Hugging Face's logo

beomi
/
KoRWKV-6B

Text Generation
Transformers
PyTorch
Safetensors
Korean
rwkv
KoRWKV
Inference Endpoints
Model card Files Community
1
beomi commited on
Commit
841b1d3
•
1 Parent(s): ae8ce18

Add >85B trained ckpt

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. config.json +1 -1
  2. model-00001-of-00030.safetensors → model-00001-of-00015.safetensors +2 -2
  3. model-00002-of-00030.safetensors → model-00002-of-00015.safetensors +2 -2
  4. model-00003-of-00030.safetensors → model-00003-of-00015.safetensors +2 -2
  5. model-00004-of-00030.safetensors → model-00004-of-00015.safetensors +2 -2
  6. model-00005-of-00015.safetensors +3 -0
  7. model-00005-of-00030.safetensors +0 -3
  8. model-00006-of-00015.safetensors +3 -0
  9. model-00006-of-00030.safetensors +0 -3
  10. model-00007-of-00015.safetensors +3 -0
  11. model-00007-of-00030.safetensors +0 -3
  12. model-00008-of-00015.safetensors +3 -0
  13. model-00008-of-00030.safetensors +0 -3
  14. model-00009-of-00015.safetensors +3 -0
  15. model-00009-of-00030.safetensors +0 -3
  16. model-00010-of-00015.safetensors +3 -0
  17. model-00010-of-00030.safetensors +0 -3
  18. model-00011-of-00015.safetensors +3 -0
  19. model-00011-of-00030.safetensors +0 -3
  20. model-00012-of-00015.safetensors +3 -0
  21. model-00012-of-00030.safetensors +0 -3
  22. model-00013-of-00015.safetensors +3 -0
  23. model-00013-of-00030.safetensors +0 -3
  24. model-00014-of-00015.safetensors +3 -0
  25. model-00014-of-00030.safetensors +0 -3
  26. model-00015-of-00015.safetensors +3 -0
  27. model-00015-of-00030.safetensors +0 -3
  28. model-00016-of-00030.safetensors +0 -3
  29. model-00017-of-00030.safetensors +0 -3
  30. model-00018-of-00030.safetensors +0 -3
  31. model-00019-of-00030.safetensors +0 -3
  32. model-00020-of-00030.safetensors +0 -3
  33. model-00021-of-00030.safetensors +0 -3
  34. model-00022-of-00030.safetensors +0 -3
  35. model-00023-of-00030.safetensors +0 -3
  36. model-00024-of-00030.safetensors +0 -3
  37. model-00025-of-00030.safetensors +0 -3
  38. model-00026-of-00030.safetensors +0 -3
  39. model-00027-of-00030.safetensors +0 -3
  40. model-00028-of-00030.safetensors +0 -3
  41. model-00029-of-00030.safetensors +0 -3
  42. model-00030-of-00030.safetensors +0 -3
  43. model.safetensors.index.json +511 -511
  44. pytorch_model-00001-of-00015.bin +3 -0
  45. pytorch_model-00002-of-00015.bin +3 -0
  46. pytorch_model-00003-of-00015.bin +3 -0
  47. pytorch_model-00004-of-00015.bin +3 -0
  48. pytorch_model-00005-of-00015.bin +3 -0
  49. pytorch_model-00006-of-00015.bin +3 -0
  50. pytorch_model-00007-of-00015.bin +3 -0
config.json CHANGED
@@ -13,7 +13,7 @@
13
  "num_hidden_layers": 28,
14
  "rescale_every": 6,
15
  "tie_word_embeddings": false,
16
- "torch_dtype": "float32",
17
  "transformers_version": "4.30.0.dev0",
18
  "use_cache": true,
19
  "vocab_size": 52000
 
13
  "num_hidden_layers": 28,
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-00030.safetensors → model-00001-of-00015.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5b1564b3facea1f85525e6d529dff4dc1f1f2abcfbff410917a25aaa4f292bb6
3
- size 986367416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21f31d9b9d24c2be5e317302e6196c0ecb64f4011d82eea868135127b72ea0e2
3
+ size 996609896
model-00002-of-00030.safetensors → model-00002-of-00015.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6155ecb531752ee842ef45e36d599208722fd431890c0d5fb7ddfb2594e442ac
3
- size 939706432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e440c42a7999400a16d88af2f1c8e96374d9fbd3b89d91c095e151a40acd1e53
3
+ size 872599456
model-00003-of-00030.safetensors → model-00003-of-00015.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6c49e16df29d1a2f527317b1e5e73f17d207ec25983f8b85904193a72f818490
3
- size 939739432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4df6d25bb68e5e98a02f052714931ae443dd5ef19c13fcc0f9730bd889abd6df
3
+ size 872599456
model-00004-of-00030.safetensors → model-00004-of-00015.safetensors RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c0c84a18564a9010ff75b385f08a82ba640a13f6695b4c824396e2c1bb8004f9
3
- size 872597456
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec574f691e99a889c0cd5fc2d017c6687bb412c71ae94d68f7c1bdb6efb7ebd4
3
+ size 872599456
model-00005-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ed7683f09b9fb7afe925b2ac87bc1cff0538408ffcf5c4035730cd87fbb64a7b
3
+ size 872599456
model-00005-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a06704ea148302d6cc5738321bf0dce23270583c9dc2404b8915950240e9c4b7
3
- size 872597456
 
 
 
 
model-00006-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8a9bd8deaf303af48ff55fb5a7eb3ea061426897ae7ca0fef92b82819941737
3
+ size 872599488
model-00006-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:14e624a8fdd7d6e8e7ebb45d491ab6eeff6e2f27e568d43c9da9e81ad6a43a69
3
- size 872597456
 
 
 
 
model-00007-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:146c55d89a29e099678dccf5e15f9c5e222e03972ef86caaeb2fa79da0259237
3
+ size 872599496
model-00007-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2a032699d64069522567bcd45219701d550eabbc75fa8962718bd3a5d5901bf
3
- size 872597456
 
 
 
 
model-00008-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f5cc091ed177e06ad46fb2d5f6cecbc5938a9cf455e544e39f827c5d74fff6d
3
+ size 872599496
model-00008-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:30118129d50be5f416ec34afaf4173f4b6b68ab3823fed9f7b4fb34a0f7eb514
3
- size 872597456
 
 
 
 
model-00009-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:223beaab951ec6f8f155dcca2582c860b724919d5b0a943e6f8a02ea8be37b2c
3
+ size 872599496
model-00009-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9e7730f1946117965a8d853e66cd24acbfaae265135490dae4100135b345639a
3
- size 872597456
 
 
 
 
model-00010-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe59b2c942a110a9240d5996c80d8a2ddabfe3359ae98acb943ada62f687e776
3
+ size 872599496
model-00010-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:48f9fce176af20011050225b6a9557a83086454239f89cd33e3d16c62b73d47a
3
- size 872597456
 
 
 
 
model-00011-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eede5189a289344a837319003f798c73e471752a9abfaa04e42d583eaaafb309
3
+ size 872599496
model-00011-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:eaf356d5f273c35ee7c1195223b6cb096437b32354da760ae04723765d549fcc
3
- size 872597472
 
 
 
 
model-00012-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43716490d178cebf11e022fc3fd9c585c50b6b2631a730774de4ac08e18f7dc7
3
+ size 872599496
model-00012-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4d700c362e3b7390f521f4fc4fda44c8bc922b31bde249eb61b3b67ee6b73d50
3
- size 872597472
 
 
 
 
model-00013-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72ecede78c405fc5720f72ea4f3afa6a2d310611f193778c518f6a0bde2cc4da
3
+ size 872599496
model-00013-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:05eba864dfddd8726699f4bd8a9690f567f5085fa40975b6a6a2d399f76d1353
3
- size 872597472
 
 
 
 
model-00014-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ca00fc3659b05010b7ff59c778e8555d2b840f1f75bc66c93c20e736caa0b85
3
+ size 872599496
model-00014-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:79fe77c0bab7ea6e126cf5a54aaeb1d57b06d09dcdb73587328053774caedacb
3
- size 872597472
 
 
 
 
model-00015-of-00015.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:432c75b3c4faafae4a001864d0cd7b6c2d94a5504b0a0982f438502505b50800
3
+ size 727990928
model-00015-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e557259c097317c3df8e2d170ca863a4cebe4608d1e9bf5d77e39ae05550ed17
3
- size 872597472
 
 
 
 
model-00016-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:fe8243e8c0e9cd54c34c8752b5cdfb61c73a831e57d34ee721c0468d8e54e50f
3
- size 872597472
 
 
 
 
model-00017-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:dd1959f2b85ba27ba4a66951ccebcc90926613b9564c29d9311ccb2fae8713ba
3
- size 872597472
 
 
 
 
model-00018-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:11475e1d7be89ba9394cb31bd10e898cf49b5d190f8beb35ecdb21d5f00288dd
3
- size 872597472
 
 
 
 
model-00019-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:13d656f43ce3d628f50c0ac23b722a9e0904f0ff997f1c01e2cc97a8fd6d8896
3
- size 872597472
 
 
 
 
model-00020-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3d6d3f1a4630f5a43601c7e8df51ebf7c3936a8e705308b40c7f9180478d094a
3
- size 872597472
 
 
 
 
model-00021-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:201e35623e89bdc06a7b08190d05b1deb2bba3c707cee88d03659936bcd264e4
3
- size 872597472
 
 
 
 
model-00022-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:46896ff2aad8a8b3113b72044ad159137e68c5bbbb1ce1f7de597499279dd96a
3
- size 872597472
 
 
 
 
model-00023-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:720df898df45852aaa64f3d8fbb1d1566b925d7cc938234d4af5e98cb24f8325
3
- size 872597472
 
 
 
 
model-00024-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:abfce58890845f482f09e76de44fe740d870df82c727fd9d80c3afa873c41df3
3
- size 872597472
 
 
 
 
model-00025-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0ca2c54bcdfe4ff6bce52f80e457e788572c72af6d95a0f5905b7130b980e226
3
- size 872597472
 
 
 
 
model-00026-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:315cac0ffa49b8d741b516d187a64864fa8ad9ace54011c158e4518cb41635e4
3
- size 872597472
 
 
 
 
model-00027-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b811d7df854a0d07a02680757159760d19bd14b369998afdb154fef53c3a399a
3
- size 872597472
 
 
 
 
model-00028-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:3a4db3e11ce563562639e71362a0186e286c10e1230789e69ea9f5489c6b3011
3
- size 872597472
 
 
 
 
model-00029-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f4db61c915178f05d0ca21b15ee0143accf89d998c208a44c73d3b5f43707c77
3
- size 604013112
 
 
 
 
model-00030-of-00030.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0f03ebe151ec769f9fec912368cf772b6b595084f0fa44acf3bf5563e136bb9d
3
- size 851968120
 
 
 
