Hugging Face's logo

Equall
/
SaulLM-54B-Instruct

Text Generation
Transformers
Safetensors
English
mixtral
text-generation-inference
Inference Endpoints
Model card Files Community
3
PierreColombo commited on
Commit
9a6b0da
•
1 Parent(s): f5185b9

Upload 27 files

Browse files
Files changed (27) hide show
  1. config.json +30 -0
  2. generation_config.json +7 -0
  3. latest +1 -0
  4. model-00001-of-00019.safetensors +3 -0
  5. model-00002-of-00019.safetensors +3 -0
  6. model-00003-of-00019.safetensors +3 -0
  7. model-00004-of-00019.safetensors +3 -0
  8. model-00005-of-00019.safetensors +3 -0
  9. model-00006-of-00019.safetensors +3 -0
  10. model-00007-of-00019.safetensors +3 -0
  11. model-00008-of-00019.safetensors +3 -0
  12. model-00009-of-00019.safetensors +3 -0
  13. model-00010-of-00019.safetensors +3 -0
  14. model-00011-of-00019.safetensors +3 -0
  15. model-00012-of-00019.safetensors +3 -0
  16. model-00013-of-00019.safetensors +3 -0
  17. model-00014-of-00019.safetensors +3 -0
  18. model-00015-of-00019.safetensors +3 -0
  19. model-00016-of-00019.safetensors +3 -0
  20. model-00017-of-00019.safetensors +3 -0
  21. model-00018-of-00019.safetensors +3 -0
  22. model-00019-of-00019.safetensors +3 -0
  23. model.safetensors.index.json +1002 -0
  24. special_tokens_map.json +30 -0
  25. tokenizer.model +3 -0
  26. tokenizer_config.json +44 -0
  27. zero_to_fp32.py +604 -0
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "checkpoints-368-ift-single-hermes",
3
+ "architectures": [
4
+ "MixtralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 4096,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 14336,
13
+ "max_position_embeddings": 32768,
14
+ "model_type": "mixtral",
15
+ "num_attention_heads": 32,
16
+ "num_experts_per_tok": 2,
17
+ "num_hidden_layers": 32,
18
+ "num_key_value_heads": 8,
19
+ "num_local_experts": 8,
20
+ "output_router_logits": false,
21
+ "rms_norm_eps": 1e-05,
22
+ "rope_theta": 1000000.0,
23
+ "router_aux_loss_coef": 0.02,
24
+ "sliding_window": null,
25
+ "tie_word_embeddings": false,
26
+ "torch_dtype": "bfloat16",
27
+ "transformers_version": "4.39.2",
28
+ "use_cache": false,
29
+ "vocab_size": 32000
30
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 2,
6
+ "transformers_version": "4.39.2"
7
+ }
latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step270
model-00001-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d42fb4b84b3adff83a5b1df9a19a9130040fe8a535970e0bb530685d8c1381c
3
+ size 4892809584
model-00002-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4adc47d39a594932e38dd3cb7ef24d2b55b453153145868bff4a900c35e19367
3
+ size 4983004016
model-00003-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6948de604e259653d8b95229d29800150c268b74104337dfa2a6301d7ab4051e
3
+ size 4983004016
model-00004-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9fcee0940bd154974dac0ed4e72c47e61c74b4cf5cda2287daff3e67d52328e
3
+ size 4899035200
model-00005-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:819bffd0053c5c0a3e76f84f216177c26f6dea825335face8378704929714100
3
+ size 4983004016
model-00006-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:346a6e6e21121695823a360237945a76a3818fc935af4887d818673b4440a8a4
3
+ size 4983004016
model-00007-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6466cb0d30202a7c69fa3049b7768f4c7bc3d3a507b6bdec87b52d373602b639
3
+ size 4899035248
model-00008-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:afa485b2f2bd8e3dc343575b92d29fd0941b11768f91207a30a3fe21471179bc
3
+ size 4983004072
model-00009-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55e0c6dff3b4fb38b5589b5a140ca5830e5fcd03e436e50a499681b613972d06
3
+ size 4983004072
model-00010-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3e528dc9fbc73bf4d5bca430ced4c721a754dad39fe534157262e2704cc7848
3
+ size 4899035248
model-00011-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e0a73ccc9533d808409eb626b6176dfa20415aac35b55d62df89d9a64d20e8d1
3
+ size 4983004072
model-00012-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2813734aa2ff934e4ab0664f1a61b90988184a26fdb43976f303e8d8583af0f4
3
+ size 4983004072
model-00013-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bc57346c9c63b0b45d873ff2618a9a49e6fc99c89a436a10f82a043072614b8
3
+ size 4983004072
model-00014-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c147148bddc7bfa38ac0e3f3ba5479646d849d790083ea9025728b367f102a6b
3
+ size 4899035248
model-00015-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4282895fc222f977c26f734d60cb4c40958a4e994b0d55e3050e6a5fa8615812
3
+ size 4983004072
model-00016-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a03e0c8368a7c84534efa697a9f7b3ddb4f4982fe69303eae53e0a7d08df98c
3
+ size 4983004072
model-00017-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c4ad29e0f8deb55c6fce8ada8ae940ef6a5aa352c28c2fd4ab9490c36915573
3
+ size 4899035248
model-00018-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6886a1f484cefefbc024240e0b50e09b9d6f92d969e60cd004cd00cb9525893
3
+ size 4983004072
model-00019-of-00019.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29f3b01f5393a3762ab86b084876e74ba8f6a95927b950da0e79f1a002be74db
3
+ size 4221679088
model.safetensors.index.json ADDED
@@ -0,0 +1,1002 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 93405585408
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00019-of-00019.safetensors",
7
+ "model.embed_tokens.weight": "model-00001-of-00019.safetensors",
8
+ "model.layers.0.block_sparse_moe.experts.0.w1.weight": "model-00001-of-00019.safetensors",
9
+ "model.layers.0.block_sparse_moe.experts.0.w2.weight": "model-00001-of-00019.safetensors",
10
+ "model.layers.0.block_sparse_moe.experts.0.w3.weight": "model-00001-of-00019.safetensors",
11
+ "model.layers.0.block_sparse_moe.experts.1.w1.weight": "model-00001-of-00019.safetensors",
12
+ "model.layers.0.block_sparse_moe.experts.1.w2.weight": "model-00001-of-00019.safetensors",
13
+ "model.layers.0.block_sparse_moe.experts.1.w3.weight": "model-00001-of-00019.safetensors",
14
+ "model.layers.0.block_sparse_moe.experts.2.w1.weight": "model-00001-of-00019.safetensors",
15
+ "model.layers.0.block_sparse_moe.experts.2.w2.weight": "model-00001-of-00019.safetensors",
16
+ "model.layers.0.block_sparse_moe.experts.2.w3.weight": "model-00001-of-00019.safetensors",
17
+ "model.layers.0.block_sparse_moe.experts.3.w1.weight": "model-00001-of-00019.safetensors",
18
+ "model.layers.0.block_sparse_moe.experts.3.w2.weight": "model-00001-of-00019.safetensors",
19
+ "model.layers.0.block_sparse_moe.experts.3.w3.weight": "model-00001-of-00019.safetensors",
20
+ "model.layers.0.block_sparse_moe.experts.4.w1.weight": "model-00001-of-00019.safetensors",
21
+ "model.layers.0.block_sparse_moe.experts.4.w2.weight": "model-00001-of-00019.safetensors",
22
+ "model.layers.0.block_sparse_moe.experts.4.w3.weight": "model-00001-of-00019.safetensors",
23
+ "model.layers.0.block_sparse_moe.experts.5.w1.weight": "model-00001-of-00019.safetensors",
24
+ "model.layers.0.block_sparse_moe.experts.5.w2.weight": "model-00001-of-00019.safetensors",
25
+ "model.layers.0.block_sparse_moe.experts.5.w3.weight": "model-00001-of-00019.safetensors",
26
+ "model.layers.0.block_sparse_moe.experts.6.w1.weight": "model-00001-of-00019.safetensors",
27
+ "model.layers.0.block_sparse_moe.experts.6.w2.weight": "model-00001-of-00019.safetensors",
28
+ "model.layers.0.block_sparse_moe.experts.6.w3.weight": "model-00001-of-00019.safetensors",
29
+ "model.layers.0.block_sparse_moe.experts.7.w1.weight": "model-00001-of-00019.safetensors",
30
+ "model.layers.0.block_sparse_moe.experts.7.w2.weight": "model-00001-of-00019.safetensors",
31
+ "model.layers.0.block_sparse_moe.experts.7.w3.weight": "model-00001-of-00019.safetensors",
32
+ "model.layers.0.block_sparse_moe.gate.weight": "model-00001-of-00019.safetensors",
33
