv.blocks.4.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 338 |
+
"rwkv.blocks.4.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 339 |
+
"rwkv.blocks.4.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 340 |
+
"rwkv.blocks.4.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 341 |
+
"rwkv.blocks.4.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 342 |
+
"rwkv.blocks.4.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 343 |
+
"rwkv.blocks.4.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 344 |
+
"rwkv.blocks.4.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 345 |
+
"rwkv.blocks.4.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 346 |
+
"rwkv.blocks.4.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 347 |
+
"rwkv.blocks.4.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 348 |
+
"rwkv.blocks.4.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 349 |
+
"rwkv.blocks.4.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 350 |
+
"rwkv.blocks.4.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 351 |
+
"rwkv.blocks.5.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 352 |
+
"rwkv.blocks.5.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 353 |
+
"rwkv.blocks.5.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 354 |
+
"rwkv.blocks.5.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 355 |
+
"rwkv.blocks.5.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 356 |
+
"rwkv.blocks.5.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 357 |
+
"rwkv.blocks.5.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 358 |
+
"rwkv.blocks.5.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 359 |
+
"rwkv.blocks.5.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 360 |
+
"rwkv.blocks.5.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 361 |
+
"rwkv.blocks.5.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 362 |
+
"rwkv.blocks.5.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 363 |
+
"rwkv.blocks.5.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 364 |
+
"rwkv.blocks.5.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 365 |
+
"rwkv.blocks.5.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 366 |
+
"rwkv.blocks.5.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 367 |
+
"rwkv.blocks.5.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 368 |
+
"rwkv.blocks.5.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 369 |
+
"rwkv.blocks.6.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 370 |
+
"rwkv.blocks.6.attention.output.weight": "pytorch_model-00001-of-00004.bin",
|
| 371 |
+
"rwkv.blocks.6.attention.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 372 |
+
"rwkv.blocks.6.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 373 |
+
"rwkv.blocks.6.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 374 |
+
"rwkv.blocks.6.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 375 |
+
"rwkv.blocks.6.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 376 |
+
"rwkv.blocks.6.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 377 |
+
"rwkv.blocks.6.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 378 |
+
"rwkv.blocks.6.feed_forward.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 379 |
+
"rwkv.blocks.6.feed_forward.receptance.weight": "pytorch_model-00001-of-00004.bin",
|
| 380 |
+
"rwkv.blocks.6.feed_forward.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 381 |
+
"rwkv.blocks.6.feed_forward.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 382 |
+
"rwkv.blocks.6.feed_forward.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 383 |
+
"rwkv.blocks.6.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 384 |
+
"rwkv.blocks.6.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 385 |
+
"rwkv.blocks.6.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 386 |
+
"rwkv.blocks.6.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 387 |
+
"rwkv.blocks.7.attention.key.weight": "pytorch_model-00001-of-00004.bin",
|
| 388 |
+
"rwkv.blocks.7.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 389 |
+
"rwkv.blocks.7.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 390 |
+
"rwkv.blocks.7.attention.time_decay": "pytorch_model-00001-of-00004.bin",
|
| 391 |
+
"rwkv.blocks.7.attention.time_first": "pytorch_model-00001-of-00004.bin",
|
| 392 |
+
"rwkv.blocks.7.attention.time_mix_key": "pytorch_model-00001-of-00004.bin",
|
| 393 |
+
"rwkv.blocks.7.attention.time_mix_receptance": "pytorch_model-00001-of-00004.bin",
|
| 394 |
+
"rwkv.blocks.7.attention.time_mix_value": "pytorch_model-00001-of-00004.bin",
|
| 395 |
+
"rwkv.blocks.7.attention.value.weight": "pytorch_model-00001-of-00004.bin",
|
| 396 |
+
"rwkv.blocks.7.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 397 |
+
"rwkv.blocks.7.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 398 |
+
"rwkv.blocks.7.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 399 |
+
"rwkv.blocks.7.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 400 |
+
"rwkv.blocks.7.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 401 |
+
"rwkv.blocks.7.ln1.bias": "pytorch_model-00001-of-00004.bin",
|
| 402 |
+
"rwkv.blocks.7.ln1.weight": "pytorch_model-00001-of-00004.bin",
|
| 403 |
+
"rwkv.blocks.7.ln2.bias": "pytorch_model-00001-of-00004.bin",
|
| 404 |
+
"rwkv.blocks.7.ln2.weight": "pytorch_model-00001-of-00004.bin",
|
| 405 |
+
"rwkv.blocks.8.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 406 |
+
"rwkv.blocks.8.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 407 |
+
"rwkv.blocks.8.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 408 |
+
"rwkv.blocks.8.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 409 |
+
"rwkv.blocks.8.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 410 |
+
"rwkv.blocks.8.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 411 |
+
"rwkv.blocks.8.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 412 |
+
"rwkv.blocks.8.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 413 |
+
"rwkv.blocks.8.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 414 |
+
"rwkv.blocks.8.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 415 |
+
"rwkv.blocks.8.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 416 |
+
"rwkv.blocks.8.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 417 |
+
"rwkv.blocks.8.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 418 |
+
"rwkv.blocks.8.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 419 |
+
"rwkv.blocks.8.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 420 |
+
"rwkv.blocks.8.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 421 |
+
"rwkv.blocks.8.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 422 |
+
"rwkv.blocks.8.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 423 |
+
"rwkv.blocks.9.attention.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 424 |
+
"rwkv.blocks.9.attention.output.weight": "pytorch_model-00002-of-00004.bin",
|
| 425 |
+
"rwkv.blocks.9.attention.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 426 |
+
"rwkv.blocks.9.attention.time_decay": "pytorch_model-00002-of-00004.bin",
|
| 427 |
+
"rwkv.blocks.9.attention.time_first": "pytorch_model-00002-of-00004.bin",
|
| 428 |
+
"rwkv.blocks.9.attention.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 429 |
+
"rwkv.blocks.9.attention.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 430 |
+
"rwkv.blocks.9.attention.time_mix_value": "pytorch_model-00002-of-00004.bin",
|
| 431 |
+
"rwkv.blocks.9.attention.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 432 |
+
"rwkv.blocks.9.feed_forward.key.weight": "pytorch_model-00002-of-00004.bin",
|
| 433 |
+
"rwkv.blocks.9.feed_forward.receptance.weight": "pytorch_model-00002-of-00004.bin",
|
| 434 |
+
"rwkv.blocks.9.feed_forward.time_mix_key": "pytorch_model-00002-of-00004.bin",
|
| 435 |
+
"rwkv.blocks.9.feed_forward.time_mix_receptance": "pytorch_model-00002-of-00004.bin",
|
| 436 |
+
"rwkv.blocks.9.feed_forward.value.weight": "pytorch_model-00002-of-00004.bin",
|
| 437 |
+
"rwkv.blocks.9.ln1.bias": "pytorch_model-00002-of-00004.bin",
|
| 438 |
+
"rwkv.blocks.9.ln1.weight": "pytorch_model-00002-of-00004.bin",
|
| 439 |
+
"rwkv.blocks.9.ln2.bias": "pytorch_model-00002-of-00004.bin",
|
| 440 |
+
"rwkv.blocks.9.ln2.weight": "pytorch_model-00002-of-00004.bin",
|
| 441 |
+
"rwkv.embeddings.weight": "pytorch_model-00001-of-00004.bin",
|
| 442 |
+
"rwkv.ln_out.bias": "pytorch_model-00003-of-00004.bin",
|
| 443 |
+
"rwkv.ln_out.weight": "pytorch_model-00003-of-00004.bin"
|
| 444 |
}
|
| 445 |
}
|