 
model.safetensors.index.json CHANGED
@@ -1,517 +1,517 @@
1
  {
2
  "metadata": {
3
- "total_size": 26136674304
4
  },
5
  "weight_map": {
6
- "head.weight": "model-00030-of-00030.safetensors",
7
- "rwkv.blocks.0.attention.key.weight": "model-00001-of-00030.safetensors",
8
- "rwkv.blocks.0.attention.output.weight": "model-00002-of-00030.safetensors",
9
- "rwkv.blocks.0.attention.receptance.weight": "model-00002-of-00030.safetensors",
10
- "rwkv.blocks.0.attention.time_decay": "model-00001-of-00030.safetensors",
11
- "rwkv.blocks.0.attention.time_first": "model-00001-of-00030.safetensors",
12
- "rwkv.blocks.0.attention.time_mix_key": "model-00001-of-00030.safetensors",
13
- "rwkv.blocks.0.attention.time_mix_receptance": "model-00001-of-00030.safetensors",
14
- "rwkv.blocks.0.attention.time_mix_value": "model-00001-of-00030.safetensors",
15
- "rwkv.blocks.0.attention.value.weight": "model-00001-of-00030.safetensors",
16
- "rwkv.blocks.0.feed_forward.key.weight": "model-00002-of-00030.safetensors",
17
- "rwkv.blocks.0.feed_forward.receptance.weight": "model-00002-of-00030.safetensors",
18
- "rwkv.blocks.0.feed_forward.time_mix_key": "model-00002-of-00030.safetensors",
19
- "rwkv.blocks.0.feed_forward.time_mix_receptance": "model-00002-of-00030.safetensors",
20
- "rwkv.blocks.0.feed_forward.value.weight": "model-00002-of-00030.safetensors",
21
- "rwkv.blocks.0.ln1.bias": "model-00001-of-00030.safetensors",
22
- "rwkv.blocks.0.ln1.weight": "model-00001-of-00030.safetensors",
23
- "rwkv.blocks.0.ln2.bias": "model-00001-of-00030.safetensors",
24
- "rwkv.blocks.0.ln2.weight": "model-00001-of-00030.safetensors",
25
- "rwkv.blocks.0.pre_ln.bias": "model-00001-of-00030.safetensors",
26
- "rwkv.blocks.0.pre_ln.weight": "model-00001-of-00030.safetensors",
27
- "rwkv.blocks.1.attention.key.weight": "model-00002-of-00030.safetensors",
28
- "rwkv.blocks.1.attention.output.weight": "model-00003-of-00030.safetensors",
29
- "rwkv.blocks.1.attention.receptance.weight": "model-00002-of-00030.safetensors",
30
- "rwkv.blocks.1.attention.time_decay": "model-00002-of-00030.safetensors",
31
- "rwkv.blocks.1.attention.time_first": "model-00002-of-00030.safetensors",
32
- "rwkv.blocks.1.attention.time_mix_key": "model-00002-of-00030.safetensors",
33
- "rwkv.blocks.1.attention.time_mix_receptance": "model-00002-of-00030.safetensors",
34
- "rwkv.blocks.1.attention.time_mix_value": "model-00002-of-00030.safetensors",
35
- "rwkv.blocks.1.attention.value.weight": "model-00002-of-00030.safetensors",
36
- "rwkv.blocks.1.feed_forward.key.weight": "model-00003-of-00030.safetensors",
37
- "rwkv.blocks.1.feed_forward.receptance.weight": "model-00003-of-00030.safetensors",
38
- "rwkv.blocks.1.feed_forward.time_mix_key": "model-00003-of-00030.safetensors",
39
- "rwkv.blocks.1.feed_forward.time_mix_receptance": "model-00003-of-00030.safetensors",
40
- "rwkv.blocks.1.feed_forward.value.weight": "model-00003-of-00030.safetensors",
41
- "rwkv.blocks.1.ln1.bias": "model-00002-of-00030.safetensors",
42
- "rwkv.blocks.1.ln1.weight": "model-00002-of-00030.safetensors",
43
- "rwkv.blocks.1.ln2.bias": "model-00002-of-00030.safetensors",
44
- "rwkv.blocks.1.ln2.weight": "model-00002-of-00030.safetensors",
45
- "rwkv.blocks.10.attention.key.weight": "model-00011-of-00030.safetensors",
46
- "rwkv.blocks.10.attention.output.weight": "model-00011-of-00030.safetensors",
47
- "rwkv.blocks.10.attention.receptance.weight": "model-00011-of-00030.safetensors",
48
- "rwkv.blocks.10.attention.time_decay": "model-00011-of-00030.safetensors",
49
- "rwkv.blocks.10.attention.time_first": "model-00011-of-00030.safetensors",
50
- "rwkv.blocks.10.attention.time_mix_key": "model-00011-of-00030.safetensors",
51
- "rwkv.blocks.10.attention.time_mix_receptance": "model-00011-of-00030.safetensors",
52
- "rwkv.blocks.10.attention.time_mix_value": "model-00011-of-00030.safetensors",
53
- "rwkv.blocks.10.attention.value.weight": "model-00011-of-00030.safetensors",
54
- "rwkv.blocks.10.feed_forward.key.weight": "model-00012-of-00030.safetensors",
55
- "rwkv.blocks.10.feed_forward.receptance.weight": "model-00012-of-00030.safetensors",
56
- "rwkv.blocks.10.feed_forward.time_mix_key": "model-00011-of-00030.safetensors",
57
- "rwkv.blocks.10.feed_forward.time_mix_receptance": "model-00011-of-00030.safetensors",
58
- "rwkv.blocks.10.feed_forward.value.weight": "model-00012-of-00030.safetensors",
59
- "rwkv.blocks.10.ln1.bias": "model-00011-of-00030.safetensors",
60
- "rwkv.blocks.10.ln1.weight": "model-00011-of-00030.safetensors",
61
- "rwkv.blocks.10.ln2.bias": "model-00011-of-00030.safetensors",
62
- "rwkv.blocks.10.ln2.weight": "model-00011-of-00030.safetensors",
63
- "rwkv.blocks.11.attention.key.weight": "model-00012-of-00030.safetensors",
64
- "rwkv.blocks.11.attention.output.weight": "model-00012-of-00030.safetensors",
65
- "rwkv.blocks.11.attention.receptance.weight": "model-00012-of-00030.safetensors",
66
- "rwkv.blocks.11.attention.time_decay": "model-00012-of-00030.safetensors",
67
- "rwkv.blocks.11.attention.time_first": "model-00012-of-00030.safetensors",
68
- "rwkv.blocks.11.attention.time_mix_key": "model-00012-of-00030.safetensors",
69
- "rwkv.blocks.11.attention.time_mix_receptance": "model-00012-of-00030.safetensors",
70
- "rwkv.blocks.11.attention.time_mix_value": "model-00012-of-00030.safetensors",
71
- "rwkv.blocks.11.attention.value.weight": "model-00012-of-00030.safetensors",
72
- "rwkv.blocks.11.feed_forward.key.weight": "model-00013-of-00030.safetensors",
73
- "rwkv.blocks.11.feed_forward.receptance.weight": "model-00013-of-00030.safetensors",
74
- "rwkv.blocks.11.feed_forward.time_mix_key": "model-00012-of-00030.safetensors",
75
- "rwkv.blocks.11.feed_forward.time_mix_receptance": "model-00012-of-00030.safetensors",
76
- "rwkv.blocks.11.feed_forward.value.weight": "model-00013-of-00030.safetensors",
77
- "rwkv.blocks.11.ln1.bias": "model-00012-of-00030.safetensors",
78
- "rwkv.blocks.11.ln1.weight": "model-00012-of-00030.safetensors",
79
- "rwkv.blocks.11.ln2.bias": "model-00012-of-00030.safetensors",
80
- "rwkv.blocks.11.ln2.weight": "model-00012-of-00030.safetensors",
81
- "rwkv.blocks.12.attention.key.weight": "model-00013-of-00030.safetensors",
82
- "rwkv.blocks.12.attention.output.weight": "model-00013-of-00030.safetensors",
83
- "rwkv.blocks.12.attention.receptance.weight": "model-00013-of-00030.safetensors",
84
- "rwkv.blocks.12.attention.time_decay": "model-00013-of-00030.safetensors",
85
- "rwkv.blocks.12.attention.time_first": "model-00013-of-00030.safetensors",
86
- "rwkv.blocks.12.attention.time_mix_key": "model-00013-of-00030.safetensors",
87
- "rwkv.blocks.12.attention.time_mix_receptance": "model-00013-of-00030.safetensors",
88
- "rwkv.blocks.12.attention.time_mix_value": "model-00013-of-00030.safetensors",
89
- "rwkv.blocks.12.attention.value.weight": "model-00013-of-00030.safetensors",
90
- "rwkv.blocks.12.feed_forward.key.weight": "model-00014-of-00030.safetensors",
91
- "rwkv.blocks.12.feed_forward.receptance.weight": "model-00014-of-00030.safetensors",
92
- "rwkv.blocks.12.feed_forward.time_mix_key": "model-00013-of-00030.safetensors",
93
- "rwkv.blocks.12.feed_forward.time_mix_receptance": "model-00013-of-00030.safetensors",
94
- "rwkv.blocks.12.feed_forward.value.weight": "model-00014-of-00030.safetensors",
95
- "rwkv.blocks.12.ln1.bias": "model-00013-of-00030.safetensors",
96
- "rwkv.blocks.12.ln1.weight": "model-00013-of-00030.safetensors",
97
- "rwkv.blocks.12.ln2.bias": "model-00013-of-00030.safetensors",
98
- "rwkv.blocks.12.ln2.weight": "model-00013-of-00030.safetensors",
99
- "rwkv.blocks.13.attention.key.weight": "model-00014-of-00030.safetensors",
100
- "rwkv.blocks.13.attention.output.weight": "model-00014-of-00030.safetensors",
101
- "rwkv.blocks.13.attention.receptance.weight": "model-00014-of-00030.safetensors",
102
- "rwkv.blocks.13.attention.time_decay": "model-00014-of-00030.safetensors",
103
- "rwkv.blocks.13.attention.time_first": "model-00014-of-00030.safetensors",
104
- "rwkv.blocks.13.attention.time_mix_key": "model-00014-of-00030.safetensors",
105
- "rwkv.blocks.13.attention.time_mix_receptance": "model-00014-of-00030.safetensors",
106
- "rwkv.blocks.13.attention.time_mix_value": "model-00014-of-00030.safetensors",
107
- "rwkv.blocks.13.attention.value.weight": "model-00014-of-00030.safetensors",
108
- "rwkv.blocks.13.feed_forward.key.weight": "model-00015-of-00030.safetensors",
109
- "rwkv.blocks.13.feed_forward.receptance.weight": "model-00015-of-00030.safetensors",
110
- "rwkv.blocks.13.feed_forward.time_mix_key": "model-00014-of-00030.safetensors",
111
- "rwkv.blocks.13.feed_forward.time_mix_receptance": "model-00014-of-00030.safetensors",
112
- "rwkv.blocks.13.feed_forward.value.weight": "model-00015-of-00030.safetensors",
113
- "rwkv.blocks.13.ln1.bias": "model-00014-of-00030.safetensors",
114
- "rwkv.blocks.13.ln1.weight": "model-00014-of-00030.safetensors",
115
- "rwkv.blocks.13.ln2.bias": "model-00014-of-00030.safetensors",
116
- "rwkv.blocks.13.ln2.weight": "model-00014-of-00030.safetensors",
117
- "rwkv.blocks.14.attention.key.weight": "model-00015-of-00030.safetensors",
118
- "rwkv.blocks.14.attention.output.weight": "model-00015-of-00030.safetensors",
119
- "rwkv.blocks.14.attention.receptance.weight": "model-00015-of-00030.safetensors",
120
- "rwkv.blocks.14.attention.time_decay": "model-00015-of-00030.safetensors",
121
- "rwkv.blocks.14.attention.time_first": "model-00015-of-00030.safetensors",
122
- "rwkv.blocks.14.attention.time_mix_key": "model-00015-of-00030.safetensors",
123
- "rwkv.blocks.14.attention.time_mix_receptance": "model-00015-of-00030.safetensors",
124
- "rwkv.blocks.14.attention.time_mix_value": "model-00015-of-00030.safetensors",
125
- "rwkv.blocks.14.attention.value.weight": "model-00015-of-00030.safetensors",
126
- "rwkv.blocks.14.feed_forward.key.weight": "model-00016-of-00030.safetensors",
127
- "rwkv.blocks.14.feed_forward.receptance.weight": "model-00016-of-00030.safetensors",
128
- "rwkv.blocks.14.feed_forward.time_mix_key": "model-00015-of-00030.safetensors",
129
- "rwkv.blocks.14.feed_forward.time_mix_receptance": "model-00015-of-00030.safetensors",
130
- "rwkv.blocks.14.feed_forward.value.weight": "model-00016-of-00030.safetensors",
131
- "rwkv.blocks.14.ln1.bias": "model-00015-of-00030.safetensors",
132
- "rwkv.blocks.14.ln1.weight": "model-00015-of-00030.safetensors",
133
- "rwkv.blocks.14.ln2.bias": "model-00015-of-00030.safetensors",
134
- "rwkv.blocks.14.ln2.weight": "model-00015-of-00030.safetensors",
135
- "rwkv.blocks.15.attention.key.weight": "model-00016-of-00030.safetensors",
136
- "rwkv.blocks.15.attention.output.weight": "model-00016-of-00030.safetensors",
137
- "rwkv.blocks.15.attention.receptance.weight": "model-00016-of-00030.safetensors",
138
- "rwkv.blocks.15.attention.time_decay": "model-00016-of-00030.safetensors",
139
- "rwkv.blocks.15.attention.time_first": "model-00016-of-00030.safetensors",
140
- "rwkv.blocks.15.attention.time_mix_key": "model-00016-of-00030.safetensors",
141
- "rwkv.blocks.15.attention.time_mix_receptance": "model-00016-of-00030.safetensors",
142
- "rwkv.blocks.15.attention.time_mix_value": "model-00016-of-00030.safetensors",
143
- "rwkv.blocks.15.attention.value.weight": "model-00016-of-00030.safetensors",
144
- "rwkv.blocks.15.feed_forward.key.weight": "model-00017-of-00030.safetensors",
145
- "rwkv.blocks.15.feed_forward.receptance.weight": "model-00017-of-00030.safetensors",
146
- "rwkv.blocks.15.feed_forward.time_mix_key": "model-00016-of-00030.safetensors",
147
- "rwkv.blocks.15.feed_forward.time_mix_receptance": "model-00016-of-00030.safetensors",
148
- "rwkv.blocks.15.feed_forward.value.weight": "model-00017-of-00030.safetensors",
149
- "rwkv.blocks.15.ln1.bias": "model-00016-of-00030.safetensors",
150
- "rwkv.blocks.15.ln1.weight": "model-00016-of-00030.safetensors",
151
- "rwkv.blocks.15.ln2.bias": "model-00016-of-00030.safetensors",
152
- "rwkv.blocks.15.ln2.weight": "model-00016-of-00030.safetensors",
153
- "rwkv.blocks.16.attention.key.weight": "model-00017-of-00030.safetensors",
154
- "rwkv.blocks.16.attention.output.weight": "model-00017-of-00030.safetensors",
155
- "rwkv.blocks.16.attention.receptance.weight": "model-00017-of-00030.safetensors",
156
- "rwkv.blocks.16.attention.time_decay": "model-00017-of-00030.safetensors",
157
- "rwkv.blocks.16.attention.time_first": "model-00017-of-00030.safetensors",
158
- "rwkv.blocks.16.attention.time_mix_key": "model-00017-of-00030.safetensors",
159
- "rwkv.blocks.16.attention.time_mix_receptance": "model-00017-of-00030.safetensors",
160
- "rwkv.blocks.16.attention.time_mix_value": "model-00017-of-00030.safetensors",
161
- "rwkv.blocks.16.attention.value.weight": "model-00017-of-00030.safetensors",
162
- "rwkv.blocks.16.feed_forward.key.weight": "model-00018-of-00030.safetensors",
163
- "rwkv.blocks.16.feed_forward.receptance.weight": "model-00018-of-00030.safetensors",
164
- "rwkv.blocks.16.feed_forward.time_mix_key": "model-00017-of-00030.safetensors",
165
- "rwkv.blocks.16.feed_forward.time_mix_receptance": "model-00017-of-00030.safetensors",
166
- "rwkv.blocks.16.feed_forward.value.weight": "model-00018-of-00030.safetensors",
167
- "rwkv.blocks.16.ln1.bias": "model-00017-of-00030.safetensors",
168
- "rwkv.blocks.16.ln1.weight": "model-00017-of-00030.safetensors",
169
- "rwkv.blocks.16.ln2.bias": "model-00017-of-00030.safetensors",
170
- "rwkv.blocks.16.ln2.weight": "model-00017-of-00030.safetensors",
171
- "rwkv.blocks.17.attention.key.weight": "model-00018-of-00030.safetensors",
172
- "rwkv.blocks.17.attention.output.weight": "model-00018-of-00030.safetensors",
173
- "rwkv.blocks.17.attention.receptance.weight": "model-00018-of-00030.safetensors",
174
- "rwkv.blocks.17.attention.time_decay": "model-00018-of-00030.safetensors",
175