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00019.safetensors",
34
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00019.safetensors",
35
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00019.safetensors",
36
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00019.safetensors",
37
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00019.safetensors",
38
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00019.safetensors",
39
+ "model.layers.1.block_sparse_moe.experts.0.w1.weight": "model-00001-of-00019.safetensors",
40
+ "model.layers.1.block_sparse_moe.experts.0.w2.weight": "model-00001-of-00019.safetensors",
41
+ "model.layers.1.block_sparse_moe.experts.0.w3.weight": "model-00001-of-00019.safetensors",
42
+ "model.layers.1.block_sparse_moe.experts.1.w1.weight": "model-00001-of-00019.safetensors",
43
+ "model.layers.1.block_sparse_moe.experts.1.w2.weight": "model-00001-of-00019.safetensors",
44
+ "model.layers.1.block_sparse_moe.experts.1.w3.weight": "model-00001-of-00019.safetensors",
45
+ "model.layers.1.block_sparse_moe.experts.2.w1.weight": "model-00001-of-00019.safetensors",
46
+ "model.layers.1.block_sparse_moe.experts.2.w2.weight": "model-00001-of-00019.safetensors",
47
+ "model.layers.1.block_sparse_moe.experts.2.w3.weight": "model-00001-of-00019.safetensors",
48
+ "model.layers.1.block_sparse_moe.experts.3.w1.weight": "model-00001-of-00019.safetensors",
49
+ "model.layers.1.block_sparse_moe.experts.3.w2.weight": "model-00001-of-00019.safetensors",
50
+ "model.layers.1.block_sparse_moe.experts.3.w3.weight": "model-00001-of-00019.safetensors",
51
+ "model.layers.1.block_sparse_moe.experts.4.w1.weight": "model-00001-of-00019.safetensors",
52
+ "model.layers.1.block_sparse_moe.experts.4.w2.weight": "model-00001-of-00019.safetensors",
53
+ "model.layers.1.block_sparse_moe.experts.4.w3.weight": "model-00002-of-00019.safetensors",
54
+ "model.layers.1.block_sparse_moe.experts.5.w1.weight": "model-00002-of-00019.safetensors",
55
+ "model.layers.1.block_sparse_moe.experts.5.w2.weight": "model-00002-of-00019.safetensors",
56
+ "model.layers.1.block_sparse_moe.experts.5.w3.weight": "model-00002-of-00019.safetensors",
57
+ "model.layers.1.block_sparse_moe.experts.6.w1.weight": "model-00002-of-00019.safetensors",
58
+ "model.layers.1.block_sparse_moe.experts.6.w2.weight": "model-00002-of-00019.safetensors",
59
+ "model.layers.1.block_sparse_moe.experts.6.w3.weight": "model-00002-of-00019.safetensors",
60
+ "model.layers.1.block_sparse_moe.experts.7.w1.weight": "model-00002-of-00019.safetensors",
61
+ "model.layers.1.block_sparse_moe.experts.7.w2.weight": "model-00002-of-00019.safetensors",
62
+ "model.layers.1.block_sparse_moe.experts.7.w3.weight": "model-00002-of-00019.safetensors",
63
+ "model.layers.1.block_sparse_moe.gate.weight": "model-00001-of-00019.safetensors",
64
+ "model.layers.1.input_layernorm.weight": "model-00002-of-00019.safetensors",
65
+ "model.layers.1.post_attention_layernorm.weight": "model-00002-of-00019.safetensors",
66
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00019.safetensors",
67
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00019.safetensors",
68
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00019.safetensors",
69
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00019.safetensors",
70
+ "model.layers.10.block_sparse_moe.experts.0.w1.weight": "model-00006-of-00019.safetensors",
71
+ "model.layers.10.block_sparse_moe.experts.0.w2.weight": "model-00006-of-00019.safetensors",
72
+ "model.layers.10.block_sparse_moe.experts.0.w3.weight": "model-00006-of-00019.safetensors",
73
+ "model.layers.10.block_sparse_moe.experts.1.w1.weight": "model-00007-of-00019.safetensors",
74
+ "model.layers.10.block_sparse_moe.experts.1.w2.weight": "model-00007-of-00019.safetensors",
75
+ "model.layers.10.block_sparse_moe.experts.1.w3.weight": "model-00007-of-00019.safetensors",
76
+ "model.layers.10.block_sparse_moe.experts.2.w1.weight": "model-00007-of-00019.safetensors",
77
+ "model.layers.10.block_sparse_moe.experts.2.w2.weight": "model-00007-of-00019.safetensors",
78
+ "model.layers.10.block_sparse_moe.experts.2.w3.weight": "model-00007-of-00019.safetensors",
79
+ "model.layers.10.block_sparse_moe.experts.3.w1.weight": "model-00007-of-00019.safetensors",
80
+ "model.layers.10.block_sparse_moe.experts.3.w2.weight": "model-00007-of-00019.safetensors",
81
+ "model.layers.10.block_sparse_moe.experts.3.w3.weight": "model-00007-of-00019.safetensors",
82
+ "model.layers.10.block_sparse_moe.experts.4.w1.weight": "model-00007-of-00019.safetensors",
83
+ "model.layers.10.block_sparse_moe.experts.4.w2.weight": "model-00007-of-00019.safetensors",
84
+ "model.layers.10.block_sparse_moe.experts.4.w3.weight": "model-00007-of-00019.safetensors",
85
+ "model.layers.10.block_sparse_moe.experts.5.w1.weight": "model-00007-of-00019.safetensors",
86
+ "model.layers.10.block_sparse_moe.experts.5.w2.weight": "model-00007-of-00019.safetensors",
87
+ "model.layers.10.block_sparse_moe.experts.5.w3.weight": "model-00007-of-00019.safetensors",
88
+ "model.layers.10.block_sparse_moe.experts.6.w1.weight": "model-00007-of-00019.safetensors",
89
+ "model.layers.10.block_sparse_moe.experts.6.w2.weight": "model-00007-of-00019.safetensors",
90
+ "model.layers.10.block_sparse_moe.experts.6.w3.weight": "model-00007-of-00019.safetensors",
91
+ "model.layers.10.block_sparse_moe.experts.7.w1.weight": "model-00007-of-00019.safetensors",
92
+ "model.layers.10.block_sparse_moe.experts.7.w2.weight": "model-00007-of-00019.safetensors",
93
+ "model.layers.10.block_sparse_moe.experts.7.w3.weight": "model-00007-of-00019.safetensors",
94
+ "model.layers.10.block_sparse_moe.gate.weight": "model-00006-of-00019.safetensors",
95
+ "model.layers.10.input_layernorm.weight": "model-00007-of-00019.safetensors",
96
+ "model.layers.10.post_attention_layernorm.weight": "model-00007-of-00019.safetensors",
97
+ "model.layers.10.self_attn.k_proj.weight": "model-00006-of-00019.safetensors",
98
+ "model.layers.10.self_attn.o_proj.weight": "model-00006-of-00019.safetensors",
99
+ "model.layers.10.self_attn.q_proj.weight": "model-00006-of-00019.safetensors",
100
+ "model.layers.10.self_attn.v_proj.weight": "model-00006-of-00019.safetensors",
101
+ "model.layers.11.block_sparse_moe.experts.0.w1.weight": "model-00007-of-00019.safetensors",
102
+ "model.layers.11.block_sparse_moe.experts.0.w2.weight": "model-00007-of-00019.safetensors",
103
+ "model.layers.11.block_sparse_moe.experts.0.w3.weight": "model-00007-of-00019.safetensors",
104
+ "model.layers.11.block_sparse_moe.experts.1.w1.weight": "model-00007-of-00019.safetensors",
105
+ "model.layers.11.block_sparse_moe.experts.1.w2.weight": "model-00007-of-00019.safetensors",
106
+ "model.layers.11.block_sparse_moe.experts.1.w3.weight": "model-00007-of-00019.safetensors",
107
+ "model.layers.11.block_sparse_moe.experts.2.w1.weight": "model-00007-of-00019.safetensors",
108
+ "model.layers.11.block_sparse_moe.experts.2.w2.weight": "model-00007-of-00019.safetensors",
109
+ "model.layers.11.block_sparse_moe.experts.2.w3.weight": "model-00007-of-00019.safetensors",
110
+ "model.layers.11.block_sparse_moe.experts.3.w1.weight": "model-00007-of-00019.safetensors",
111
+ "model.layers.11.block_sparse_moe.experts.3.w2.weight": "model-00007-of-00019.safetensors",
112
+ "model.layers.11.block_sparse_moe.experts.3.w3.weight": "model-00007-of-00019.safetensors",
113
+ "model.layers.11.block_sparse_moe.experts.4.w1.weight": "model-00007-of-00019.safetensors",
114
+ "model.layers.11.block_sparse_moe.experts.4.w2.weight": "model-00007-of-00019.safetensors",
115
+ "model.layers.11.block_sparse_moe.experts.4.w3.weight": "model-00007-of-00019.safetensors",
116
+ "model.layers.11.block_sparse_moe.experts.5.w1.weight": "model-00007-of-00019.safetensors",
117
+ "model.layers.11.block_sparse_moe.experts.5.w2.weight": "model-00007-of-00019.safetensors",