- "rwkv.blocks.17.attention.time_first": "model-00018-of-00030.safetensors",
176
- "rwkv.blocks.17.attention.time_mix_key": "model-00018-of-00030.safetensors",
177
- "rwkv.blocks.17.attention.time_mix_receptance": "model-00018-of-00030.safetensors",
178
- "rwkv.blocks.17.attention.time_mix_value": "model-00018-of-00030.safetensors",
179
- "rwkv.blocks.17.attention.value.weight": "model-00018-of-00030.safetensors",
180
- "rwkv.blocks.17.feed_forward.key.weight": "model-00019-of-00030.safetensors",
181
- "rwkv.blocks.17.feed_forward.receptance.weight": "model-00019-of-00030.safetensors",
182
- "rwkv.blocks.17.feed_forward.time_mix_key": "model-00018-of-00030.safetensors",
183
- "rwkv.blocks.17.feed_forward.time_mix_receptance": "model-00018-of-00030.safetensors",
184
- "rwkv.blocks.17.feed_forward.value.weight": "model-00019-of-00030.safetensors",
185
- "rwkv.blocks.17.ln1.bias": "model-00018-of-00030.safetensors",
186
- "rwkv.blocks.17.ln1.weight": "model-00018-of-00030.safetensors",
187
- "rwkv.blocks.17.ln2.bias": "model-00018-of-00030.safetensors",
188
- "rwkv.blocks.17.ln2.weight": "model-00018-of-00030.safetensors",
189
- "rwkv.blocks.18.attention.key.weight": "model-00019-of-00030.safetensors",
190
- "rwkv.blocks.18.attention.output.weight": "model-00019-of-00030.safetensors",
191
- "rwkv.blocks.18.attention.receptance.weight": "model-00019-of-00030.safetensors",
192
- "rwkv.blocks.18.attention.time_decay": "model-00019-of-00030.safetensors",
193
- "rwkv.blocks.18.attention.time_first": "model-00019-of-00030.safetensors",
194
- "rwkv.blocks.18.attention.time_mix_key": "model-00019-of-00030.safetensors",
195
- "rwkv.blocks.18.attention.time_mix_receptance": "model-00019-of-00030.safetensors",
196
- "rwkv.blocks.18.attention.time_mix_value": "model-00019-of-00030.safetensors",
197
- "rwkv.blocks.18.attention.value.weight": "model-00019-of-00030.safetensors",
198
- "rwkv.blocks.18.feed_forward.key.weight": "model-00020-of-00030.safetensors",
199
- "rwkv.blocks.18.feed_forward.receptance.weight": "model-00020-of-00030.safetensors",
200
- "rwkv.blocks.18.feed_forward.time_mix_key": "model-00019-of-00030.safetensors",
201
- "rwkv.blocks.18.feed_forward.time_mix_receptance": "model-00019-of-00030.safetensors",
202
- "rwkv.blocks.18.feed_forward.value.weight": "model-00020-of-00030.safetensors",
203
- "rwkv.blocks.18.ln1.bias": "model-00019-of-00030.safetensors",
204
- "rwkv.blocks.18.ln1.weight": "model-00019-of-00030.safetensors",
205
- "rwkv.blocks.18.ln2.bias": "model-00019-of-00030.safetensors",
206
- "rwkv.blocks.18.ln2.weight": "model-00019-of-00030.safetensors",
207
- "rwkv.blocks.19.attention.key.weight": "model-00020-of-00030.safetensors",
208
- "rwkv.blocks.19.attention.output.weight": "model-00020-of-00030.safetensors",
209
- "rwkv.blocks.19.attention.receptance.weight": "model-00020-of-00030.safetensors",
210
- "rwkv.blocks.19.attention.time_decay": "model-00020-of-00030.safetensors",
211
- "rwkv.blocks.19.attention.time_first": "model-00020-of-00030.safetensors",
212
- "rwkv.blocks.19.attention.time_mix_key": "model-00020-of-00030.safetensors",
213
- "rwkv.blocks.19.attention.time_mix_receptance": "model-00020-of-00030.safetensors",
214
- "rwkv.blocks.19.attention.time_mix_value": "model-00020-of-00030.safetensors",
215
- "rwkv.blocks.19.attention.value.weight": "model-00020-of-00030.safetensors",
216
- "rwkv.blocks.19.feed_forward.key.weight": "model-00021-of-00030.safetensors",
217
- "rwkv.blocks.19.feed_forward.receptance.weight": "model-00021-of-00030.safetensors",
218
- "rwkv.blocks.19.feed_forward.time_mix_key": "model-00020-of-00030.safetensors",
219
- "rwkv.blocks.19.feed_forward.time_mix_receptance": "model-00020-of-00030.safetensors",
220
- "rwkv.blocks.19.feed_forward.value.weight": "model-00021-of-00030.safetensors",
221
- "rwkv.blocks.19.ln1.bias": "model-00020-of-00030.safetensors",
222
- "rwkv.blocks.19.ln1.weight": "model-00020-of-00030.safetensors",
223
- "rwkv.blocks.19.ln2.bias": "model-00020-of-00030.safetensors",
224
- "rwkv.blocks.19.ln2.weight": "model-00020-of-00030.safetensors",
225
- "rwkv.blocks.2.attention.key.weight": "model-00003-of-00030.safetensors",
226
- "rwkv.blocks.2.attention.output.weight": "model-00003-of-00030.safetensors",
227
- "rwkv.blocks.2.attention.receptance.weight": "model-00003-of-00030.safetensors",
228
- "rwkv.blocks.2.attention.time_decay": "model-00003-of-00030.safetensors",
229
- "rwkv.blocks.2.attention.time_first": "model-00003-of-00030.safetensors",
230
- "rwkv.blocks.2.attention.time_mix_key": "model-00003-of-00030.safetensors",
231
- "rwkv.blocks.2.attention.time_mix_receptance": "model-00003-of-00030.safetensors",
232
- "rwkv.blocks.2.attention.time_mix_value": "model-00003-of-00030.safetensors",
233
- "rwkv.blocks.2.attention.value.weight": "model-00003-of-00030.safetensors",
234
- "rwkv.blocks.2.feed_forward.key.weight": "model-00004-of-00030.safetensors",
235
- "rwkv.blocks.2.feed_forward.receptance.weight": "model-00004-of-00030.safetensors",
236
- "rwkv.blocks.2.feed_forward.time_mix_key": "model-00003-of-00030.safetensors",
237
- "rwkv.blocks.2.feed_forward.time_mix_receptance": "model-00003-of-00030.safetensors",
238
- "rwkv.blocks.2.feed_forward.value.weight": "model-00004-of-00030.safetensors",
239
- "rwkv.blocks.2.ln1.bias": "model-00003-of-00030.safetensors",
240
- "rwkv.blocks.2.ln1.weight": "model-00003-of-00030.safetensors",
241
- "rwkv.blocks.2.ln2.bias": "model-00003-of-00030.safetensors",
242
- "rwkv.blocks.2.ln2.weight": "model-00003-of-00030.safetensors",
243
- "rwkv.blocks.20.attention.key.weight": "model-00021-of-00030.safetensors",
244
- "rwkv.blocks.20.attention.output.weight": "model-00021-of-00030.safetensors",
245
- "rwkv.blocks.20.attention.receptance.weight": "model-00021-of-00030.safetensors",
246
- "rwkv.blocks.20.attention.time_decay": "model-00021-of-00030.safetensors",
247
- "rwkv.blocks.20.attention.time_first": "model-00021-of-00030.safetensors",
248
- "rwkv.blocks.20.attention.time_mix_key": "model-00021-of-00030.safetensors",
249
- "rwkv.blocks.20.attention.time_mix_receptance": "model-00021-of-00030.safetensors",
250
- "rwkv.blocks.20.attention.time_mix_value": "model-00021-of-00030.safetensors",
251
- "rwkv.blocks.20.attention.value.weight": "model-00021-of-00030.safetensors",
252
- "rwkv.blocks.20.feed_forward.key.weight": "model-00022-of-00030.safetensors",
253
- "rwkv.blocks.20.feed_forward.receptance.weight": "model-00022-of-00030.safetensors",
254
- "rwkv.blocks.20.feed_forward.time_mix_key": "model-00021-of-00030.safetensors",
255
- "rwkv.blocks.20.feed_forward.time_mix_receptance": "model-00021-of-00030.safetensors",
256
- "rwkv.blocks.20.feed_forward.value.weight": "model-00022-of-00030.safetensors",
257
- "rwkv.blocks.20.ln1.bias": "model-00021-of-00030.safetensors",
258
- "rwkv.blocks.20.ln1.weight": "model-00021-of-00030.safetensors",
259
- "rwkv.blocks.20.ln2.bias": "model-00021-of-00030.safetensors",
260
- "rwkv.blocks.20.ln2.weight": "model-00021-of-00030.safetensors",
261
- "rwkv.blocks.21.attention.key.weight": "model-00022-of-00030.safetensors",
262
- "rwkv.blocks.21.attention.output.weight": "model-00022-of-00030.safetensors",
263
- "rwkv.blocks.21.attention.receptance.weight": "model-00022-of-00030.safetensors",
264
- "rwkv.blocks.21.attention.time_decay": "model-00022-of-00030.safetensors",
265
- "rwkv.blocks.21.attention.time_first": "model-00022-of-00030.safetensors",
266
- "rwkv.blocks.21.attention.time_mix_key": "model-00022-of-00030.safetensors",
267
- "rwkv.blocks.21.attention.time_mix_receptance": "model-00022-of-00030.safetensors",
268
- "rwkv.blocks.21.attention.time_mix_value": "model-00022-of-00030.safetensors",
269
- "rwkv.blocks.21.attention.value.weight": "model-00022-of-00030.safetensors",
270
- "rwkv.blocks.21.feed_forward.key.weight": "model-00023-of-00030.safetensors",
271
- "rwkv.blocks.21.feed_forward.receptance.weight": "model-00023-of-00030.safetensors",
272
- "rwkv.blocks.21.feed_forward.time_mix_key": "model-00022-of-00030.safetensors",
273
- "rwkv.blocks.21.feed_forward.time_mix_receptance": "model-00022-of-00030.safetensors",
274
- "rwkv.blocks.21.feed_forward.value.weight": "model-00023-of-00030.safetensors",
275
- "rwkv.blocks.21.ln1.bias": "model-00022-of-00030.safetensors",
276
- "rwkv.blocks.21.ln1.weight": "model-00022-of-00030.safetensors",
277
- "rwkv.blocks.21.ln2.bias": "model-00022-of-00030.safetensors",
278
- "rwkv.blocks.21.ln2.weight": "model-00022-of-00030.safetensors",
279
- "rwkv.blocks.22.attention.key.weight": "model-00023-of-00030.safetensors",
280
- "rwkv.blocks.22.attention.output.weight": "model-00023-of-00030.safetensors",
281
- "rwkv.blocks.22.attention.receptance.weight": "model-00023-of-00030.safetensors",
282
- "rwkv.blocks.22.attention.time_decay": "model-00023-of-00030.safetensors",
283
- "rwkv.blocks.22.attention.time_first": "model-00023-of-00030.safetensors",
284
- "rwkv.blocks.22.attention.time_mix_key": "model-00023-of-00030.safetensors",
285
- "rwkv.blocks.22.attention.time_mix_receptance": "model-00023-of-00030.safetensors",
286
- "rwkv.blocks.22.attention.time_mix_value": "model-00023-of-00030.safetensors",
287
- "rwkv.blocks.22.attention.value.weight": "model-00023-of-00030.safetensors",
288
- "rwkv.blocks.22.feed_forward.key.weight": "model-00024-of-00030.safetensors",
289
- "rwkv.blocks.22.feed_forward.receptance.weight": "model-00024-of-00030.safetensors",
290
- "rwkv.blocks.22.feed_forward.time_mix_key": "model-00023-of-00030.safetensors",
291
- "rwkv.blocks.22.feed_forward.time_mix_receptance": "model-00023-of-00030.safetensors",
292
- "rwkv.blocks.22.feed_forward.value.weight": "model-00024-of-00030.safetensors",
293
- "rwkv.blocks.22.ln1.bias": "model-00023-of-00030.safetensors",
294
- "rwkv.blocks.22.ln1.weight": "model-00023-of-00030.safetensors",
295
- "rwkv.blocks.22.ln2.bias": "model-00023-of-00030.safetensors",
296
- "rwkv.blocks.22.ln2.weight": "model-00023-of-00030.safetensors",
297
- "rwkv.blocks.23.attention.key.weight": "model-00024-of-00030.safetensors",
298
- "rwkv.blocks.23.attention.output.weight": "model-00024-of-00030.safetensors",
299
- "rwkv.blocks.23.attention.receptance.weight": "model-00024-of-00030.safetensors",
300
- "rwkv.blocks.23.attention.time_decay": "model-00024-of-00030.safetensors",
301
- "rwkv.blocks.23.attention.time_first": "model-00024-of-00030.safetensors",
302
- "rwkv.blocks.23.attention.time_mix_key": "model-00024-of-00030.safetensors",
303
- "rwkv.blocks.23.attention.time_mix_receptance": "model-00024-of-00030.safetensors",
304
- "rwkv.blocks.23.attention.time_mix_value": "model-00024-of-00030.safetensors",
305
- "rwkv.blocks.23.attention.value.weight": "model-00024-of-00030.safetensors",
306
- "rwkv.blocks.23.feed_forward.key.weight": "model-00025-of-00030.safetensors",
307
- "rwkv.blocks.23.feed_forward.receptance.weight": "model-00025-of-00030.safetensors",
308
- "rwkv.blocks.23.feed_forward.time_mix_key": "model-00024-of-00030.safetensors",
309
- "rwkv.blocks.23.feed_forward.time_mix_receptance": "model-00024-of-00030.safetensors",
310
- "rwkv.blocks.23.feed_forward.value.weight": "model-00025-of-00030.safetensors",
311
- "rwkv.blocks.23.ln1.bias": "model-00024-of-00030.safetensors",
312
- "rwkv.blocks.23.ln1.weight": "model-00024-of-00030.safetensors",
313
- "rwkv.blocks.23.ln2.bias": "model-00024-of-00030.safetensors",
314
- "rwkv.blocks.23.ln2.weight": "model-00024-of-00030.safetensors",
315
- "rwkv.blocks.24.attention.key.weight": "model-00025-of-00030.safetensors",
316
- "rwkv.blocks.24.attention.output.weight": "model-00025-of-00030.safetensors",
317
- "rwkv.blocks.24.attention.receptance.weight": "model-00025-of-00030.safetensors",
318
- "rwkv.blocks.24.attention.time_decay": "model-00025-of-00030.safetensors",
319
- "rwkv.blocks.24.attention.time_first": "model-00025-of-00030.safetensors",
320
- "rwkv.blocks.24.attention.time_mix_key": "model-00025-of-00030.safetensors",
321
- "rwkv.blocks.24.attention.time_mix_receptance": "model-00025-of-00030.safetensors",
322
- "rwkv.blocks.24.attention.time_mix_value": "model-00025-of-00030.safetensors",
323
- "rwkv.blocks.24.attention.value.weight": "model-00025-of-00030.safetensors",
324
- "rwkv.blocks.24.feed_forward.key.weight": "model-00026-of-00030.safetensors",
325
- "rwkv.blocks.24.feed_forward.receptance.weight": "model-00026-of-00030.safetensors",
326
- "rwkv.blocks.24.feed_forward.time_mix_key": "model-00025-of-00030.safetensors",
327
- "rwkv.blocks.24.feed_forward.time_mix_receptance": "model-00025-of-00030.safetensors",
328
- "rwkv.blocks.24.feed_forward.value.weight": "model-00026-of-00030.safetensors",
329
- "rwkv.blocks.24.ln1.bias": "model-00025-of-00030.safetensors",
330
- "rwkv.blocks.24.ln1.weight": "model-00025-of-00030.safetensors",
331
- "rwkv.blocks.24.ln2.bias": "model-00025-of-00030.safetensors",
332
- "rwkv.blocks.24.ln2.weight": "model-00025-of-00030.safetensors",
333
- "rwkv.blocks.25.attention.key.weight": "model-00026-of-00030.safetensors",
334
- "rwkv.blocks.25.attention.output.weight": "model-00026-of-00030.safetensors",
335
- "rwkv.blocks.25.attention.receptance.weight": "model-00026-of-00030.safetensors",
336
- "rwkv.blocks.25.attention.time_decay": "model-00026-of-00030.safetensors",
337
- "rwkv.blocks.25.attention.time_first": "model-00026-of-00030.safetensors",
338
- "rwkv.blocks.25.attention.time_mix_key": "model-00026-of-00030.safetensors",
339
- "rwkv.blocks.25.attention.time_mix_receptance": "model-00026-of-00030.safetensors",
340
- "rwkv.blocks.25.attention.time_mix_value": "model-00026-of-00030.safetensors",
341
- "rwkv.blocks.25.attention.value.weight": "model-00026-of-00030.safetensors",
342
- "rwkv.blocks.25.feed_forward.key.weight": "model-00027-of-00030.safetensors",
343
- "rwkv.blocks.25.feed_forward.receptance.weight": "model-00027-of-00030.safetensors",
344
- "rwkv.blocks.25.feed_forward.time_mix_key": "model-00026-of-00030.safetensors",
345
- "rwkv.blocks.25.feed_forward.time_mix_receptance": "model-00026-of-00030.safetensors",
346
- "rwkv.blocks.25.feed_forward.value.weight": "model-00027-of-00030.safetensors",
347
- "rwkv.blocks.25.ln1.bias": "model-00026-of-00030.safetensors",
348
- "rwkv.blocks.25.ln1.weight": "model-00026-of-00030.safetensors",
349
- "rwkv.blocks.25.ln2.bias": "model-00026-of-00030.safetensors",
350
- "rwkv.blocks.25.ln2.weight": "model-00026-of-00030.safetensors",
351
- "rwkv.blocks.26.attention.key.weight": "model-00027-of-00030.safetensors",
352
- "rwkv.blocks.26.attention.output.weight": "model-00027-of-00030.safetensors",
353
- "rwkv.blocks.26.attention.receptance.weight": "model-00027-of-00030.safetensors",
354
- "rwkv.blocks.26.attention.time_decay": "model-00027-of-00030.safetensors",
355
- "rwkv.blocks.26.attention.time_first": "model-00027-of-00030.safetensors",
356
- "rwkv.blocks.26.attention.time_mix_key": "model-00027-of-00030.safetensors",
357
- "rwkv.blocks.26.attention.time_mix_receptance": "model-00027-of-00030.safetensors",
358
- "rwkv.blocks.26.attention.time_mix_value": "model-00027-of-00030.safetensors",
359
- "rwkv.blocks.26.attention.value.weight": "model-00027-of-00030.safetensors",
360
- "rwkv.blocks.26.feed_forward.key.weight": "model-00028-of-00030.safetensors",
361
- "rwkv.blocks.26.feed_forward.receptance.weight": "model-00028-of-00030.safetensors",
362
- "rwkv.blocks.26.feed_forward.time_mix_key": "model-00027-of-00030.safetensors",
363
- "rwkv.blocks.26.feed_forward.time_mix_receptance": "model-00027-of-00030.safetensors",
364
- "rwkv.blocks.26.feed_forward.value.weight": "model-00028-of-00030.safetensors",
365
- "rwkv.blocks.26.ln1.bias": "model-00027-of-00030.safetensors",
366
- "rwkv.blocks.26.ln1.weight": "model-00027-of-00030.safetensors",
367
- "rwkv.blocks.26.ln2.bias": "model-00027-of-00030.safetensors",
368
- "rwkv.blocks.26.ln2.weight": "model-00027-of-00030.safetensors",
369
- "rwkv.blocks.27.attention.key.weight": "model-00028-of-00030.safetensors",
370
- "rwkv.blocks.27.attention.output.weight": "model-00028-of-00030.safetensors",
371
- "rwkv.blocks.27.attention.receptance.weight": "model-00028-of-00030.safetensors",
372
- "rwkv.blocks.27.attention.time_decay": "model-00028-of-00030.safetensors",
373
- "rwkv.blocks.27.attention.time_first": "model-00028-of-00030.safetensors",
374
- "rwkv.blocks.27.attention.time_mix_key": "model-00028-of-00030.safetensors",
375
- "rwkv.blocks.27.attention.time_mix_receptance": "model-00028-of-00030.safetensors",
376
- "rwkv.blocks.27.attention.time_mix_value": "model-00028-of-00030.safetensors",
377
- "rwkv.blocks.27.attention.value.weight": "model-00028-of-00030.safetensors",
378
- "rwkv.blocks.27.feed_forward.key.weight": "model-00029-of-00030.safetensors",
379
- "rwkv.blocks.27.feed_forward.receptance.weight": "model-00029-of-00030.safetensors",
380
- "rwkv.blocks.27.feed_forward.time_mix_key": "model-00028-of-00030.safetensors",
381
- "rwkv.blocks.27.feed_forward.time_mix_receptance": "model-00028-of-00030.safetensors",
382
- "rwkv.blocks.27.feed_forward.value.weight": "model-00029-of-00030.safetensors",
383
- "rwkv.blocks.27.ln1.bias": "model-00028-of-00030.safetensors",
384
- "rwkv.blocks.27.ln1.weight": "model-00028-of-00030.safetensors",
385
- "rwkv.blocks.27.ln2.bias": "model-00028-of-00030.safetensors",
386
- "rwkv.blocks.27.ln2.weight": "model-00028-of-00030.safetensors",
387
- "rwkv.blocks.3.attention.key.weight": "model-00004-of-00030.safetensors",
388
- "rwkv.blocks.3.attention.output.weight": "model-00004-of-00030.safetensors",
389
- "rwkv.blocks.3.attention.receptance.weight": "model-00004-of-00030.safetensors",
390
- "rwkv.blocks.3.attention.time_decay": "model-00004-of-00030.safetensors",
391
- "rwkv.blocks.3.attention.time_first": "model-00004-of-00030.safetensors",
392
- "rwkv.blocks.3.attention.time_mix_key": "model-00004-of-00030.safetensors",
393
- "rwkv.blocks.3.attention.time_mix_receptance": "model-00004-of-00030.safetensors",
394
- "rwkv.blocks.3.attention.time_mix_value": "model-00004-of-00030.safetensors",
395
- "rwkv.blocks.3.attention.value.weight": "model-00004-of-00030.safetensors",
396
- "rwkv.blocks.3.feed_forward.key.weight": "model-00005-of-00030.safetensors",
397
- "rwkv.blocks.3.feed_forward.receptance.weight": "model-00005-of-00030.safetensors",
398
- "rwkv.blocks.3.feed_forward.time_mix_key": "model-00004-of-00030.safetensors",
399
- "rwkv.blocks.3.feed_forward.time_mix_receptance": "model-00004-of-00030.safetensors",
400
- "rwkv.blocks.3.feed_forward.value.weight": "model-00005-of-00030.safetensors",
401
- "rwkv.blocks.3.ln1.bias": "model-00004-of-00030.safetensors",
402
- "rwkv.blocks.3.ln1.weight": "model-00004-of-00030.safetensors",
403
- "rwkv.blocks.3.ln2.bias": "model-00004-of-00030.safetensors",
404
- "rwkv.blocks.3.ln2.weight": "model-00004-of-00030.safetensors",
405
- "rwkv.blocks.4.attention.key.weight": "model-00005-of-00030.safetensors",
406
- "rwkv.blocks.4.attention.output.weight": "model-00005-of-00030.safetensors",
407
- "rwkv.blocks.4.attention.receptance.weight": "model-00005-of-00030.safetensors",
408
- "rwkv.blocks.4.attention.time_decay": "model-00005-of-00030.safetensors",
409
- "rwkv.blocks.4.attention.time_first": "model-00005-of-00030.safetensors",
410
- "rwkv.blocks.4.attention.time_mix_key": "model-00005-of-00030.safetensors",
411
- "rwkv.blocks.4.attention.time_mix_receptance": "model-00005-of-00030.safetensors",
412
- "rwkv.blocks.4.attention.time_mix_value": "model-00005-of-00030.safetensors",
413
- "rwkv.blocks.4.attention.value.weight": "model-00005-of-00030.safetensors",
414
- "rwkv.blocks.4.feed_forward.key.weight": "model-00006-of-00030.safetensors",
415
- "rwkv.blocks.4.feed_forward.receptance.weight": "model-00006-of-00030.safetensors",
416
- "rwkv.blocks.4.feed_forward.time_mix_key": "model-00005-of-00030.safetensors",
417
- "rwkv.blocks.4.feed_forward.time_mix_receptance": "model-00005-of-00030.safetensors",
418
- "rwkv.blocks.4.feed_forward.value.weight": "model-00006-of-00030.safetensors",
419
- "rwkv.blocks.4.ln1.bias": "model-00005-of-00030.safetensors",
420
- "rwkv.blocks.4.ln1.weight": "model-00005-of-00030.safetensors",
421
- "rwkv.blocks.4.ln2.bias": "model-00005-of-00030.safetensors",
422
- "rwkv.blocks.4.ln2.weight": "model-00005-of-00030.safetensors",
423
- "rwkv.blocks.5.attention.key.weight": "model-00006-of-00030.safetensors",
424
- "rwkv.blocks.5.attention.output.weight": "model-00006-of-00030.safetensors",
425
- "rwkv.blocks.5.attention.receptance.weight": "model-00006-of-00030.safetensors",
426
- "rwkv.blocks.5.attention.time_decay": "model-00006-of-00030.safetensors",
427
- "rwkv.blocks.5.attention.time_first": "model-00006-of-00030.safetensors",
428
- "rwkv.blocks.5.attention.time_mix_key": "model-00006-of-00030.safetensors",
429
- "rwkv.blocks.5.attention.time_mix_receptance": "model-00006-of-00030.safetensors",
430
- "rwkv.blocks.5.attention.time_mix_value": "model-00006-of-00030.safetensors",
431
- "rwkv.blocks.5.attention.value.weight": "model-00006-of-00030.safetensors",
432
- "rwkv.blocks.5.feed_forward.key.weight": "model-00007-of-00030.safetensors",
433
- "rwkv.blocks.5.feed_forward.receptance.weight": "model-00007-of-00030.safetensors",
434
- "rwkv.blocks.5.feed_forward.time_mix_key": "model-00006-of-00030.safetensors",
435
- "rwkv.blocks.5.feed_forward.time_mix_receptance": "model-00006-of-00030.safetensors",
436
- "rwkv.blocks.5.feed_forward.value.weight": "model-00007-of-00030.safetensors",
437
- "rwkv.blocks.5.ln1.bias": "model-00006-of-00030.safetensors",
438
- "rwkv.blocks.5.ln1.weight": "model-00006-of-00030.safetensors",
439
- "rwkv.blocks.5.ln2.bias": "model-00006-of-00030.safetensors",
440
- "rwkv.blocks.5.ln2.weight": "model-00006-of-00030.safetensors",
441
- "rwkv.blocks.6.attention.key.weight": "model-00007-of-00030.safetensors",
442
- "rwkv.blocks.6.attention.output.weight": "model-00007-of-00030.safetensors",
443
- "rwkv.blocks.6.attention.receptance.weight": "model-00007-of-00030.safetensors",
444
- "rwkv.blocks.6.attention.time_decay": "model-00007-of-00030.safetensors",
445
- "rwkv.blocks.6.attention.time_first": "model-00007-of-00030.safetensors",
446
- "rwkv.blocks.6.attention.time_mix_key": "model-00007-of-00030.safetensors",
447
- "rwkv.blocks.6.attention.time_mix_receptance": "model-00007-of-00030.safetensors",
448
- "rwkv.blocks.6.attention.time_mix_value": "model-00007-of-00030.safetensors",
449
- "rwkv.blocks.6.attention.value.weight": "model-00007-of-00030.safetensors",
450
- "rwkv.blocks.6.feed_forward.key.weight": "model-00008-of-00030.safetensors",
451
- "rwkv.blocks.6.feed_forward.receptance.weight": "model-00008-of-00030.safetensors",
452
- "rwkv.blocks.6.feed_forward.time_mix_key": "model-00007-of-00030.safetensors",
453
- "rwkv.blocks.6.feed_forward.time_mix_receptance": "model-00007-of-00030.safetensors",
454
- "rwkv.blocks.6.feed_forward.value.weight": "model-00008-of-00030.safetensors",
455
- "rwkv.blocks.6.ln1.bias": "model-00007-of-00030.safetensors",
456
- "rwkv.blocks.6.ln1.weight": "model-00007-of-00030.safetensors",
457
- "rwkv.blocks.6.ln2.bias": "model-00007-of-00030.safetensors",
458
- "rwkv.blocks.6.ln2.weight": "model-00007-of-00030.safetensors",
459
- "rwkv.blocks.7.attention.key.weight": "model-00008-of-00030.safetensors",
460
- "rwkv.blocks.7.attention.output.weight": "model-00008-of-00030.safetensors",
461
- "rwkv.blocks.7.attention.receptance.weight": "model-00008-of-00030.safetensors",
462
- "rwkv.blocks.7.attention.time_decay": "model-00008-of-00030.safetensors",
463
- "rwkv.blocks.7.attention.time_first": "model-00008-of-00030.safetensors",
464
- "rwkv.blocks.7.attention.time_mix_key": "model-00008-of-00030.safetensors",
465
- "rwkv.blocks.7.attention.time_mix_receptance": "model-00008-of-00030.safetensors",
466
- "rwkv.blocks.7.attention.time_mix_value": "model-00008-of-00030.safetensors",
467
- "rwkv.blocks.7.attention.value.weight": "model-00008-of-00030.safetensors",
468
- "rwkv.blocks.7.feed_forward.key.weight": "model-00009-of-00030.safetensors",
469
- "rwkv.blocks.7.feed_forward.receptance.weight": "model-00009-of-00030.safetensors",
470
- "rwkv.blocks.7.feed_forward.time_mix_key": "model-00008-of-00030.safetensors",
471
- "rwkv.blocks.7.feed_forward.time_mix_receptance": "model-00008-of-00030.safetensors",
472
- "rwkv.blocks.7.feed_forward.value.weight": "model-00009-of-00030.safetensors",
473
- "rwkv.blocks.7.ln1.bias": "model-00008-of-00030.safetensors",
474
- "rwkv.blocks.7.ln1.weight": "model-00008-of-00030.safetensors",
475
- "rwkv.blocks.7.ln2.bias": "model-00008-of-00030.safetensors",
476
- "rwkv.blocks.7.ln2.weight": "model-00008-of-00030.safetensors",
477
- "rwkv.blocks.8.attention.key.weight": "model-00009-of-00030.safetensors",
478
- "rwkv.blocks.8.attention.output.weight": "model-00009-of-00030.safetensors",
479
- "rwkv.blocks.8.attention.receptance.weight": "model-00009-of-00030.safetensors",
480
- "rwkv.blocks.8.attention.time_decay": "model-00009-of-00030.safetensors",
481
- "rwkv.blocks.8.attention.time_first": "model-00009-of-00030.safetensors",
482
- "rwkv.blocks.8.attention.time_mix_key": "model-00009-of-00030.safetensors",
483
- "rwkv.blocks.8.attention.time_mix_receptance": "model-00009-of-00030.safetensors",
484
- "rwkv.blocks.8.attention.time_mix_value": "model-00009-of-00030.safetensors",
485
- "rwkv.blocks.8.attention.value.weight": "model-00009-of-00030.safetensors",
486
- "rwkv.blocks.8.feed_forward.key.weight": "model-00010-of-00030.safetensors",
487
- "rwkv.blocks.8.feed_forward.receptance.weight": "model-00010-of-00030.safetensors",
488
- "rwkv.blocks.8.feed_forward.time_mix_key": "model-00009-of-00030.safetensors",
489
- "rwkv.blocks.8.feed_forward.time_mix_receptance": "model-00009-of-00030.safetensors",
490
- "rwkv.blocks.8.feed_forward.value.weight": "model-00010-of-00030.safetensors",
491
- "rwkv.blocks.8.ln1.bias": "model-00009-of-00030.safetensors",
492
- "rwkv.blocks.8.ln1.weight": "model-00009-of-00030.safetensors",
493
- "rwkv.blocks.8.ln2.bias": "model-00009-of-00030.safetensors",
494
- "rwkv.blocks.8.ln2.weight": "model-00009-of-00030.safetensors",
495
- "rwkv.blocks.9.attention.key.weight": "model-00010-of-00030.safetensors",
496
- "rwkv.blocks.9.attention.output.weight": "model-00010-of-00030.safetensors",
497
- "rwkv.blocks.9.attention.receptance.weight": "model-00010-of-00030.safetensors",
498
- "rwkv.blocks.9.attention.time_decay": "model-00010-of-00030.safetensors",
499
- "rwkv.blocks.9.attention.time_first": "model-00010-of-00030.safetensors",
500
- "rwkv.blocks.9.attention.time_mix_key": "model-00010-of-00030.safetensors",
501
- "rwkv.blocks.9.attention.time_mix_receptance": "model-00010-of-00030.safetensors",
502
- "rwkv.blocks.9.attention.time_mix_value": "model-00010-of-00030.safetensors",
503
- "rwkv.blocks.9.attention.value.weight": "model-00010-of-00030.safetensors",
504
- "rwkv.blocks.9.feed_forward.key.weight": "model-00011-of-00030.safetensors",
505
- "rwkv.blocks.9.feed_forward.receptance.weight": "model-00011-of-00030.safetensors",
506
- "rwkv.blocks.9.feed_forward.time_mix_key": "model-00010-of-00030.safetensors",
507
- "rwkv.blocks.9.feed_forward.time_mix_receptance": "model-00010-of-00030.safetensors",
508
- "rwkv.blocks.9.feed_forward.value.weight": "model-00011-of-00030.safetensors",
509
- "rwkv.blocks.9.ln1.bias": "model-00010-of-00030.safetensors",
510
- "rwkv.blocks.9.ln1.weight": "model-00010-of-00030.safetensors",
511
- "rwkv.blocks.9.ln2.bias": "model-00010-of-00030.safetensors",
512
- "rwkv.blocks.9.ln2.weight": "model-00010-of-00030.safetensors",
513
- "rwkv.embeddings.weight": "model-00001-of-00030.safetensors",
514
- "rwkv.ln_out.bias": "model-00029-of-00030.safetensors",
515
- "rwkv.ln_out.weight": "model-00029-of-00030.safetensors"
516
  }
517
  }
 