118
+ "model.layers.11.block_sparse_moe.experts.5.w3.weight": "model-00007-of-00019.safetensors",
119
+ "model.layers.11.block_sparse_moe.experts.6.w1.weight": "model-00007-of-00019.safetensors",
120
+ "model.layers.11.block_sparse_moe.experts.6.w2.weight": "model-00007-of-00019.safetensors",
121
+ "model.layers.11.block_sparse_moe.experts.6.w3.weight": "model-00008-of-00019.safetensors",
122
+ "model.layers.11.block_sparse_moe.experts.7.w1.weight": "model-00008-of-00019.safetensors",
123
+ "model.layers.11.block_sparse_moe.experts.7.w2.weight": "model-00008-of-00019.safetensors",
124
+ "model.layers.11.block_sparse_moe.experts.7.w3.weight": "model-00008-of-00019.safetensors",
125
+ "model.layers.11.block_sparse_moe.gate.weight": "model-00007-of-00019.safetensors",
126
+ "model.layers.11.input_layernorm.weight": "model-00008-of-00019.safetensors",
127
+ "model.layers.11.post_attention_layernorm.weight": "model-00008-of-00019.safetensors",
128
+ "model.layers.11.self_attn.k_proj.weight": "model-00007-of-00019.safetensors",
129
+ "model.layers.11.self_attn.o_proj.weight": "model-00007-of-00019.safetensors",
130
+ "model.layers.11.self_attn.q_proj.weight": "model-00007-of-00019.safetensors",
131
+ "model.layers.11.self_attn.v_proj.weight": "model-00007-of-00019.safetensors",
132
+ "model.layers.12.block_sparse_moe.experts.0.w1.weight": "model-00008-of-00019.safetensors",
133
+ "model.layers.12.block_sparse_moe.experts.0.w2.weight": "model-00008-of-00019.safetensors",
134
+ "model.layers.12.block_sparse_moe.experts.0.w3.weight": "model-00008-of-00019.safetensors",
135
+ "model.layers.12.block_sparse_moe.experts.1.w1.weight": "model-00008-of-00019.safetensors",
136
+ "model.layers.12.block_sparse_moe.experts.1.w2.weight": "model-00008-of-00019.safetensors",
137
+ "model.layers.12.block_sparse_moe.experts.1.w3.weight": "model-00008-of-00019.safetensors",
138
+ "model.layers.12.block_sparse_moe.experts.2.w1.weight": "model-00008-of-00019.safetensors",
139
+ "model.layers.12.block_sparse_moe.experts.2.w2.weight": "model-00008-of-00019.safetensors",
140
+ "model.layers.12.block_sparse_moe.experts.2.w3.weight": "model-00008-of-00019.safetensors",
141
+ "model.layers.12.block_sparse_moe.experts.3.w1.weight": "model-00008-of-00019.safetensors",
142
+ "model.layers.12.block_sparse_moe.experts.3.w2.weight": "model-00008-of-00019.safetensors",
143
+ "model.layers.12.block_sparse_moe.experts.3.w3.weight": "model-00008-of-00019.safetensors",
144
+ "model.layers.12.block_sparse_moe.experts.4.w1.weight": "model-00008-of-00019.safetensors",
145
+ "model.layers.12.block_sparse_moe.experts.4.w2.weight": "model-00008-of-00019.safetensors",
146
+ "model.layers.12.block_sparse_moe.experts.4.w3.weight": "model-00008-of-00019.safetensors",
147
+ "model.layers.12.block_sparse_moe.experts.5.w1.weight": "model-00008-of-00019.safetensors",
148
+ "model.layers.12.block_sparse_moe.experts.5.w2.weight": "model-00008-of-00019.safetensors",
149
+ "model.layers.12.block_sparse_moe.experts.5.w3.weight": "model-00008-of-00019.safetensors",
150
+ "model.layers.12.block_sparse_moe.experts.6.w1.weight": "model-00008-of-00019.safetensors",
151
+ "model.layers.12.block_sparse_moe.experts.6.w2.weight": "model-00008-of-00019.safetensors",
152
+ "model.layers.12.block_sparse_moe.experts.6.w3.weight": "model-00008-of-00019.safetensors",
153
+ "model.layers.12.block_sparse_moe.experts.7.w1.weight": "model-00008-of-00019.safetensors",
154
+ "model.layers.12.block_sparse_moe.experts.7.w2.weight": "model-00008-of-00019.safetensors",
155
+ "model.layers.12.block_sparse_moe.experts.7.w3.weight": "model-00008-of-00019.safetensors",
156
+ "model.layers.12.block_sparse_moe.gate.weight": "model-00008-of-00019.safetensors",
157
+ "model.layers.12.input_layernorm.weight": "model-00008-of-00019.safetensors",
158
+ "model.layers.12.post_attention_layernorm.weight": "model-00008-of-00019.safetensors",
159
+ "model.layers.12.self_attn.k_proj.weight": "model-00008-of-00019.safetensors",
160
+ "model.layers.12.self_attn.o_proj.weight": "model-00008-of-00019.safetensors",
161
+ "model.layers.12.self_attn.q_proj.weight": "model-00008-of-00019.safetensors",
162
+ "model.layers.12.self_attn.v_proj.weight": "model-00008-of-00019.safetensors",
163
+ "model.layers.13.block_sparse_moe.experts.0.w1.weight": "model-00008-of-00019.safetensors",
164
+ "model.layers.13.block_sparse_moe.experts.0.w2.weight": "model-00008-of-00019.safetensors",
165
+ "model.layers.13.block_sparse_moe.experts.0.w3.weight": "model-00008-of-00019.safetensors",
166
+ "model.layers.13.block_sparse_moe.experts.1.w1.weight": "model-00008-of-00019.safetensors",
167
+ "model.layers.13.block_sparse_moe.experts.1.w2.weight": "model-00008-of-00019.safetensors",
168
+ "model.layers.13.block_sparse_moe.experts.1.w3.weight": "model-00008-of-00019.safetensors",
169
+ "model.layers.13.block_sparse_moe.experts.2.w1.weight": "model-00008-of-00019.safetensors",
170
+ "model.layers.13.block_sparse_moe.experts.2.w2.weight": "model-00008-of-00019.safetensors",
171
+ "model.layers.13.block_sparse_moe.experts.2.w3.weight": "model-00008-of-00019.safetensors",
172
+ "model.layers.13.block_sparse_moe.experts.3.w1.weight": "model-00008-of-00019.safetensors",
173
+ "model.layers.13.block_sparse_moe.experts.3.w2.weight": "model-00008-of-00019.safetensors",
174
+ "model.layers.13.block_sparse_moe.experts.3.w3.weight": "model-00008-of-00019.safetensors",
175
+ "model.layers.13.block_sparse_moe.experts.4.w1.weight": "model-00008-of-00019.safetensors",
176
+ "model.layers.13.block_sparse_moe.experts.4.w2.weight": "model-00009-of-00019.safetensors",
177
+ "model.layers.13.block_sparse_moe.experts.4.w3.weight": "model-00009-of-00019.safetensors",
178
+ "model.layers.13.block_sparse_moe.experts.5.w1.weight": "model-00009-of-00019.safetensors",
179
+ "model.layers.13.block_sparse_moe.experts.5.w2.weight": "model-00009-of-00019.safetensors",
180
+ "model.layers.13.block_sparse_moe.experts.5.w3.weight": "model-00009-of-00019.safetensors",
181
+ "model.layers.13.block_sparse_moe.experts.6.w1.weight": "model-00009-of-00019.safetensors",
182
+ "model.layers.13.block_sparse_moe.experts.6.w2.weight": "model-00009-of-00019.safetensors",
183
+ "model.layers.13.block_sparse_moe.experts.6.w3.weight": "model-00009-of-00019.safetensors",
184
+ "model.layers.13.block_sparse_moe.experts.7.w1.weight": "model-00009-of-00019.safetensors",
185
+ "model.layers.13.block_sparse_moe.experts.7.w2.weight": "model-00009-of-00019.safetensors",
186
+ "model.layers.13.block_sparse_moe.experts.7.w3.weight": "model-00009-of-00019.safetensors",
187
+ "model.layers.13.block_sparse_moe.gate.weight": "model-00008-of-00019.safetensors",
188
+ "model.layers.13.input_layernorm.weight": "model-00009-of-00019.safetensors",
189
+ "model.layers.13.post_attention_layernorm.weight": "model-00009-of-00019.safetensors",
190
+ "model.layers.13.self_attn.k_proj.weight": "model-00008-of-00019.safetensors",
191
+ "model.layers.13.self_attn.o_proj.weight": "model-00008-of-00019.safetensors",
192
+ "model.layers.13.self_attn.q_proj.weight": "model-00008-of-00019.safetensors",
193
+ "model.layers.13.self_attn.v_proj.weight": "model-00008-of-00019.safetensors",
194
+ "model.layers.14.block_sparse_moe.experts.0.w1.weight": "model-00009-of-00019.safetensors",
195
+ "model.layers.14.block_sparse_moe.experts.0.w2.weight": "model-00009-of-00019.safetensors",
196
+ "model.layers.14.block_sparse_moe.experts.0.w3.weight": "model-00009-of-00019.safetensors",
197
+ "model.layers.14.block_sparse_moe.experts.1.w1.weight": "model-00009-of-00019.safetensors",
198
+ "model.layers.14.block_sparse_moe.experts.1.w2.weight": "model-00009-of-00019.safetensors",
199
+ "model.layers.14.block_sparse_moe.experts.1.w3.weight": "model-00009-of-00019.safetensors",
200
+ "model.layers.14.block_sparse_moe.experts.2.w1.weight": "model-00009-of-00019.safetensors",