1
  {
2
  "metadata": {
3
+ "total_size": 13068337152
4
  },
5
  "weight_map": {
6
+ "head.weight": "model-00015-of-00015.safetensors",
7
+ "rwkv.blocks.0.attention.key.weight": "model-00001-of-00015.safetensors",
8
+ "rwkv.blocks.0.attention.output.weight": "model-00001-of-00015.safetensors",
9
+ "rwkv.blocks.0.attention.receptance.weight": "model-00001-of-00015.safetensors",
10
+ "rwkv.blocks.0.attention.time_decay": "model-00001-of-00015.safetensors",
11
+ "rwkv.blocks.0.attention.time_first": "model-00001-of-00015.safetensors",
12
+ "rwkv.blocks.0.attention.time_mix_key": "model-00001-of-00015.safetensors",
13
+ "rwkv.blocks.0.attention.time_mix_receptance": "model-00001-of-00015.safetensors",
14
+ "rwkv.blocks.0.attention.time_mix_value": "model-00001-of-00015.safetensors",
15
+ "rwkv.blocks.0.attention.value.weight": "model-00001-of-00015.safetensors",
16
+ "rwkv.blocks.0.feed_forward.key.weight": "model-00001-of-00015.safetensors",
17
+ "rwkv.blocks.0.feed_forward.receptance.weight": "model-00001-of-00015.safetensors",
18
+ "rwkv.blocks.0.feed_forward.time_mix_key": "model-00001-of-00015.safetensors",
19
+ "rwkv.blocks.0.feed_forward.time_mix_receptance": "model-00001-of-00015.safetensors",
20
+ "rwkv.blocks.0.feed_forward.value.weight": "model-00001-of-00015.safetensors",
21
+ "rwkv.blocks.0.ln1.bias": "model-00001-of-00015.safetensors",
22
+ "rwkv.blocks.0.ln1.weight": "model-00001-of-00015.safetensors",
23
+ "rwkv.blocks.0.ln2.bias": "model-00001-of-00015.safetensors",
24
+ "rwkv.blocks.0.ln2.weight": "model-00001-of-00015.safetensors",
25
+ "rwkv.blocks.0.pre_ln.bias": "model-00001-of-00015.safetensors",
26
+ "rwkv.blocks.0.pre_ln.weight": "model-00001-of-00015.safetensors",
27
+ "rwkv.blocks.1.attention.key.weight": "model-00001-of-00015.safetensors",
28
+ "rwkv.blocks.1.attention.output.weight": "model-00001-of-00015.safetensors",
29
+ "rwkv.blocks.1.attention.receptance.weight": "model-00001-of-00015.safetensors",
30
+ "rwkv.blocks.1.attention.time_decay": "model-00001-of-00015.safetensors",
31
+ "rwkv.blocks.1.attention.time_first": "model-00001-of-00015.safetensors",
32
+ "rwkv.blocks.1.attention.time_mix_key": "model-00001-of-00015.safetensors",
33
+ "rwkv.blocks.1.attention.time_mix_receptance": "model-00001-of-00015.safetensors",
34
+ "rwkv.blocks.1.attention.time_mix_value": "model-00001-of-00015.safetensors",
35
+ "rwkv.blocks.1.attention.value.weight": "model-00001-of-00015.safetensors",
36
+ "rwkv.blocks.1.feed_forward.key.weight": "model-00002-of-00015.safetensors",
37
+ "rwkv.blocks.1.feed_forward.receptance.weight": "model-00002-of-00015.safetensors",
38
+ "rwkv.blocks.1.feed_forward.time_mix_key": "model-00001-of-00015.safetensors",
39
+ "rwkv.blocks.1.feed_forward.time_mix_receptance": "model-00001-of-00015.safetensors",
40
+ "rwkv.blocks.1.feed_forward.value.weight": "model-00002-of-00015.safetensors",
41
+ "rwkv.blocks.1.ln1.bias": "model-00001-of-00015.safetensors",
42
+ "rwkv.blocks.1.ln1.weight": "model-00001-of-00015.safetensors",
43
+ "rwkv.blocks.1.ln2.bias": "model-00001-of-00015.safetensors",
44
+ "rwkv.blocks.1.ln2.weight": "model-00001-of-00015.safetensors",
45
+ "rwkv.blocks.10.attention.key.weight": "model-00006-of-00015.safetensors",
46
+ "rwkv.blocks.10.attention.output.weight": "model-00006-of-00015.safetensors",
47
+ "rwkv.blocks.10.attention.receptance.weight": "model-00006-of-00015.safetensors",
48
+ "rwkv.blocks.10.attention.time_decay": "model-00006-of-00015.safetensors",
49
+ "rwkv.blocks.10.attention.time_first": "model-00006-of-00015.safetensors",
50
+ "rwkv.blocks.10.attention.time_mix_key": "model-00006-of-00015.safetensors",
51
+ "rwkv.blocks.10.attention.time_mix_receptance": "model-00006-of-00015.safetensors",
52
+ "rwkv.blocks.10.attention.time_mix_value": "model-00006-of-00015.safetensors",
53
+ "rwkv.blocks.10.attention.value.weight": "model-00006-of-00015.safetensors",
54
+ "rwkv.blocks.10.feed_forward.key.weight": "model-00006-of-00015.safetensors",
55
+ "rwkv.blocks.10.feed_forward.receptance.weight": "model-00006-of-00015.safetensors",
56
+ "rwkv.blocks.10.feed_forward.time_mix_key": "model-00006-of-00015.safetensors",
57
+ "rwkv.blocks.10.feed_forward.time_mix_receptance": "model-00006-of-00015.safetensors",
58
+ "rwkv.blocks.10.feed_forward.value.weight": "model-00006-of-00015.safetensors",
59
+ "rwkv.blocks.10.ln1.bias": "model-00006-of-00015.safetensors",
60
+ "rwkv.blocks.10.ln1.weight": "model-00006-of-00015.safetensors",
61
+ "rwkv.blocks.10.ln2.bias": "model-00006-of-00015.safetensors",
62
+ "rwkv.blocks.10.ln2.weight": "model-00006-of-00015.safetensors",
63
+ "rwkv.blocks.11.attention.key.weight": "model-00006-of-00015.safetensors",
64
+ "rwkv.blocks.11.attention.output.weight": "model-00006-of-00015.safetensors",
65
+ "rwkv.blocks.11.attention.receptance.weight": "model-00006-of-00015.safetensors",
66
+ "rwkv.blocks.11.attention.time_decay": "model-00006-of-00015.safetensors",
67
+ "rwkv.blocks.11.attention.time_first": "model-00006-of-00015.safetensors",
68
+ "rwkv.blocks.11.attention.time_mix_key": "model-00006-of-00015.safetensors",
69
+ "rwkv.blocks.11.attention.time_mix_receptance": "model-00006-of-00015.safetensors",
70
+ "rwkv.blocks.11.attention.time_mix_value": "model-00006-of-00015.safetensors",
71
+ "rwkv.blocks.11.attention.value.weight": "model-00006-of-00015.safetensors",
72
+ "rwkv.blocks.11.feed_forward.key.weight": "model-00007-of-00015.safetensors",
73
+ "rwkv.blocks.11.feed_forward.receptance.weight": "model-00007-of-00015.safetensors",
74
+ "rwkv.blocks.11.feed_forward.time_mix_key": "model-00006-of-00015.safetensors",
75
+ "rwkv.blocks.11.feed_forward.time_mix_receptance": "model-00006-of-00015.safetensors",
76
+ "rwkv.blocks.11.feed_forward.value.weight": "model-00007-of-00015.safetensors",
77
+ "rwkv.blocks.11.ln1.bias": "model-00006-of-00015.safetensors",
78
+ "rwkv.blocks.11.ln1.weight": "model-00006-of-00015.safetensors",
79
+ "rwkv.blocks.11.ln2.bias": "model-00006-of-00015.safetensors",
80
+ "rwkv.blocks.11.ln2.weight": "model-00006-of-00015.safetensors",
81
+ "rwkv.blocks.12.attention.key.weight": "model-00007-of-00015.safetensors",
82
+ "rwkv.blocks.12.attention.output.weight": "model-00007-of-00015.safetensors",
83
+ "rwkv.blocks.12.attention.receptance.weight": "model-00007-of-00015.safetensors",
84
+ "rwkv.blocks.12.attention.time_decay": "model-00007-of-00015.safetensors",
85
+ "rwkv.blocks.12.attention.time_first": "model-00007-of-00015.safetensors",
86
+ "rwkv.blocks.12.attention.time_mix_key": "model-00007-of-00015.safetensors",
87
+ "rwkv.blocks.12.attention.time_mix_receptance": "model-00007-of-00015.safetensors",
88
+ "rwkv.blocks.12.attention.time_mix_value": "model-00007-of-00015.safetensors",
89
+ "rwkv.blocks.12.attention.value.weight": "model-00007-of-00015.safetensors",
90
+ "rwkv.blocks.12.feed_forward.key.weight": "model-00007-of-00015.safetensors",
91
+ "rwkv.blocks.12.feed_forward.receptance.weight": "model-00007-of-00015.safetensors",
92
+ "rwkv.blocks.12.feed_forward.time_mix_key": "model-00007-of-00015.safetensors",
93
+ "rwkv.blocks.12.feed_forward.time_mix_receptance": "model-00007-of-00015.safetensors",
94
+ "rwkv.blocks.12.feed_forward.value.weight": "model-00007-of-00015.safetensors",
95
+ "rwkv.blocks.12.ln1.bias": "model-00007-of-00015.safetensors",
96
+ "rwkv.blocks.12.ln1.weight": "model-00007-of-00015.safetensors",
97
+ "rwkv.blocks.12.ln2.bias": "model-00007-of-00015.safetensors",
98
+ "rwkv.blocks.12.ln2.weight": "model-00007-of-00015.safetensors",
99
+ "rwkv.blocks.13.attention.key.weight": "model-00007-of-00015.safetensors",
100
+ "rwkv.blocks.13.attention.output.weight": "model-00007-of-00015.safetensors",
101
+ "rwkv.blocks.13.attention.receptance.weight": "model-00007-of-00015.safetensors",
102
+ "rwkv.blocks.13.attention.time_decay": "model-00007-of-00015.safetensors",
103
+ "rwkv.blocks.13.attention.time_first": "model-00007-of-00015.safetensors",
104
+ "rwkv.blocks.13.attention.time_mix_key": "model-00007-of-00015.safetensors",
105
+ "rwkv.blocks.13.attention.time_mix_receptance": "model-00007-of-00015.safetensors",
106
+ "rwkv.blocks.13.attention.time_mix_value": "model-00007-of-00015.safetensors",
107
+ "rwkv.blocks.13.attention.value.weight": "model-00007-of-00015.safetensors",
108
+ "rwkv.blocks.13.feed_forward.key.weight": "model-00008-of-00015.safetensors",
109
+ "rwkv.blocks.13.feed_forward.receptance.weight": "model-00008-of-00015.safetensors",
110
+ "rwkv.blocks.13.feed_forward.time_mix_key": "model-00007-of-00015.safetensors",
111
+ "rwkv.blocks.13.feed_forward.time_mix_receptance": "model-00007-of-00015.safetensors",
112
+ "rwkv.blocks.13.feed_forward.value.weight": "model-00008-of-00015.safetensors",
113
+ "rwkv.blocks.13.ln1.bias": "model-00007-of-00015.safetensors",
114
+ "rwkv.blocks.13.ln1.weight": "model-00007-of-00015.safetensors",
115
+ "rwkv.blocks.13.ln2.bias": "model-00007-of-00015.safetensors",
116
+ "rwkv.blocks.13.ln2.weight": "model-00007-of-00015.safetensors",
117
+ "rwkv.blocks.14.attention.key.weight": "model-00008-of-00015.safetensors",
118
+ "rwkv.blocks.14.attention.output.weight": "model-00008-of-00015.safetensors",
119
+ "rwkv.blocks.14.attention.receptance.weight": "model-00008-of-00015.safetensors",
120
+ "rwkv.blocks.14.attention.time_decay": "model-00008-of-00015.safetensors",
121
+ "rwkv.blocks.14.attention.time_first": "model-00008-of-00015.safetensors",
122
+ "rwkv.blocks.14.attention.time_mix_key": "model-00008-of-00015.safetensors",
123
+ "rwkv.blocks.14.attention.time_mix_receptance": "model-00008-of-00015.safetensors",
124
+ "rwkv.blocks.14.attention.time_mix_value": "model-00008-of-00015.safetensors",
125
+ "rwkv.blocks.14.attention.value.weight": "model-00008-of-00015.safetensors",
126
+ "rwkv.blocks.14.feed_forward.key.weight": "model-00008-of-00015.safetensors",
127
+ "rwkv.blocks.14.feed_forward.receptance.weight": "model-00008-of-00015.safetensors",
128
+ "rwkv.blocks.14.feed_forward.time_mix_key": "model-00008-of-00015.safetensors",
129
+ "rwkv.blocks.14.feed_forward.time_mix_receptance": "model-00008-of-00015.safetensors",
130
+ "rwkv.blocks.14.feed_forward.value.weight": "model-00008-of-00015.safetensors",
131
+ "rwkv.blocks.14.ln1.bias": "model-00008-of-00015.safetensors",
132
+ "rwkv.blocks.14.ln1.weight": "model-00008-of-00015.safetensors",
133
+ "rwkv.blocks.14.ln2.bias": "model-00008-of-00015.safetensors",
134
+ "rwkv.blocks.14.ln2.weight": "model-00008-of-00015.safetensors",
135
+ "rwkv.blocks.15.attention.key.weight": "model-00008-of-00015.safetensors",
136
+ "rwkv.blocks.15.attention.output.weight": "model-00008-of-00015.safetensors",
137
+ "rwkv.blocks.15.attention.receptance.weight": "model-00008-of-00015.safetensors",
138
+ "rwkv.blocks.15.attention.time_decay": "model-00008-of-00015.safetensors",
139
+ "rwkv.blocks.15.attention.time_first": "model-00008-of-00015.safetensors",
140
+ "rwkv.blocks.15.attention.time_mix_key": "model-00008-of-00015.safetensors",
141
+ "rwkv.blocks.15.attention.time_mix_receptance": "model-00008-of-00015.safetensors",
142
+ "rwkv.blocks.15.attention.time_mix_value": "model-00008-of-00015.safetensors",
143
+ "rwkv.blocks.15.attention.value.weight": "model-00008-of-00015.safetensors",
144
+ "rwkv.blocks.15.feed_forward.key.weight": "model-00009-of-00015.safetensors",
145
+ "rwkv.blocks.15.feed_forward.receptance.weight": "model-00009-of-00015.safetensors",
146
+ "rwkv.blocks.15.feed_forward.time_mix_key": "model-00008-of-00015.safetensors",
147
+ "rwkv.blocks.15.feed_forward.time_mix_receptance": "model-00008-of-00015.safetensors",
148
+ "rwkv.blocks.15.feed_forward.value.weight": "model-00009-of-00015.safetensors",
149
+ "rwkv.blocks.15.ln1.bias": "model-00008-of-00015.safetensors",
150
+ "rwkv.blocks.15.ln1.weight": "model-00008-of-00015.safetensors",
151
+ "rwkv.blocks.15.ln2.bias": "model-00008-of-00015.safetensors",
152
+ "rwkv.blocks.15.ln2.weight": "model-00008-of-00015.safetensors",
153
+ "rwkv.blocks.16.attention.key.weight": "model-00009-of-00015.safetensors",
154
+ "rwkv.blocks.16.attention.output.weight": "model-00009-of-00015.safetensors",
155
+ "rwkv.blocks.16.attention.receptance.weight": "model-00009-of-00015.safetensors",
156
+ "rwkv.blocks.16.attention.time_decay": "model-00009-of-00015.safetensors",
157
+ "rwkv.blocks.16.attention.time_first": "model-00009-of-00015.safetensors",
158
+ "rwkv.blocks.16.attention.time_mix_key": "model-00009-of-00015.safetensors",
159
+ "rwkv.blocks.16.attention.time_mix_receptance": "model-00009-of-00015.safetensors",
160
+ "rwkv.blocks.16.attention.time_mix_value": "model-00009-of-00015.safetensors",
161
+ "rwkv.blocks.16.attention.value.weight": "model-00009-of-00015.safetensors",
162
+ "rwkv.blocks.16.feed_forward.key.weight": "model-00009-of-00015.safetensors",
163
+ "rwkv.blocks.16.feed_forward.receptance.weight": "model-00009-of-00015.safetensors",
164
+ "rwkv.blocks.16.feed_forward.time_mix_key": "model-00009-of-00015.safetensors",
165
+ "rwkv.blocks.16.feed_forward.time_mix_receptance": "model-00009-of-00015.safetensors",
166
+ "rwkv.blocks.16.feed_forward.value.weight": "model-00009-of-00015.safetensors",
167
+ "rwkv.blocks.16.ln1.bias": "model-00009-of-00015.safetensors",
168
+ "rwkv.blocks.16.ln1.weight": "model-00009-of-00015.safetensors",
169
+ "rwkv.blocks.16.ln2.bias": "model-00009-of-00015.safetensors",
170
+ "rwkv.blocks.16.ln2.weight": "model-00009-of-00015.safetensors",
171
+ "rwkv.blocks.17.attention.key.weight": "model-00009-of-00015.safetensors",
172
+ "rwkv.blocks.17.attention.output.weight": "model-00009-of-00015.safetensors",
173
+ "rwkv.blocks.17.attention.receptance.weight": "model-00009-of-00015.safetensors",
174
+ "rwkv.blocks.17.attention.time_decay": "model-00009-of-00015.safetensors",
175
+ "rwkv.blocks.17.attention.time_first": "model-00009-of-00015.safetensors",
176
+ "rwkv.blocks.17.attention.time_mix_key": "model-00009-of-00015.safetensors",
177
+ "rwkv.blocks.17.attention.time_mix_receptance": "model-00009-of-00015.safetensors",
178
+ "rwkv.blocks.17.attention.time_mix_value": "model-00009-of-00015.safetensors",
179
+ "rwkv.blocks.17.attention.value.weight": "model-00009-of-00015.safetensors",
180
+ "rwkv.blocks.17.feed_forward.key.weight": "model-00010-of-00015.safetensors",
181
+ "rwkv.blocks.17.feed_forward.receptance.weight": "model-00010-of-00015.safetensors",
182
+ "rwkv.blocks.17.feed_forward.time_mix_key": "model-00009-of-00015.safetensors",
183
+ "rwkv.blocks.17.feed_forward.time_mix_receptance": "model-00009-of-00015.safetensors",
184
+ "rwkv.blocks.17.feed_forward.value.weight": "model-00010-of-00015.safetensors",
185
+ "rwkv.blocks.17.ln1.bias": "model-00009-of-00015.safetensors",
186
+ "rwkv.blocks.17.ln1.weight": "model-00009-of-00015.safetensors",
187
+ "rwkv.blocks.17.ln2.bias": "model-00009-of-00015.safetensors",
188
+ "rwkv.blocks.17.ln2.weight": "model-00009-of-00015.safetensors",
189
+ "rwkv.blocks.18.attention.key.weight": "model-00010-of-00015.safetensors",
190
+ "rwkv.blocks.18.attention.output.weight": "model-00010-of-00015.safetensors",
191
+ "rwkv.blocks.18.attention.receptance.weight": "model-00010-of-00015.safetensors",
192
+ "rwkv.blocks.18.attention.time_decay": "model-00010-of-00015.safetensors",
193
+ "rwkv.blocks.18.attention.time_first": "model-00010-of-00015.safetensors",
194
+ "rwkv.blocks.18.attention.time_mix_key": "model-00010-of-00015.safetensors",
195
+ "rwkv.blocks.18.attention.time_mix_receptance": "model-00010-of-00015.safetensors",
196
+ "rwkv.blocks.18.attention.time_mix_value": "model-00010-of-00015.safetensors",
197
+ "rwkv.blocks.18.attention.value.weight": "model-00010-of-00015.safetensors",
198
+ "rwkv.blocks.18.feed_forward.key.weight": "model-00010-of-00015.safetensors",
199
+ "rwkv.blocks.18.feed_forward.receptance.weight": "model-00010-of-00015.safetensors",
200
+ "rwkv.blocks.18.feed_forward.time_mix_key": "model-00010-of-00015.safetensors",
201
+ "rwkv.blocks.18.feed_forward.time_mix_receptance": "model-00010-of-00015.safetensors",
202
+ "rwkv.blocks.18.feed_forward.value.weight": "model-00010-of-00015.safetensors",
203
+ "rwkv.blocks.18.ln1.bias": "model-00010-of-00015.safetensors",
204
+ "rwkv.blocks.18.ln1.weight": "model-00010-of-00015.safetensors",
205
+ "rwkv.blocks.18.ln2.bias": "model-00010-of-00015.safetensors",
206
+ "rwkv.blocks.18.ln2.weight": "model-00010-of-00015.safetensors",
207
+ "rwkv.blocks.19.attention.key.weight": "model-00010-of-00015.safetensors",
208
+ "rwkv.blocks.19.attention.output.weight": "model-00010-of-00015.safetensors",
209
+ "rwkv.blocks.19.attention.receptance.weight": "model-00010-of-00015.safetensors",
210
+ "rwkv.blocks.19.attention.time_decay": "model-00010-of-00015.safetensors",
211
+ "rwkv.blocks.19.attention.time_first": "model-00010-of-00015.safetensors",
212
+ "rwkv.blocks.19.attention.time_mix_key": "model-00010-of-00015.safetensors",
213
+ "rwkv.blocks.19.attention.time_mix_receptance": "model-00010-of-00015.safetensors",
214
+ "rwkv.blocks.19.attention.time_mix_value": "model-00010-of-00015.safetensors",
215
+ "rwkv.blocks.19.attention.value.weight": "model-00010-of-00015.safetensors",
216
+ "rwkv.blocks.19.feed_forward.key.weight": "model-00011-of-00015.safetensors",
217
+ "rwkv.blocks.19.feed_forward.receptance.weight": "model-00011-of-00015.safetensors",
218
+ "rwkv.blocks.19.feed_forward.time_mix_key": "model-00010-of-00015.safetensors",
219
+ "rwkv.blocks.19.feed_forward.time_mix_receptance": "model-00010-of-00015.safetensors",
220
+ "rwkv.blocks.19.feed_forward.value.weight": "model-00011-of-00015.safetensors",
221
+ "rwkv.blocks.19.ln1.bias": "model-00010-of-00015.safetensors",
222
+ "rwkv.blocks.19.ln1.weight": "model-00010-of-00015.safetensors",
223
+ "rwkv.blocks.19.ln2.bias": "model-00010-of-00015.safetensors",
224
+ "rwkv.blocks.19.ln2.weight": "model-00010-of-00015.safetensors",
225
+ "rwkv.blocks.2.attention.key.weight": "model-00002-of-00015.safetensors",
226
+ "rwkv.blocks.2.attention.output.weight": "model-00002-of-00015.safetensors",
227
+ "rwkv.blocks.2.attention.receptance.weight": "model-00002-of-00015.safetensors",
228
+ "rwkv.blocks.2.attention.time_decay": "model-00002-of-00015.safetensors",
229
+ "rwkv.blocks.2.attention.time_first": "model-00002-of-00015.safetensors",
230
+ "rwkv.blocks.2.attention.time_mix_key": "model-00002-of-00015.safetensors",
231
+ "rwkv.blocks.2.attention.time_mix_receptance": "model-00002-of-00015.safetensors",
232
+ "rwkv.blocks.2.attention.time_mix_value": "model-00002-of-00015.safetensors",
233
+ "rwkv.blocks.2.attention.value.weight": "model-00002-of-00015.safetensors",
234
+ "rwkv.blocks.2.feed_forward.key.weight": "model-00002-of-00015.safetensors",
235
+ "rwkv.blocks.2.feed_forward.receptance.weight": "model-00002-of-00015.safetensors",
236
+ "rwkv.blocks.2.feed_forward.time_mix_key": "model-00002-of-00015.safetensors",
237
+ "rwkv.blocks.2.feed_forward.time_mix_receptance": "model-00002-of-00015.safetensors",
238
+ "rwkv.blocks.2.feed_forward.value.weight": "model-00002-of-00015.safetensors",
239
+ "rwkv.blocks.2.ln1.bias": "model-00002-of-00015.safetensors",
240
+ "rwkv.blocks.2.ln1.weight": "model-00002-of-00015.safetensors",
241
+ "rwkv.blocks.2.ln2.bias": "model-00002-of-00015.safetensors",
242
+ "rwkv.blocks.2.ln2.weight": "model-00002-of-00015.safetensors",
243
+ "rwkv.blocks.20.attention.key.weight": "model-00011-of-00015.safetensors",
244
+ "rwkv.blocks.20.attention.output.weight": "model-00011-of-00015.safetensors",
245
+ "rwkv.blocks.20.attention.receptance.weight": "model-00011-of-00015.safetensors",
246
+ "rwkv.blocks.20.attention.time_decay": "model-00011-of-00015.safetensors",
247
+ "rwkv.blocks.20.attention.time_first": "model-00011-of-00015.safetensors",
248
+ "rwkv.blocks.20.attention.time_mix_key": "model-00011-of-00015.safetensors",
249
+ "rwkv.blocks.20.attention.time_mix_receptance": "model-00011-of-00015.safetensors",
250
+ "rwkv.blocks.20.attention.time_mix_value": "model-00011-of-00015.safetensors",
251
+ "rwkv.blocks.20.attention.value.weight": "model-00011-of-00015.safetensors",
252
+ "rwkv.blocks.20.feed_forward.key.weight": "model-00011-of-00015.safetensors",
253
+ "rwkv.blocks.20.feed_forward.receptance.weight": "model-00011-of-00015.safetensors",
254
+ "rwkv.blocks.20.feed_forward.time_mix_key": "model-00011-of-00015.safetensors",
255
+ "rwkv.blocks.20.feed_forward.time_mix_receptance": "model-00011-of-00015.safetensors",
256
+ "rwkv.blocks.20.feed_forward.value.weight": "model-00011-of-00015.safetensors",
257
+ "rwkv.blocks.20.ln1.bias": "model-00011-of-00015.safetensors",
258
+ "rwkv.blocks.20.ln1.weight": "model-00011-of-00015.safetensors",
259
+ "rwkv.blocks.20.ln2.bias": "model-00011-of-00015.safetensors",
260
+ "rwkv.blocks.20.ln2.weight": "model-00011-of-00015.safetensors",
261
+ "rwkv.blocks.21.attention.key.weight": "model-00011-of-00015.safetensors",
262
+ "rwkv.blocks.21.attention.output.weight": "model-00011-of-00015.safetensors",
263
+ "rwkv.blocks.21.attention.receptance.weight": "model-00011-of-00015.safetensors",
264
+ "rwkv.blocks.21.attention.time_decay": "model-00011-of-00015.safetensors",
265
+ "rwkv.blocks.21.attention.time_first": "model-00011-of-00015.safetensors",
266
+ "rwkv.blocks.21.attention.time_mix_key": "model-00011-of-00015.safetensors",
267