201
+ "model.layers.14.block_sparse_moe.experts.2.w2.weight": "model-00009-of-00019.safetensors",
202
+ "model.layers.14.block_sparse_moe.experts.2.w3.weight": "model-00009-of-00019.safetensors",
203
+ "model.layers.14.block_sparse_moe.experts.3.w1.weight": "model-00009-of-00019.safetensors",
204
+ "model.layers.14.block_sparse_moe.experts.3.w2.weight": "model-00009-of-00019.safetensors",
205
+ "model.layers.14.block_sparse_moe.experts.3.w3.weight": "model-00009-of-00019.safetensors",
206
+ "model.layers.14.block_sparse_moe.experts.4.w1.weight": "model-00009-of-00019.safetensors",
207
+ "model.layers.14.block_sparse_moe.experts.4.w2.weight": "model-00009-of-00019.safetensors",
208
+ "model.layers.14.block_sparse_moe.experts.4.w3.weight": "model-00009-of-00019.safetensors",
209
+ "model.layers.14.block_sparse_moe.experts.5.w1.weight": "model-00009-of-00019.safetensors",
210
+ "model.layers.14.block_sparse_moe.experts.5.w2.weight": "model-00009-of-00019.safetensors",
211
+ "model.layers.14.block_sparse_moe.experts.5.w3.weight": "model-00009-of-00019.safetensors",
212
+ "model.layers.14.block_sparse_moe.experts.6.w1.weight": "model-00009-of-00019.safetensors",
213
+ "model.layers.14.block_sparse_moe.experts.6.w2.weight": "model-00009-of-00019.safetensors",
214
+ "model.layers.14.block_sparse_moe.experts.6.w3.weight": "model-00009-of-00019.safetensors",
215
+ "model.layers.14.block_sparse_moe.experts.7.w1.weight": "model-00009-of-00019.safetensors",
216
+ "model.layers.14.block_sparse_moe.experts.7.w2.weight": "model-00009-of-00019.safetensors",
217
+ "model.layers.14.block_sparse_moe.experts.7.w3.weight": "model-00009-of-00019.safetensors",
218
+ "model.layers.14.block_sparse_moe.gate.weight": "model-00009-of-00019.safetensors",
219
+ "model.layers.14.input_layernorm.weight": "model-00009-of-00019.safetensors",
220
+ "model.layers.14.post_attention_layernorm.weight": "model-00009-of-00019.safetensors",
221
+ "model.layers.14.self_attn.k_proj.weight": "model-00009-of-00019.safetensors",
222
+ "model.layers.14.self_attn.o_proj.weight": "model-00009-of-00019.safetensors",
223
+ "model.layers.14.self_attn.q_proj.weight": "model-00009-of-00019.safetensors",
224
+ "model.layers.14.self_attn.v_proj.weight": "model-00009-of-00019.safetensors",
225
+ "model.layers.15.block_sparse_moe.experts.0.w1.weight": "model-00009-of-00019.safetensors",
226
+ "model.layers.15.block_sparse_moe.experts.0.w2.weight": "model-00009-of-00019.safetensors",
227
+ "model.layers.15.block_sparse_moe.experts.0.w3.weight": "model-00009-of-00019.safetensors",
228
+ "model.layers.15.block_sparse_moe.experts.1.w1.weight": "model-00009-of-00019.safetensors",
229
+ "model.layers.15.block_sparse_moe.experts.1.w2.weight": "model-00009-of-00019.safetensors",
230
+ "model.layers.15.block_sparse_moe.experts.1.w3.weight": "model-00009-of-00019.safetensors",
231
+ "model.layers.15.block_sparse_moe.experts.2.w1.weight": "model-00010-of-00019.safetensors",
232
+ "model.layers.15.block_sparse_moe.experts.2.w2.weight": "model-00010-of-00019.safetensors",
233
+ "model.layers.15.block_sparse_moe.experts.2.w3.weight": "model-00010-of-00019.safetensors",
234
+ "model.layers.15.block_sparse_moe.experts.3.w1.weight": "model-00010-of-00019.safetensors",
235
+ "model.layers.15.block_sparse_moe.experts.3.w2.weight": "model-00010-of-00019.safetensors",
236
+ "model.layers.15.block_sparse_moe.experts.3.w3.weight": "model-00010-of-00019.safetensors",
237
+ "model.layers.15.block_sparse_moe.experts.4.w1.weight": "model-00010-of-00019.safetensors",
238
+ "model.layers.15.block_sparse_moe.experts.4.w2.weight": "model-00010-of-00019.safetensors",
239
+ "model.layers.15.block_sparse_moe.experts.4.w3.weight": "model-00010-of-00019.safetensors",
240
+ "model.layers.15.block_sparse_moe.experts.5.w1.weight": "model-00010-of-00019.safetensors",
241
+ "model.layers.15.block_sparse_moe.experts.5.w2.weight": "model-00010-of-00019.safetensors",
242
+ "model.layers.15.block_sparse_moe.experts.5.w3.weight": "model-00010-of-00019.safetensors",
243
+ "model.layers.15.block_sparse_moe.experts.6.w1.weight": "model-00010-of-00019.safetensors",
244
+ "model.layers.15.block_sparse_moe.experts.6.w2.weight": "model-00010-of-00019.safetensors",
245
+ "model.layers.15.block_sparse_moe.experts.6.w3.weight": "model-00010-of-00019.safetensors",
246
+ "model.layers.15.block_sparse_moe.experts.7.w1.weight": "model-00010-of-00019.safetensors",
247
+ "model.layers.15.block_sparse_moe.experts.7.w2.weight": "model-00010-of-00019.safetensors",
248
+ "model.layers.15.block_sparse_moe.experts.7.w3.weight": "model-00010-of-00019.safetensors",
249
+ "model.layers.15.block_sparse_moe.gate.weight": "model-00009-of-00019.safetensors",
250
+ "model.layers.15.input_layernorm.weight": "model-00010-of-00019.safetensors",
251
+ "model.layers.15.post_attention_layernorm.weight": "model-00010-of-00019.safetensors",
252
+ "model.layers.15.self_attn.k_proj.weight": "model-00009-of-00019.safetensors",
253
+ "model.layers.15.self_attn.o_proj.weight": "model-00009-of-00019.safetensors",
254
+ "model.layers.15.self_attn.q_proj.weight": "model-00009-of-00019.safetensors",
255
+ "model.layers.15.self_attn.v_proj.weight": "model-00009-of-00019.safetensors",
256
+ "model.layers.16.block_sparse_moe.experts.0.w1.weight": "model-00010-of-00019.safetensors",
257
+ "model.layers.16.block_sparse_moe.experts.0.w2.weight": "model-00010-of-00019.safetensors",
258
+ "model.layers.16.block_sparse_moe.experts.0.w3.weight": "model-00010-of-00019.safetensors",
259
+ "model.layers.16.block_sparse_moe.experts.1.w1.weight": "model-00010-of-00019.safetensors",
260
+ "model.layers.16.block_sparse_moe.experts.1.w2.weight": "model-00010-of-00019.safetensors",
261
+ "model.layers.16.block_sparse_moe.experts.1.w3.weight": "model-00010-of-00019.safetensors",
262
+ "model.layers.16.block_sparse_moe.experts.2.w1.weight": "model-00010-of-00019.safetensors",
263
+ "model.layers.16.block_sparse_moe.experts.2.w2.weight": "model-00010-of-00019.safetensors",
264
+ "model.layers.16.block_sparse_moe.experts.2.w3.weight": "model-00010-of-00019.safetensors",
265
+ "model.layers.16.block_sparse_moe.experts.3.w1.weight": "model-00010-of-00019.safetensors",
266
+ "model.layers.16.block_sparse_moe.experts.3.w2.weight": "model-00010-of-00019.safetensors",
267
+ "model.layers.16.block_sparse_moe.experts.3.w3.weight": "model-00010-of-00019.safetensors",
268
+ "model.layers.16.block_sparse_moe.experts.4.w1.weight": "model-00010-of-00019.safetensors",
269
+ "model.layers.16.block_sparse_moe.experts.4.w2.weight": "model-00010-of-00019.safetensors",
270
+ "model.layers.16.block_sparse_moe.experts.4.w3.weight": "model-00010-of-00019.safetensors",
271
+ "model.layers.16.block_sparse_moe.experts.5.w1.weight": "model-00010-of-00019.safetensors",
272
+ "model.layers.16.block_sparse_moe.experts.5.w2.weight": "model-00010-of-00019.safetensors",
273
+ "model.layers.16.block_sparse_moe.experts.5.w3.weight": "model-00010-of-00019.safetensors",
274
+ "model.layers.16.block_sparse_moe.experts.6.w1.weight": "model-00010-of-00019.safetensors",
275
+ "model.layers.16.block_sparse_moe.experts.6.w2.weight": "model-00010-of-00019.safetensors",
276
+ "model.layers.16.block_sparse_moe.experts.6.w3.weight": "model-00010-of-00019.safetensors",
277
+ "model.layers.16.block_sparse_moe.experts.7.w1.weight": "model-00010-of-00019.safetensors",
278
+ "model.layers.16.block_sparse_moe.experts.7.w2.weight": "model-00010-of-00019.safetensors",
279
+ "model.layers.16.block_sparse_moe.experts.7.w3.weight": "model-00011-of-00019.safetensors",
280
+ "model.layers.16.block_sparse_moe.gate.weight": "model-00010-of-00019.safetensors",
281
+ "model.layers.16.input_layernorm.weight": "model-00011-of-00019.safetensors",
282
+ "model.layers.16.post_attention_layernorm.weight": "model-00011-of-00019.safetensors",
283