+ "rwkv.blocks.21.attention.time_mix_receptance": "model-00011-of-00015.safetensors",
268
+ "rwkv.blocks.21.attention.time_mix_value": "model-00011-of-00015.safetensors",
269
+ "rwkv.blocks.21.attention.value.weight": "model-00011-of-00015.safetensors",
270
+ "rwkv.blocks.21.feed_forward.key.weight": "model-00012-of-00015.safetensors",
271
+ "rwkv.blocks.21.feed_forward.receptance.weight": "model-00012-of-00015.safetensors",
272
+ "rwkv.blocks.21.feed_forward.time_mix_key": "model-00011-of-00015.safetensors",
273
+ "rwkv.blocks.21.feed_forward.time_mix_receptance": "model-00011-of-00015.safetensors",
274
+ "rwkv.blocks.21.feed_forward.value.weight": "model-00012-of-00015.safetensors",
275
+ "rwkv.blocks.21.ln1.bias": "model-00011-of-00015.safetensors",
276
+ "rwkv.blocks.21.ln1.weight": "model-00011-of-00015.safetensors",
277
+ "rwkv.blocks.21.ln2.bias": "model-00011-of-00015.safetensors",
278
+ "rwkv.blocks.21.ln2.weight": "model-00011-of-00015.safetensors",
279
+ "rwkv.blocks.22.attention.key.weight": "model-00012-of-00015.safetensors",
280
+ "rwkv.blocks.22.attention.output.weight": "model-00012-of-00015.safetensors",
281
+ "rwkv.blocks.22.attention.receptance.weight": "model-00012-of-00015.safetensors",
282
+ "rwkv.blocks.22.attention.time_decay": "model-00012-of-00015.safetensors",
283
+ "rwkv.blocks.22.attention.time_first": "model-00012-of-00015.safetensors",
284
+ "rwkv.blocks.22.attention.time_mix_key": "model-00012-of-00015.safetensors",
285
+ "rwkv.blocks.22.attention.time_mix_receptance": "model-00012-of-00015.safetensors",
286
+ "rwkv.blocks.22.attention.time_mix_value": "model-00012-of-00015.safetensors",
287
+ "rwkv.blocks.22.attention.value.weight": "model-00012-of-00015.safetensors",
288
+ "rwkv.blocks.22.feed_forward.key.weight": "model-00012-of-00015.safetensors",
289
+ "rwkv.blocks.22.feed_forward.receptance.weight": "model-00012-of-00015.safetensors",
290
+ "rwkv.blocks.22.feed_forward.time_mix_key": "model-00012-of-00015.safetensors",
291
+ "rwkv.blocks.22.feed_forward.time_mix_receptance": "model-00012-of-00015.safetensors",
292
+ "rwkv.blocks.22.feed_forward.value.weight": "model-00012-of-00015.safetensors",
293
+ "rwkv.blocks.22.ln1.bias": "model-00012-of-00015.safetensors",
294
+ "rwkv.blocks.22.ln1.weight": "model-00012-of-00015.safetensors",
295
+ "rwkv.blocks.22.ln2.bias": "model-00012-of-00015.safetensors",
296
+ "rwkv.blocks.22.ln2.weight": "model-00012-of-00015.safetensors",
297
+ "rwkv.blocks.23.attention.key.weight": "model-00012-of-00015.safetensors",
298
+ "rwkv.blocks.23.attention.output.weight": "model-00012-of-00015.safetensors",
299
+ "rwkv.blocks.23.attention.receptance.weight": "model-00012-of-00015.safetensors",
300
+ "rwkv.blocks.23.attention.time_decay": "model-00012-of-00015.safetensors",
301
+ "rwkv.blocks.23.attention.time_first": "model-00012-of-00015.safetensors",
302
+ "rwkv.blocks.23.attention.time_mix_key": "model-00012-of-00015.safetensors",
303
+ "rwkv.blocks.23.attention.time_mix_receptance": "model-00012-of-00015.safetensors",
304
+ "rwkv.blocks.23.attention.time_mix_value": "model-00012-of-00015.safetensors",
305
+ "rwkv.blocks.23.attention.value.weight": "model-00012-of-00015.safetensors",
306
+ "rwkv.blocks.23.feed_forward.key.weight": "model-00013-of-00015.safetensors",
307
+ "rwkv.blocks.23.feed_forward.receptance.weight": "model-00013-of-00015.safetensors",
308
+ "rwkv.blocks.23.feed_forward.time_mix_key": "model-00012-of-00015.safetensors",
309
+ "rwkv.blocks.23.feed_forward.time_mix_receptance": "model-00012-of-00015.safetensors",
310
+ "rwkv.blocks.23.feed_forward.value.weight": "model-00013-of-00015.safetensors",
311
+ "rwkv.blocks.23.ln1.bias": "model-00012-of-00015.safetensors",
312
+ "rwkv.blocks.23.ln1.weight": "model-00012-of-00015.safetensors",
313
+ "rwkv.blocks.23.ln2.bias": "model-00012-of-00015.safetensors",
314
+ "rwkv.blocks.23.ln2.weight": "model-00012-of-00015.safetensors",
315
+ "rwkv.blocks.24.attention.key.weight": "model-00013-of-00015.safetensors",
316
+ "rwkv.blocks.24.attention.output.weight": "model-00013-of-00015.safetensors",
317
+ "rwkv.blocks.24.attention.receptance.weight": "model-00013-of-00015.safetensors",
318
+ "rwkv.blocks.24.attention.time_decay": "model-00013-of-00015.safetensors",
319
+ "rwkv.blocks.24.attention.time_first": "model-00013-of-00015.safetensors",
320
+ "rwkv.blocks.24.attention.time_mix_key": "model-00013-of-00015.safetensors",
321
+ "rwkv.blocks.24.attention.time_mix_receptance": "model-00013-of-00015.safetensors",
322
+ "rwkv.blocks.24.attention.time_mix_value": "model-00013-of-00015.safetensors",
323
+ "rwkv.blocks.24.attention.value.weight": "model-00013-of-00015.safetensors",
324
+ "rwkv.blocks.24.feed_forward.key.weight": "model-00013-of-00015.safetensors",
325
+ "rwkv.blocks.24.feed_forward.receptance.weight": "model-00013-of-00015.safetensors",
326
+ "rwkv.blocks.24.feed_forward.time_mix_key": "model-00013-of-00015.safetensors",
327
+ "rwkv.blocks.24.feed_forward.time_mix_receptance": "model-00013-of-00015.safetensors",
328
+ "rwkv.blocks.24.feed_forward.value.weight": "model-00013-of-00015.safetensors",
329
+ "rwkv.blocks.24.ln1.bias": "model-00013-of-00015.safetensors",
330
+ "rwkv.blocks.24.ln1.weight": "model-00013-of-00015.safetensors",
331
+ "rwkv.blocks.24.ln2.bias": "model-00013-of-00015.safetensors",
332
+ "rwkv.blocks.24.ln2.weight": "model-00013-of-00015.safetensors",
333
+ "rwkv.blocks.25.attention.key.weight": "model-00013-of-00015.safetensors",
334
+ "rwkv.blocks.25.attention.output.weight": "model-00013-of-00015.safetensors",
335
+ "rwkv.blocks.25.attention.receptance.weight": "model-00013-of-00015.safetensors",
336
+ "rwkv.blocks.25.attention.time_decay": "model-00013-of-00015.safetensors",
337
+ "rwkv.blocks.25.attention.time_first": "model-00013-of-00015.safetensors",
338
+ "rwkv.blocks.25.attention.time_mix_key": "model-00013-of-00015.safetensors",
339
+ "rwkv.blocks.25.attention.time_mix_receptance": "model-00013-of-00015.safetensors",
340
+ "rwkv.blocks.25.attention.time_mix_value": "model-00013-of-00015.safetensors",
341
+ "rwkv.blocks.25.attention.value.weight": "model-00013-of-00015.safetensors",
342
+ "rwkv.blocks.25.feed_forward.key.weight": "model-00014-of-00015.safetensors",
343
+ "rwkv.blocks.25.feed_forward.receptance.weight": "model-00014-of-00015.safetensors",
344
+ "rwkv.blocks.25.feed_forward.time_mix_key": "model-00013-of-00015.safetensors",
345
+ "rwkv.blocks.25.feed_forward.time_mix_receptance": "model-00013-of-00015.safetensors",
346
+ "rwkv.blocks.25.feed_forward.value.weight": "model-00014-of-00015.safetensors",
347
+ "rwkv.blocks.25.ln1.bias": "model-00013-of-00015.safetensors",
348
+ "rwkv.blocks.25.ln1.weight": "model-00013-of-00015.safetensors",
349
+ "rwkv.blocks.25.ln2.bias": "model-00013-of-00015.safetensors",
350
+ "rwkv.blocks.25.ln2.weight": "model-00013-of-00015.safetensors",
351
+ "rwkv.blocks.26.attention.key.weight": "model-00014-of-00015.safetensors",
352
+ "rwkv.blocks.26.attention.output.weight": "model-00014-of-00015.safetensors",
353
+ "rwkv.blocks.26.attention.receptance.weight": "model-00014-of-00015.safetensors",
354
+ "rwkv.blocks.26.attention.time_decay": "model-00014-of-00015.safetensors",
355
+ "rwkv.blocks.26.attention.time_first": "model-00014-of-00015.safetensors",
356
+ "rwkv.blocks.26.attention.time_mix_key": "model-00014-of-00015.safetensors",
357
+ "rwkv.blocks.26.attention.time_mix_receptance": "model-00014-of-00015.safetensors",
358
+ "rwkv.blocks.26.attention.time_mix_value": "model-00014-of-00015.safetensors",
359
+ "rwkv.blocks.26.attention.value.weight": "model-00014-of-00015.safetensors",
360
+ "rwkv.blocks.26.feed_forward.key.weight": "model-00014-of-00015.safetensors",
361
+ "rwkv.blocks.26.feed_forward.receptance.weight": "model-00014-of-00015.safetensors",
362
+ "rwkv.blocks.26.feed_forward.time_mix_key": "model-00014-of-00015.safetensors",
363
+ "rwkv.blocks.26.feed_forward.time_mix_receptance": "model-00014-of-00015.safetensors",
364
+ "rwkv.blocks.26.feed_forward.value.weight": "model-00014-of-00015.safetensors",
365
+ "rwkv.blocks.26.ln1.bias": "model-00014-of-00015.safetensors",
366
+ "rwkv.blocks.26.ln1.weight": "model-00014-of-00015.safetensors",
367
+ "rwkv.blocks.26.ln2.bias": "model-00014-of-00015.safetensors",
368
+ "rwkv.blocks.26.ln2.weight": "model-00014-of-00015.safetensors",
369
+ "rwkv.blocks.27.attention.key.weight": "model-00014-of-00015.safetensors",
370
+ "rwkv.blocks.27.attention.output.weight": "model-00014-of-00015.safetensors",
371
+ "rwkv.blocks.27.attention.receptance.weight": "model-00014-of-00015.safetensors",
372
+ "rwkv.blocks.27.attention.time_decay": "model-00014-of-00015.safetensors",
373
+ "rwkv.blocks.27.attention.time_first": "model-00014-of-00015.safetensors",
374
+ "rwkv.blocks.27.attention.time_mix_key": "model-00014-of-00015.safetensors",
375
+ "rwkv.blocks.27.attention.time_mix_receptance": "model-00014-of-00015.safetensors",
376
+ "rwkv.blocks.27.attention.time_mix_value": "model-00014-of-00015.safetensors",
377
+ "rwkv.blocks.27.attention.value.weight": "model-00014-of-00015.safetensors",
378
+ "rwkv.blocks.27.feed_forward.key.weight": "model-00015-of-00015.safetensors",
379
+ "rwkv.blocks.27.feed_forward.receptance.weight": "model-00015-of-00015.safetensors",
380
+ "rwkv.blocks.27.feed_forward.time_mix_key": "model-00014-of-00015.safetensors",
381
+ "rwkv.blocks.27.feed_forward.time_mix_receptance": "model-00014-of-00015.safetensors",
382
+ "rwkv.blocks.27.feed_forward.value.weight": "model-00015-of-00015.safetensors",
383
+ "rwkv.blocks.27.ln1.bias": "model-00014-of-00015.safetensors",
384
+ "rwkv.blocks.27.ln1.weight": "model-00014-of-00015.safetensors",
385
+ "rwkv.blocks.27.ln2.bias": "model-00014-of-00015.safetensors",
386
+ "rwkv.blocks.27.ln2.weight": "model-00014-of-00015.safetensors",
387
+ "rwkv.blocks.3.attention.key.weight": "model-00002-of-00015.safetensors",
388
+ "rwkv.blocks.3.attention.output.weight": "model-00002-of-00015.safetensors",
389
+ "rwkv.blocks.3.attention.receptance.weight": "model-00002-of-00015.safetensors",
390
+ "rwkv.blocks.3.attention.time_decay": "model-00002-of-00015.safetensors",
391
+ "rwkv.blocks.3.attention.time_first": "model-00002-of-00015.safetensors",
392
+ "rwkv.blocks.3.attention.time_mix_key": "model-00002-of-00015.safetensors",
393
+ "rwkv.blocks.3.attention.time_mix_receptance": "model-00002-of-00015.safetensors",
394
+ "rwkv.blocks.3.attention.time_mix_value": "model-00002-of-00015.safetensors",
395
+ "rwkv.blocks.3.attention.value.weight": "model-00002-of-00015.safetensors",
396
+ "rwkv.blocks.3.feed_forward.key.weight": "model-00003-of-00015.safetensors",
397