+ "model.layers.16.self_attn.k_proj.weight": "model-00010-of-00019.safetensors",
284
+ "model.layers.16.self_attn.o_proj.weight": "model-00010-of-00019.safetensors",
285
+ "model.layers.16.self_attn.q_proj.weight": "model-00010-of-00019.safetensors",
286
+ "model.layers.16.self_attn.v_proj.weight": "model-00010-of-00019.safetensors",
287
+ "model.layers.17.block_sparse_moe.experts.0.w1.weight": "model-00011-of-00019.safetensors",
288
+ "model.layers.17.block_sparse_moe.experts.0.w2.weight": "model-00011-of-00019.safetensors",
289
+ "model.layers.17.block_sparse_moe.experts.0.w3.weight": "model-00011-of-00019.safetensors",
290
+ "model.layers.17.block_sparse_moe.experts.1.w1.weight": "model-00011-of-00019.safetensors",
291
+ "model.layers.17.block_sparse_moe.experts.1.w2.weight": "model-00011-of-00019.safetensors",
292
+ "model.layers.17.block_sparse_moe.experts.1.w3.weight": "model-00011-of-00019.safetensors",
293
+ "model.layers.17.block_sparse_moe.experts.2.w1.weight": "model-00011-of-00019.safetensors",
294
+ "model.layers.17.block_sparse_moe.experts.2.w2.weight": "model-00011-of-00019.safetensors",
295
+ "model.layers.17.block_sparse_moe.experts.2.w3.weight": "model-00011-of-00019.safetensors",
296
+ "model.layers.17.block_sparse_moe.experts.3.w1.weight": "model-00011-of-00019.safetensors",
297
+ "model.layers.17.block_sparse_moe.experts.3.w2.weight": "model-00011-of-00019.safetensors",
298
+ "model.layers.17.block_sparse_moe.experts.3.w3.weight": "model-00011-of-00019.safetensors",
299
+ "model.layers.17.block_sparse_moe.experts.4.w1.weight": "model-00011-of-00019.safetensors",
300
+ "model.layers.17.block_sparse_moe.experts.4.w2.weight": "model-00011-of-00019.safetensors",
301
+ "model.layers.17.block_sparse_moe.experts.4.w3.weight": "model-00011-of-00019.safetensors",
302
+ "model.layers.17.block_sparse_moe.experts.5.w1.weight": "model-00011-of-00019.safetensors",
303
+ "model.layers.17.block_sparse_moe.experts.5.w2.weight": "model-00011-of-00019.safetensors",
304
+ "model.layers.17.block_sparse_moe.experts.5.w3.weight": "model-00011-of-00019.safetensors",
305
+ "model.layers.17.block_sparse_moe.experts.6.w1.weight": "model-00011-of-00019.safetensors",
306
+ "model.layers.17.block_sparse_moe.experts.6.w2.weight": "model-00011-of-00019.safetensors",
307
+ "model.layers.17.block_sparse_moe.experts.6.w3.weight": "model-00011-of-00019.safetensors",
308
+ "model.layers.17.block_sparse_moe.experts.7.w1.weight": "model-00011-of-00019.safetensors",
309
+ "model.layers.17.block_sparse_moe.experts.7.w2.weight": "model-00011-of-00019.safetensors",
310
+ "model.layers.17.block_sparse_moe.experts.7.w3.weight": "model-00011-of-00019.safetensors",
311
+ "model.layers.17.block_sparse_moe.gate.weight": "model-00011-of-00019.safetensors",
312
+ "model.layers.17.input_layernorm.weight": "model-00011-of-00019.safetensors",
313
+ "model.layers.17.post_attention_layernorm.weight": "model-00011-of-00019.safetensors",
314
+ "model.layers.17.self_attn.k_proj.weight": "model-00011-of-00019.safetensors",
315
+ "model.layers.17.self_attn.o_proj.weight": "model-00011-of-00019.safetensors",
316
+ "model.layers.17.self_attn.q_proj.weight": "model-00011-of-00019.safetensors",
317
+ "model.layers.17.self_attn.v_proj.weight": "model-00011-of-00019.safetensors",
318
+ "model.layers.18.block_sparse_moe.experts.0.w1.weight": "model-00011-of-00019.safetensors",
319
+ "model.layers.18.block_sparse_moe.experts.0.w2.weight": "model-00011-of-00019.safetensors",
320
+ "model.layers.18.block_sparse_moe.experts.0.w3.weight": "model-00011-of-00019.safetensors",
321
+ "model.layers.18.block_sparse_moe.experts.1.w1.weight": "model-00011-of-00019.safetensors",
322
+ "model.layers.18.block_sparse_moe.experts.1.w2.weight": "model-00011-of-00019.safetensors",
323
+ "model.layers.18.block_sparse_moe.experts.1.w3.weight": "model-00011-of-00019.safetensors",
324
+ "model.layers.18.block_sparse_moe.experts.2.w1.weight": "model-00011-of-00019.safetensors",
325
+ "model.layers.18.block_sparse_moe.experts.2.w2.weight": "model-00011-of-00019.safetensors",
326
+ "model.layers.18.block_sparse_moe.experts.2.w3.weight": "model-00011-of-00019.safetensors",
327
+ "model.layers.18.block_sparse_moe.experts.3.w1.weight": "model-00011-of-00019.safetensors",
328
+ "model.layers.18.block_sparse_moe.experts.3.w2.weight": "model-00011-of-00019.safetensors",
329
+ "model.layers.18.block_sparse_moe.experts.3.w3.weight": "model-00011-of-00019.safetensors",
330
+ "model.layers.18.block_sparse_moe.experts.4.w1.weight": "model-00011-of-00019.safetensors",
331
+ "model.layers.18.block_sparse_moe.experts.4.w2.weight": "model-00011-of-00019.safetensors",
332
+ "model.layers.18.block_sparse_moe.experts.4.w3.weight": "model-00011-of-00019.safetensors",
333
+ "model.layers.18.block_sparse_moe.experts.5.w1.weight": "model-00011-of-00019.safetensors",
334
+ "model.layers.18.block_sparse_moe.experts.5.w2.weight": "model-00012-of-00019.safetensors",
335
+ "model.layers.18.block_sparse_moe.experts.5.w3.weight": "model-00012-of-00019.safetensors",
336
+ "model.layers.18.block_sparse_moe.experts.6.w1.weight": "model-00012-of-00019.safetensors",
337
+ "model.layers.18.block_sparse_moe.experts.6.w2.weight": "model-00012-of-00019.safetensors",
338
+ "model.layers.18.block_sparse_moe.experts.6.w3.weight": "model-00012-of-00019.safetensors",
339
+ "model.layers.18.block_sparse_moe.experts.7.w1.weight": "model-00012-of-00019.safetensors",
340
+ "model.layers.18.block_sparse_moe.experts.7.w2.weight": "model-00012-of-00019.safetensors",
341
+ "model.layers.18.block_sparse_moe.experts.7.w3.weight": "model-00012-of-00019.safetensors",
342
+ "model.layers.18.block_sparse_moe.gate.weight": "model-00011-of-00019.safetensors",
343
+ "model.layers.18.input_layernorm.weight": "model-00012-of-00019.safetensors",
344
+ "model.layers.18.post_attention_layernorm.weight": "model-00012-of-00019.safetensors",
345
+ "model.layers.18.self_attn.k_proj.weight": "model-00011-of-00019.safetensors",
346
+ "model.layers.18.self_attn.o_proj.weight": "model-00011-of-00019.safetensors",
347
+ "model.layers.18.self_attn.q_proj.weight": "model-00011-of-00019.safetensors",
348
+ "model.layers.18.self_attn.v_proj.weight": "model-00011-of-00019.safetensors",
349
+ "model.layers.19.block_sparse_moe.experts.0.w1.weight": "model-00012-of-00019.safetensors",
350
+ "model.layers.19.block_sparse_moe.experts.0.w2.weight": "model-00012-of-00019.safetensors",
351
+ "model.layers.19.block_sparse_moe.experts.0.w3.weight": "model-00012-of-00019.safetensors",
352
+ "model.layers.19.block_sparse_moe.experts.1.w1.weight": "model-00012-of-00019.safetensors",
353
+ "model.layers.19.block_sparse_moe.experts.1.w2.weight": "model-00012-of-00019.safetensors",
354
+ "model.layers.19.block_sparse_moe.experts.1.w3.weight": "model-00012-of-00019.safetensors",
355
+ "model.layers.19.block_sparse_moe.experts.2.w1.weight": "model-00012-of-00019.safetensors",
356
+ "model.layers.19.block_sparse_moe.experts.2.w2.weight": "model-00012-of-00019.safetensors",
357
+ "model.layers.19.block_sparse_moe.experts.2.w3.weight": "model-00012-of-00019.safetensors",
358
+ "model.layers.19.block_sparse_moe.experts.3.w1.weight": "model-00012-of-00019.safetensors",
359
+ "model.layers.19.block_sparse_moe.experts.3.w2.weight": "model-00012-of-00019.safetensors",
360
+ "model.layers.19.block_sparse_moe.experts.3.w3.weight": "model-00012-of-00019.safetensors",
361
+ "model.layers.19.block_sparse_moe.experts.4.w1.weight": "model-00012-of-00019.safetensors",
362
+ "model.layers.19.block_sparse_moe.experts.4.w2.weight": "model-00012-of-00019.safetensors",
363
+ "model.layers.19.block_sparse_moe.experts.4.w3.weight": "model-00012-of-00019.safetensors",
364
+ "model.layers.19.block_sparse_moe.experts.5.w1.weight": "model-00012-of-00019.safetensors",
365
+ "model.layers.19.block_sparse_moe.experts.5.w2.weight": "model-00012-of-00019.safetensors",