+ "rwkv.blocks.3.feed_forward.receptance.weight": "model-00003-of-00015.safetensors",
398
+ "rwkv.blocks.3.feed_forward.time_mix_key": "model-00002-of-00015.safetensors",
399
+ "rwkv.blocks.3.feed_forward.time_mix_receptance": "model-00002-of-00015.safetensors",
400
+ "rwkv.blocks.3.feed_forward.value.weight": "model-00003-of-00015.safetensors",
401
+ "rwkv.blocks.3.ln1.bias": "model-00002-of-00015.safetensors",
402
+ "rwkv.blocks.3.ln1.weight": "model-00002-of-00015.safetensors",
403
+ "rwkv.blocks.3.ln2.bias": "model-00002-of-00015.safetensors",
404
+ "rwkv.blocks.3.ln2.weight": "model-00002-of-00015.safetensors",
405
+ "rwkv.blocks.4.attention.key.weight": "model-00003-of-00015.safetensors",
406
+ "rwkv.blocks.4.attention.output.weight": "model-00003-of-00015.safetensors",
407
+ "rwkv.blocks.4.attention.receptance.weight": "model-00003-of-00015.safetensors",
408
+ "rwkv.blocks.4.attention.time_decay": "model-00003-of-00015.safetensors",
409
+ "rwkv.blocks.4.attention.time_first": "model-00003-of-00015.safetensors",
410
+ "rwkv.blocks.4.attention.time_mix_key": "model-00003-of-00015.safetensors",
411
+ "rwkv.blocks.4.attention.time_mix_receptance": "model-00003-of-00015.safetensors",
412
+ "rwkv.blocks.4.attention.time_mix_value": "model-00003-of-00015.safetensors",
413
+ "rwkv.blocks.4.attention.value.weight": "model-00003-of-00015.safetensors",
414
+ "rwkv.blocks.4.feed_forward.key.weight": "model-00003-of-00015.safetensors",
415
+ "rwkv.blocks.4.feed_forward.receptance.weight": "model-00003-of-00015.safetensors",
416
+ "rwkv.blocks.4.feed_forward.time_mix_key": "model-00003-of-00015.safetensors",
417
+ "rwkv.blocks.4.feed_forward.time_mix_receptance": "model-00003-of-00015.safetensors",
418
+ "rwkv.blocks.4.feed_forward.value.weight": "model-00003-of-00015.safetensors",
419
+ "rwkv.blocks.4.ln1.bias": "model-00003-of-00015.safetensors",
420
+ "rwkv.blocks.4.ln1.weight": "model-00003-of-00015.safetensors",
421
+ "rwkv.blocks.4.ln2.bias": "model-00003-of-00015.safetensors",
422
+ "rwkv.blocks.4.ln2.weight": "model-00003-of-00015.safetensors",
423
+ "rwkv.blocks.5.attention.key.weight": "model-00003-of-00015.safetensors",
424
+ "rwkv.blocks.5.attention.output.weight": "model-00003-of-00015.safetensors",
425
+ "rwkv.blocks.5.attention.receptance.weight": "model-00003-of-00015.safetensors",
426
+ "rwkv.blocks.5.attention.time_decay": "model-00003-of-00015.safetensors",
427
+ "rwkv.blocks.5.attention.time_first": "model-00003-of-00015.safetensors",
428
+ "rwkv.blocks.5.attention.time_mix_key": "model-00003-of-00015.safetensors",
429
+ "rwkv.blocks.5.attention.time_mix_receptance": "model-00003-of-00015.safetensors",
430
+ "rwkv.blocks.5.attention.time_mix_value": "model-00003-of-00015.safetensors",
431
+ "rwkv.blocks.5.attention.value.weight": "model-00003-of-00015.safetensors",
432
+ "rwkv.blocks.5.feed_forward.key.weight": "model-00004-of-00015.safetensors",
433
+ "rwkv.blocks.5.feed_forward.receptance.weight": "model-00004-of-00015.safetensors",
434
+ "rwkv.blocks.5.feed_forward.time_mix_key": "model-00003-of-00015.safetensors",
435
+ "rwkv.blocks.5.feed_forward.time_mix_receptance": "model-00003-of-00015.safetensors",
436
+ "rwkv.blocks.5.feed_forward.value.weight": "model-00004-of-00015.safetensors",
437
+ "rwkv.blocks.5.ln1.bias": "model-00003-of-00015.safetensors",
438
+ "rwkv.blocks.5.ln1.weight": "model-00003-of-00015.safetensors",
439
+ "rwkv.blocks.5.ln2.bias": "model-00003-of-00015.safetensors",
440
+ "rwkv.blocks.5.ln2.weight": "model-00003-of-00015.safetensors",
441
+ "rwkv.blocks.6.attention.key.weight": "model-00004-of-00015.safetensors",
442
+ "rwkv.blocks.6.attention.output.weight": "model-00004-of-00015.safetensors",
443
+ "rwkv.blocks.6.attention.receptance.weight": "model-00004-of-00015.safetensors",
444
+ "rwkv.blocks.6.attention.time_decay": "model-00004-of-00015.safetensors",
445
+ "rwkv.blocks.6.attention.time_first": "model-00004-of-00015.safetensors",
446
+ "rwkv.blocks.6.attention.time_mix_key": "model-00004-of-00015.safetensors",
447
+ "rwkv.blocks.6.attention.time_mix_receptance": "model-00004-of-00015.safetensors",
448
+ "rwkv.blocks.6.attention.time_mix_value": "model-00004-of-00015.safetensors",
449
+ "rwkv.blocks.6.attention.value.weight": "model-00004-of-00015.safetensors",
450
+ "rwkv.blocks.6.feed_forward.key.weight": "model-00004-of-00015.safetensors",
451
+ "rwkv.blocks.6.feed_forward.receptance.weight": "model-00004-of-00015.safetensors",
452
+ "rwkv.blocks.6.feed_forward.time_mix_key": "model-00004-of-00015.safetensors",
453
+ "rwkv.blocks.6.feed_forward.time_mix_receptance": "model-00004-of-00015.safetensors",
454
+ "rwkv.blocks.6.feed_forward.value.weight": "model-00004-of-00015.safetensors",
455
+ "rwkv.blocks.6.ln1.bias": "model-00004-of-00015.safetensors",
456
+ "rwkv.blocks.6.ln1.weight": "model-00004-of-00015.safetensors",
457
+ "rwkv.blocks.6.ln2.bias": "model-00004-of-00015.safetensors",
458
+ "rwkv.blocks.6.ln2.weight": "model-00004-of-00015.safetensors",
459
+ "rwkv.blocks.7.attention.key.weight": "model-00004-of-00015.safetensors",
460
+ "rwkv.blocks.7.attention.output.weight": "model-00004-of-00015.safetensors",
461
+ "rwkv.blocks.7.attention.receptance.weight": "model-00004-of-00015.safetensors",
462
+ "rwkv.blocks.7.attention.time_decay": "model-00004-of-00015.safetensors",
463
+ "rwkv.blocks.7.attention.time_first": "model-00004-of-00015.safetensors",
464
+ "rwkv.blocks.7.attention.time_mix_key": "model-00004-of-00015.safetensors",
465
+ "rwkv.blocks.7.attention.time_mix_receptance": "model-00004-of-00015.safetensors",
466
+ "rwkv.blocks.7.attention.time_mix_value": "model-00004-of-00015.safetensors",
467
+ "rwkv.blocks.7.attention.value.weight": "model-00004-of-00015.safetensors",
468
+ "rwkv.blocks.7.feed_forward.key.weight": "model-00005-of-00015.safetensors",
469
+ "rwkv.blocks.7.feed_forward.receptance.weight": "model-00005-of-00015.safetensors",
470
+ "rwkv.blocks.7.feed_forward.time_mix_key": "model-00004-of-00015.safetensors",
471
+ "rwkv.blocks.7.feed_forward.time_mix_receptance": "model-00004-of-00015.safetensors",
472
+ "rwkv.blocks.7.feed_forward.value.weight": "model-00005-of-00015.safetensors",
473
+ "rwkv.blocks.7.ln1.bias": "model-00004-of-00015.safetensors",
474
+ "rwkv.blocks.7.ln1.weight": "model-00004-of-00015.safetensors",
475
+ "rwkv.blocks.7.ln2.bias": "model-00004-of-00015.safetensors",
476
+ "rwkv.blocks.7.ln2.weight": "model-00004-of-00015.safetensors",
477
+ "rwkv.blocks.8.attention.key.weight": "model-00005-of-00015.safetensors",
478
+ "rwkv.blocks.8.attention.output.weight": "model-00005-of-00015.safetensors",
479
+ "rwkv.blocks.8.attention.receptance.weight": "model-00005-of-00015.safetensors",
480
+ "rwkv.blocks.8.attention.time_decay": "model-00005-of-00015.safetensors",
481
+ "rwkv.blocks.8.attention.time_first": "model-00005-of-00015.safetensors",
482
+ "rwkv.blocks.8.attention.time_mix_key": "model-00005-of-00015.safetensors",
483
+ "rwkv.blocks.8.attention.time_mix_receptance": "model-00005-of-00015.safetensors",
484
+ "rwkv.blocks.8.attention.time_mix_value": "model-00005-of-00015.safetensors",
485
+ "rwkv.blocks.8.attention.value.weight": "model-00005-of-00015.safetensors",
486
+ "rwkv.blocks.8.feed_forward.key.weight": "model-00005-of-00015.safetensors",
487
+ "rwkv.blocks.8.feed_forward.receptance.weight": "model-00005-of-00015.safetensors",
488
+ "rwkv.blocks.8.feed_forward.time_mix_key": "model-00005-of-00015.safetensors",
489
+ "rwkv.blocks.8.feed_forward.time_mix_receptance": "model-00005-of-00015.safetensors",
490
+ "rwkv.blocks.8.feed_forward.value.weight": "model-00005-of-00015.safetensors",
491
+ "rwkv.blocks.8.ln1.bias": "model-00005-of-00015.safetensors",
492
+ "rwkv.blocks.8.ln1.weight": "model-00005-of-00015.safetensors",
493
+ "rwkv.blocks.8.ln2.bias": "model-00005-of-00015.safetensors",
494
+ "rwkv.blocks.8.ln2.weight": "model-00005-of-00015.safetensors",
495
+ "rwkv.blocks.9.attention.key.weight": "model-00005-of-00015.safetensors",
496
+ "rwkv.blocks.9.attention.output.weight": "model-00005-of-00015.safetensors",
497
+ "rwkv.blocks.9.attention.receptance.weight": "model-00005-of-00015.safetensors",
498
+ "rwkv.blocks.9.attention.time_decay": "model-00005-of-00015.safetensors",
499
+ "rwkv.blocks.9.attention.time_first": "model-00005-of-00015.safetensors",
500
+ "rwkv.blocks.9.attention.time_mix_key": "model-00005-of-00015.safetensors",
501
+ "rwkv.blocks.9.attention.time_mix_receptance": "model-00005-of-00015.safetensors",
502
+ "rwkv.blocks.9.attention.time_mix_value": "model-00005-of-00015.safetensors",
503
+ "rwkv.blocks.9.attention.value.weight": "model-00005-of-00015.safetensors",
504
+ "rwkv.blocks.9.feed_forward.key.weight": "model-00006-of-00015.safetensors",
505
+ "rwkv.blocks.9.feed_forward.receptance.weight": "model-00006-of-00015.safetensors",
506
+ "rwkv.blocks.9.feed_forward.time_mix_key": "model-00005-of-00015.safetensors",
507
+ "rwkv.blocks.9.feed_forward.time_mix_receptance": "model-00005-of-00015.safetensors",
508
+ "rwkv.blocks.9.feed_forward.value.weight": "model-00006-of-00015.safetensors",
509
+ "rwkv.blocks.9.ln1.bias": "model-00005-of-00015.safetensors",
510
+ "rwkv.blocks.9.ln1.weight": "model-00005-of-00015.safetensors",
511
+ "rwkv.blocks.9.ln2.bias": "model-00005-of-00015.safetensors",
512
+ "rwkv.blocks.9.ln2.weight": "model-00005-of-00015.safetensors",
513
+ "rwkv.embeddings.weight": "model-00001-of-00015.safetensors",
514
+ "rwkv.ln_out.bias": "model-00015-of-00015.safetensors",
515
+ "rwkv.ln_out.weight": "model-00015-of-00015.safetensors"
516
  }
517
  }
pytorch_model-00001-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1dde9f1b7de8e05d30655366e0cd56e273ed15a6021558c6f180d473d4f843e
3
+ size 996617751
pytorch_model-00002-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ee0832da23d20cb04f20b339c55e6c8dd692f66e53315743e39dee06cc4c66e8
3
+ size 872607319
pytorch_model-00003-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d4e223ebf87c21e58e21f6cdab00592771e1c8d73c3a16b6d2e36c79724f2c9
3
+ size 872607319
pytorch_model-00004-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1e5d5b83ec9a662a2e1b53ac8aaf2df5617150740eacd8abd64dbcaef91c1e0
3
+ size 872607319
pytorch_model-00005-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33ac092e51936c5ee3dd836368b5bf13024948ba75684ce3355d00032477aa53
3
+ size 872607319
pytorch_model-00006-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f19b4c1078b2692aca4dda5c2e7d90b890f456a5234ed9dc015408e1f79fede3
3
+ size 872607383
pytorch_model-00007-of-00015.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70bfe750348156eb394979c82c8e4ff5d09756271d34e10edac48ae5430f3f15
3
+ size 872607383