366
+ "model.layers.19.block_sparse_moe.experts.5.w3.weight": "model-00012-of-00019.safetensors",
367
+ "model.layers.19.block_sparse_moe.experts.6.w1.weight": "model-00012-of-00019.safetensors",
368
+ "model.layers.19.block_sparse_moe.experts.6.w2.weight": "model-00012-of-00019.safetensors",
369
+ "model.layers.19.block_sparse_moe.experts.6.w3.weight": "model-00012-of-00019.safetensors",
370
+ "model.layers.19.block_sparse_moe.experts.7.w1.weight": "model-00012-of-00019.safetensors",
371
+ "model.layers.19.block_sparse_moe.experts.7.w2.weight": "model-00012-of-00019.safetensors",
372
+ "model.layers.19.block_sparse_moe.experts.7.w3.weight": "model-00012-of-00019.safetensors",
373
+ "model.layers.19.block_sparse_moe.gate.weight": "model-00012-of-00019.safetensors",
374
+ "model.layers.19.input_layernorm.weight": "model-00012-of-00019.safetensors",
375
+ "model.layers.19.post_attention_layernorm.weight": "model-00012-of-00019.safetensors",
376
+ "model.layers.19.self_attn.k_proj.weight": "model-00012-of-00019.safetensors",
377
+ "model.layers.19.self_attn.o_proj.weight": "model-00012-of-00019.safetensors",
378
+ "model.layers.19.self_attn.q_proj.weight": "model-00012-of-00019.safetensors",
379
+ "model.layers.19.self_attn.v_proj.weight": "model-00012-of-00019.safetensors",
380
+ "model.layers.2.block_sparse_moe.experts.0.w1.weight": "model-00002-of-00019.safetensors",
381
+ "model.layers.2.block_sparse_moe.experts.0.w2.weight": "model-00002-of-00019.safetensors",
382
+ "model.layers.2.block_sparse_moe.experts.0.w3.weight": "model-00002-of-00019.safetensors",
383
+ "model.layers.2.block_sparse_moe.experts.1.w1.weight": "model-00002-of-00019.safetensors",
384
+ "model.layers.2.block_sparse_moe.experts.1.w2.weight": "model-00002-of-00019.safetensors",
385
+ "model.layers.2.block_sparse_moe.experts.1.w3.weight": "model-00002-of-00019.safetensors",
386
+ "model.layers.2.block_sparse_moe.experts.2.w1.weight": "model-00002-of-00019.safetensors",
387
+ "model.layers.2.block_sparse_moe.experts.2.w2.weight": "model-00002-of-00019.safetensors",
388
+ "model.layers.2.block_sparse_moe.experts.2.w3.weight": "model-00002-of-00019.safetensors",
389
+ "model.layers.2.block_sparse_moe.experts.3.w1.weight": "model-00002-of-00019.safetensors",
390
+ "model.layers.2.block_sparse_moe.experts.3.w2.weight": "model-00002-of-00019.safetensors",
391
+ "model.layers.2.block_sparse_moe.experts.3.w3.weight": "model-00002-of-00019.safetensors",
392
+ "model.layers.2.block_sparse_moe.experts.4.w1.weight": "model-00002-of-00019.safetensors",
393
+ "model.layers.2.block_sparse_moe.experts.4.w2.weight": "model-00002-of-00019.safetensors",
394
+ "model.layers.2.block_sparse_moe.experts.4.w3.weight": "model-00002-of-00019.safetensors",
395
+ "model.layers.2.block_sparse_moe.experts.5.w1.weight": "model-00002-of-00019.safetensors",
396
+ "model.layers.2.block_sparse_moe.experts.5.w2.weight": "model-00002-of-00019.safetensors",
397
+ "model.layers.2.block_sparse_moe.experts.5.w3.weight": "model-00002-of-00019.safetensors",
398
+ "model.layers.2.block_sparse_moe.experts.6.w1.weight": "model-00002-of-00019.safetensors",
399
+ "model.layers.2.block_sparse_moe.experts.6.w2.weight": "model-00002-of-00019.safetensors",
400
+ "model.layers.2.block_sparse_moe.experts.6.w3.weight": "model-00002-of-00019.safetensors",
401
+ "model.layers.2.block_sparse_moe.experts.7.w1.weight": "model-00002-of-00019.safetensors",
402
+ "model.layers.2.block_sparse_moe.experts.7.w2.weight": "model-00002-of-00019.safetensors",
403
+ "model.layers.2.block_sparse_moe.experts.7.w3.weight": "model-00002-of-00019.safetensors",
404
+ "model.layers.2.block_sparse_moe.gate.weight": "model-00002-of-00019.safetensors",
405
+ "model.layers.2.input_layernorm.weight": "model-00002-of-00019.safetensors",
406
+ "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00019.safetensors",
407
+ "model.layers.2.self_attn.k_proj.weight": "model-00002-of-00019.safetensors",
408
+ "model.layers.2.self_attn.o_proj.weight": "model-00002-of-00019.safetensors",
409
+ "model.layers.2.self_attn.q_proj.weight": "model-00002-of-00019.safetensors",
410
+ "model.layers.2.self_attn.v_proj.weight": "model-00002-of-00019.safetensors",
411
+ "model.layers.20.block_sparse_moe.experts.0.w1.weight": "model-00012-of-00019.safetensors",
412
+ "model.layers.20.block_sparse_moe.experts.0.w2.weight": "model-00012-of-00019.safetensors",
413
+ "model.layers.20.block_sparse_moe.experts.0.w3.weight": "model-00012-of-00019.safetensors",
414
+ "model.layers.20.block_sparse_moe.experts.1.w1.weight": "model-00012-of-00019.safetensors",
415
+ "model.layers.20.block_sparse_moe.experts.1.w2.weight": "model-00012-of-00019.safetensors",
416
+ "model.layers.20.block_sparse_moe.experts.1.w3.weight": "model-00012-of-00019.safetensors",
417
+ "model.layers.20.block_sparse_moe.experts.2.w1.weight": "model-00012-of-00019.safetensors",
418
+ "model.layers.20.block_sparse_moe.experts.2.w2.weight": "model-00012-of-00019.safetensors",
419
+ "model.layers.20.block_sparse_moe.experts.2.w3.weight": "model-00012-of-00019.safetensors",
420
+ "model.layers.20.block_sparse_moe.experts.3.w1.weight": "model-00013-of-00019.safetensors",
421
+ "model.layers.20.block_sparse_moe.experts.3.w2.weight": "model-00013-of-00019.safetensors",
422
+ "model.layers.20.block_sparse_moe.experts.3.w3.weight": "model-00013-of-00019.safetensors",
423
+ "model.layers.20.block_sparse_moe.experts.4.w1.weight": "model-00013-of-00019.safetensors",
424
+ "model.layers.20.block_sparse_moe.experts.4.w2.weight": "model-00013-of-00019.safetensors",
425
+ "model.layers.20.block_sparse_moe.experts.4.w3.weight": "model-00013-of-00019.safetensors",
426
+ "model.layers.20.block_sparse_moe.experts.5.w1.weight": "model-00013-of-00019.safetensors",
427
+ "model.layers.20.block_sparse_moe.experts.5.w2.weight": "model-00013-of-00019.safetensors",
428
+ "model.layers.20.block_sparse_moe.experts.5.w3.weight": "model-00013-of-00019.safetensors",
429
+ "model.layers.20.block_sparse_moe.experts.6.w1.weight": "model-00013-of-00019.safetensors",
430
+ "model.layers.20.block_sparse_moe.experts.6.w2.weight": "model-00013-of-00019.safetensors",
431
+ "model.layers.20.block_sparse_moe.experts.6.w3.weight": "model-00013-of-00019.safetensors",
432
+ "model.layers.20.block_sparse_moe.experts.7.w1.weight": "model-00013-of-00019.safetensors",
433
+ "model.layers.20.block_sparse_moe.experts.7.w2.weight": "model-00013-of-00019.safetensors",
434
+ "model.layers.20.block_sparse_moe.experts.7.w3.weight": "model-00013-of-00019.safetensors",
435
+ "model.layers.20.block_sparse_moe.gate.weight": "model-00012-of-00019.safetensors",
436
+ "model.layers.20.input_layernorm.weight": "model-00013-of-00019.safetensors",
437
+ "model.layers.20.post_attention_layernorm.weight": "model-00013-of-00019.safetensors",
438
+ "model.layers.20.self_attn.k_proj.weight": "model-00012-of-00019.safetensors",
439
+ "model.layers.20.self_attn.o_proj.weight": "model-00012-of-00019.safetensors",
440
+ "model.layers.20.self_attn.q_proj.weight": "model-00012-of-00019.safetensors",
441
+ "model.layers.20.self_attn.v_proj.weight": "model-00012-of-00019.safetensors",
442
+ "model.layers.21.block_sparse_moe.experts.0.w1.weight": "model-00013-of-00019.safetensors",
443
+ "model.layers.21.block_sparse_moe.experts.0.w2.weight": "model-00013-of-00019.safetensors",
444
+ "model.layers.21.block_sparse_moe.experts.0.w3.weight": "model-00013-of-00019.safetensors",
445
+ "model.layers.21.block_sparse_moe.experts.1.w1.weight": "model-00013-of-00019.safetensors",
446
+ "model.layers.21.block_sparse_moe.experts.1.w2.weight": "model-00013-of-00019.safetensors",
447
+ "model.layers.21.block_sparse_moe.experts.1.w3.weight": "model-00013-of-00019.safetensors",
448
+ "model.layers.21.block_sparse_moe.experts.2.w1.weight": "model-00013-of-00019.safetensors",
449
+ "model.layers.21.block_sparse_moe.experts.2.w2.weight": "model-00013-of-00019.safetensors",
450
+ "model.layers.21.block_sparse_moe.experts.2.w3.weight": "model-00013-of-00019.safetensors",
451
+ "model.layers.21.block_sparse_moe.experts.3.w1.weight": "model-00013-of-00019.safetensors",
452
+ "model.layers.21.block_sparse_moe.experts.3.w2.weight": "model-00013-of-00019.safetensors",
453
+ "model.layers.21.block_sparse_moe.experts.3.w3.weight": "model-00013-of-00019.safetensors",
454
+ "model.layers.21.block_sparse_moe.experts.4.w1.weight": "model-00013-of-00019.safetensors",
455
+ "model.layers.21.block_sparse_moe.experts.4.w2.weight": "model-00013-of-00019.safetensors",
456
+ "model.layers.21.block_sparse_moe.experts.4.w3.weight": "model-00013-of-00019.safetensors",
457
+ "model.layers.21.block_sparse_moe.experts.5.w1.weight": "model-00013-of-00019.safetensors",
458
+ "model.layers.21.block_sparse_moe.experts.5.w2.weight": "model-00013-of-00019.safetensors",
459
+ "model.layers.21.block_sparse_moe.experts.5.w3.weight": "model-00013-of-00019.safetensors",
460
+ "model.layers.21.block_sparse_moe.experts.6.w1.weight": "model-00013-of-00019.safetensors",
461
+ "model.layers.21.block_sparse_moe.experts.6.w2.weight": "model-00013-of-00019.safetensors",
462
+ "model.layers.21.block_sparse_moe.experts.6.w3.weight": "model-00013-of-00019.safetensors",
463
+ "model.layers.21.block_sparse_moe.experts.7.w1.weight": "model-00013-of-00019.safetensors",
464
+ "model.layers.21.block_sparse_moe.experts.7.w2.weight": "model-00013-of-00019.safetensors",
465
+ "model.layers.21.block_sparse_moe.experts.7.w3.weight": "model-00013-of-00019.safetensors",
466
+ "model.layers.21.block_sparse_moe.gate.weight": "model-00013-of-00019.safetensors",
467
+ "model.layers.21.input_layernorm.weight": "model-00013-of-00019.safetensors",
468
+ "model.layers.21.post_attention_layernorm.weight": "model-00013-of-00019.safetensors",
469
+ "model.layers.21.self_attn.k_proj.weight": "model-00013-of-00019.safetensors",
470
+ "model.layers.21.self_attn.o_proj.weight": "model-00013-of-00019.safetensors",
471
+ "model.layers.21.self_attn.q_proj.weight": "model-00013-of-00019.safetensors",
472
+ "model.layers.21.self_attn.v_proj.weight": "model-00013-of-00019.safetensors",
473
+ "model.layers.22.block_sparse_moe.experts.0.w1.weight": "model-00013-of-00019.safetensors",
474
+ "model.layers.22.block_sparse_moe.experts.0.w2.weight": "model-00013-of-00019.safetensors",
475
+ "model.layers.22.block_sparse_moe.experts.0.w3.weight": "model-00014-of-00019.safetensors",
476
+ "model.layers.22.block_sparse_moe.experts.1.w1.weight": "model-00014-of-00019.safetensors",
477
+ "model.layers.22.block_sparse_moe.experts.1.w2.weight": "model-00014-of-00019.safetensors",
478
+ "model.layers.22.block_sparse_moe.experts.1.w3.weight": "model-00014-of-00019.safetensors",
479
+ "model.layers.22.block_sparse_moe.experts.2.w1.weight": "model-00014-of-00019.safetensors",
480
+ "model.layers.22.block_sparse_moe.experts.2.w2.weight": "model-00014-of-00019.safetensors",
481
+ "model.layers.22.block_sparse_moe.experts.2.w3.weight": "model-00014-of-00019.safetensors",
482
+ "model.layers.22.block_sparse_moe.experts.3.w1.weight": "model-00014-of-00019.safetensors",
483
+ "model.layers.22.block_sparse_moe.experts.3.w2.weight": "model-00014-of-00019.safetensors",
484
+ "model.layers.22.block_sparse_moe.experts.3.w3.weight": "model-00014-of-00019.safetensors",
485
+ "model.layers.22.block_sparse_moe.experts.4.w1.weight": "model-00014-of-00019.safetensors",
486
+ "model.layers.22.block_sparse_moe.experts.4.w2.weight": "model-00014-of-00019.safetensors",
487
+ "model.layers.22.block_sparse_moe.experts.4.w3.weight": "model-00014-of-00019.safetensors",
488
+ "model.layers.22.block_sparse_moe.experts.5.w1.weight": "model-00014-of-00019.safetensors",
489
+ "model.layers.22.block_sparse_moe.experts.5.w2.weight": "model-00014-of-00019.safetensors",
490
+ "model.layers.22.block_sparse_moe.experts.5.w3.weight": "model-00014-of-00019.safetensors",
491
+ "model.layers.22.block_sparse_moe.experts.6.w1.weight": "model-00014-of-00019.safetensors",
492
+ "model.layers.22.block_sparse_moe.experts.6.w2.weight": "model-00014-of-00019.safetensors",
493
+ "model.layers.22.block_sparse_moe.experts.6.w3.weight": "model-00014-of-00019.safetensors",
494
+ "model.layers.22.block_sparse_moe.experts.7.w1.weight": "model-00014-of-00019.safetensors",
495
+ "model.layers.22.block_sparse_moe.experts.7.w2.weight": "model-00014-of-00019.safetensors",
496
+ "model.layers.22.block_sparse_moe.experts.7.w3.weight": "model-00014-of-00019.safetensors",
497
+ "model.layers.22.block_sparse_moe.gate.weight": "model-00013-of-00019.safetensors",
498
+ "model.layers.22.input_layernorm.weight": "model-00014-of-00019.safetensors",
499
+ "model.layers.22.post_attention_layernorm.weight": "model-00014-of-00019.safetensors",
500
+ "model.layers.22.self_attn.k_proj.weight": "model-00013-of-00019.safetensors",
501
+ "model.layers.22.self_attn.o_proj.weight": "model-00013-of-00019.safetensors",
502
+ "model.layers.22.self_attn.q_proj.weight": "model-00013-of-00019.safetensors",
503
+ "model.layers.22.self_attn.v_proj.weight": "model-00013-of-00019.safetensors",
504
+ "model.layers.23.block_sparse_moe.experts.0.w1.weight": "model-00014-of-00019.safetensors",
505
+ "model.layers.23.block_sparse_moe.experts.0.w2.weight": "model-00014-of-00019.safetensors",
506
+ "model.layers.23.block_sparse_moe.experts.0.w3.weight": "model-00014-of-00019.safetensors",
507
+ "model.layers.23.block_sparse_moe.experts.1.w1.weight": "model-00014-of-00019.safetensors",
508
+ "model.layers.23.block_sparse_moe.experts.1.w2.weight": "model-00014-of-00019.safetensors",
509
+ "model.layers.23.block_sparse_moe.experts.1.w3.weight": "model-00014-of-00019.safetensors",
510
+ "model.layers.23.block_sparse_moe.experts.2.w1.weight": "model-00014-of-00019.safetensors",
511
+ "model.layers.23.block_sparse_moe.experts.2.w2.weight": "model-00014-of-00019.safetensors",
512
+ "model.layers.23.block_sparse_moe.experts.2.w3.weight": "model-00014-of-00019.safetensors",
513
+ "model.layers.23.block_sparse_moe.experts.3.w1.weight": "model-00014-of-00019.safetensors",
514
+ "model.layers.23.block_sparse_moe.experts.3.w2.weight": "model-00014-of-00019.safetensors",
515
+ "model.layers.23.block_sparse_moe.experts.3.w3.weight": "model-00014-of-00019.safetensors",
516
+ "model.layers.23.block_sparse_moe.experts.4.w1.weight": "model-00014-of-00019.safetensors",
517
+ "model.layers.23.block_sparse_moe.experts.4.w2.weight": "model-00014-of-00019.safetensors",
518
+ "model.layers.23.block_sparse_moe.experts.4.w3.weight": "model-00014-of-00019.safetensors",
519
+ "model.layers.23.block_sparse_moe.experts.5.w1.weight": "model-00014-of-00019.safetensors",
520
+ "model.layers.23.block_sparse_moe.experts.5.w2.weight": "model-00014-of-00019.safetensors",
521
+ "model.layers.23.block_sparse_moe.experts.5.w3.weight": "model-00014-of-00019.safetensors",
522
+ "model.layers.23.block_sparse_moe.experts.6.w1.weight": "model-00014-of-00019.safetensors",
523
+ "model.layers.23.block_sparse_moe.experts.6.w2.weight": "model-00015-of-00019.safetensors",
524
+ "model.layers.23.block_sparse_moe.experts.6.w3.weight": "model-00015-of-00019.safetensors",
525
+ "model.layers.23.block_sparse_moe.experts.7.w1.weight": "model-00015-of-00019.safetensors",
526
+ "model.layers.23.block_sparse_moe.experts.7.w2.weight": "model-00015-of-00019.safetensors",
527
+ "model.layers.23.block_sparse_moe.experts.7.w3.weight": "model-00015-of-00019.safetensors",
528
+ "model.layers.23.block_sparse_moe.gate.weight": "model-00014-of-00019.safetensors",
529
+ "model.layers.23.input_layernorm.weight": "model-00015-of-00019.safetensors",
530
+ "model.layers.23.post_attention_layernorm.weight": "model-00015-of-00019.safetensors",
531
+ "model.layers.23.self_attn.k_proj.weight": "model-00014-of-00019.safetensors",
532
+ "model.layers.23.self_attn.o_proj.weight": "model-00014-of-00019.safetensors",
533
+ "model.layers.23.self_attn.q_proj.weight": "model-00014-of-00019.safetensors",
534
+ "model.layers.23.self_attn.v_proj.weight": "model-00014-of-00019.safetensors",
535
+ "model.layers.24.block_sparse_moe.experts.0.w1.weight": "model-00015-of-00019.safetensors",
536
+ "model.layers.24.block_sparse_moe.experts.0.w2.weight": "model-00015-of-00019.safetensors",
537
+ "model.layers.24.block_sparse_moe.experts.0.w3.weight": "model-00015-of-00019.safetensors",
538
+ "model.layers.24.block_sparse_moe.experts.1.w1.weight": "model-00015-of-00019.safetensors",
539
+ "model.layers.24.block_sparse_moe.experts.1.w2.weight": "model-00015-of-00019.safetensors",
540
+ "model.layers.24.block_sparse_moe.experts.1.w3.weight": "model-00015-of-00019.safetensors",
541
+ "model.layers.24.block_sparse_moe.experts.2.w1.weight": "model-00015-of-00019.safetensors",
542
+ "model.layers.24.block_sparse_moe.experts.2.w2.weight": "model-00015-of-00019.safetensors",
543
+ "model.layers.24.block_sparse_moe.experts.2.w3.weight": "model-00015-of-00019.safetensors",
544
+ "model.layers.24.block_sparse_moe.experts.3.w1.weight": "model-00015-of-00019.safetensors",
545
+ "model.layers.24.block_sparse_moe.experts.3.w2.weight": "model-00015-of-00019.safetensors",
546
+ "model.layers.24.block_sparse_moe.experts.3.w3.weight": "model-00015-of-00019.safetensors",
547
+ "model.layers.24.block_sparse_moe.experts.4.w1.weight": "model-00015-of-00019.safetensors",
548
+ "model.layers.24.block_sparse_moe.experts.4.w2.weight": "model-00015-of-00019.safetensors",
549
+ "model.layers.24.block_sparse_moe.experts.4.w3.weight": "model-00015-of-00019.safetensors",
550
+ "model.layers.24.block_sparse_moe.experts.5.w1.weight": "model-00015-of-00019.safetensors",
551
+ "model.layers.24.block_sparse_moe.experts.5.w2.weight": "model-00015-of-00019.safetensors",
552
+ "model.layers.24.block_sparse_moe.experts.5.w3.weight": "model-00015-of-00019.safetensors",
553
+ "model.layers.24.block_sparse_moe.experts.6.w1.weight": "model-00015-of-00019.safetensors",
554
+ "model.layers.24.block_sparse_moe.experts.6.w2.weight": "model-00015-of-00019.safetensors",
555
+ "model.layers.24.block_sparse_moe.experts.6.w3.weight": "model-00015-of-00019.safetensors",
556
+ "model.layers.24.block_sparse_moe.experts.7.w1.weight": "model-00015-of-00019.safetensors",
557
+ "model.layers.24.block_sparse_moe.experts.7.w2.weight": "model-00015-of-00019.safetensors",
558
+ "model.layers.24.block_sparse_moe.experts.7.w3.weight": "model-00015-of-00019.safetensors",
559
+ "model.layers.24.block_sparse_moe.gate.weight": "model-00015-of-00019.safetensors",
560
+ "model.layers.24.input_layernorm.weight": "model-00015-of-00019.safetensors",
561
+ "model.layers.24.post_attention_layernorm.weight": "model-00015-of-00019.safetensors",
562
+ "model.layers.24.self_attn.k_proj.weight": "model-00015-of-00019.safetensors",
563
+ "model.layers.24.self_attn.o_proj.weight": "model-00015-of-00019.safetensors",
564
+ "model.layers.24.self_attn.q_proj.weight": "model-00015-of-00019.safetensors",
565
+ "model.layers.24.self_attn.v_proj.weight": "model-00015-of-00019.safetensors",
566
+ "model.layers.25.block_sparse_moe.experts.0.w1.weight": "model-00015-of-00019.safetensors",
567
+ "model.layers.25.block_sparse_moe.experts.0.w2.weight": "model-00015-of-00019.safetensors",
568
+ "model.layers.25.block_sparse_moe.experts.0.w3.weight": "model-00015-of-00019.safetensors",
569
+ "model.layers.25.block_sparse_moe.experts.1.w1.weight": "model-00015-of-00019.safetensors",
570
+ "model.layers.25.block_sparse_moe.experts.1.w2.weight": "model-00015-of-00019.safetensors",
571
+ "model.layers.25.block_sparse_moe.experts.1.w3.weight": "model-00015-of-00019.safetensors",
572
+ "model.layers.25.block_sparse_moe.experts.2.w1.weight": "model-00015-of-00019.safetensors",
573
+ "model.layers.25.block_sparse_moe.experts.2.w2.weight": "model-00015-of-00019.safetensors",
574
+ "model.layers.25.block_sparse_moe.experts.2.w3.weight": "model-00015-of-00019.safetensors",
575
+ "model.layers.25.block_sparse_moe.experts.3.w1.weight": "model-00015-of-00019.safetensors",
576
+ "model.layers.25.block_sparse_moe.experts.3.w2.weight": "model-00015-of-00019.safetensors",
577
+ "model.layers.25.block_sparse_moe.experts.3.w3.weight": "model-00015-of-00019.safetensors",
578
+ "model.layers.25.block_sparse_moe.experts.4.w1.weight": "model-00016-of-00019.safetensors",
579
+ "model.layers.25.block_sparse_moe.experts.4.w2.weight": "model-00016-of-00019.safetensors",
580
+ "model.layers.25.block_sparse_moe.experts.4.w3.weight": "model-00016-of-00019.safetensors",
581
+ "model.layers.25.block_sparse_moe.experts.5.w1.weight": "model-00016-of-00019.safetensors",
582
+ "model.layers.25.block_sparse_moe.experts.5.w2.weight": "model-00016-of-00019.safetensors",
583
+ "model.layers.25.block_sparse_moe.experts.5.w3.weight": "model-00016-of-00019.safetensors",
584
+ "model.layers.25.block_sparse_moe.experts.6.w1.weight": "model-00016-of-00019.safetensors",
585
+ "model.layers.25.block_sparse_moe.experts.6.w2.weight": "model-00016-of-00019.safetensors",
586
+ "model.layers.25.block_sparse_moe.experts.6.w3.weight": "model-00016-of-00019.safetensors",
587
+ "model.layers.25.block_sparse_moe.experts.7.w1.weight": "model-00016-of-00019.safetensors",
588
+ "model.layers.25.block_sparse_moe.experts.7.w2.weight": "model-00016-of-00019.safetensors",
589
+ "model.layers.25.block_sparse_moe.experts.7.w3.weight": "model-00016-of-00019.safetensors",
590
+ "model.layers.25.block_sparse_moe.gate.weight": "model-00015-of-00019.safetensors",
591
+ "model.layers.25.input_layernorm.weight": "model-00016-of-00019.safetensors",
592
+ "model.layers.25.post_attention_layernorm.weight": "model-00016-of-00019.safetensors",
593
+ "model.layers.25.self_attn.k_proj.weight": "model-00015-of-00019.safetensors",
594
+ "model.layers.25.self_attn.o_proj.weight": "model-00015-of-00019.safetensors",
595
+ "model.layers.25.self_attn.q_proj.weight": "model-00015-of-00019.safetensors",
596
+ "model.layers.25.self_attn.v_proj.weight": "model-00015-of-00019.safetensors",
597
+ "model.layers.26.block_sparse_moe.experts.0.w1.weight": "model-00016-of-00019.safetensors",
598
+ "model.layers.26.block_sparse_moe.experts.0.w2.weight": "model-00016-of-00019.safetensors",
599
+ "model.layers.26.block_sparse_moe.experts.0.w3.weight": "model-00016-of-00019.safetensors",
600
+ "model.layers.26.block_sparse_moe.experts.1.w1.weight": "model-00016-of-00019.safetensors",
601
+ "model.layers.26.block_sparse_moe.experts.1.w2.weight": "model-00016-of-00019.safetensors",
602
+ "model.layers.26.block_sparse_moe.experts.1.w3.weight": "model-00016-of-00019.safetensors",
603
+ "model.layers.26.block_sparse_moe.experts.2.w1.weight": "model-00016-of-00019.safetensors",
604
+ "model.layers.26.block_sparse_moe.experts.2.w2.weight": "model-00016-of-00019.safetensors",
605
+ "model.layers.26.block_sparse_moe.experts.2.w3.weight": "model-00016-of-00019.safetensors",
606
+ "model.layers.26.block_sparse_moe.experts.3.w1.weight": "model-00016-of-00019.safetensors",
607
+ "model.layers.26.block_sparse_moe.experts.3.w2.weight": "model-00016-of-00019.safetensors",
608
+ "model.layers.26.block_sparse_moe.experts.3.w3.weight": "model-00016-of-00019.safetensors",
609
+ "model.layers.26.block_sparse_moe.experts.4.w1.weight": "model-00016-of-00019.safetensors",
610
+ "model.layers.26.block_sparse_moe.experts.4.w2.weight": "model-00016-of-00019.safetensors",
611
+ "model.layers.26.block_sparse_moe.experts.4.w3.weight": "model-00016-of-00019.safetensors",
612
+ "model.layers.26.block_sparse_moe.experts.5.w1.weight": "model-00016-of-00019.safetensors",
613
+ "model.layers.26.block_sparse_moe.experts.5.w2.weight": "model-00016-of-00019.safetensors",
614
+ "model.layers.26.block_sparse_moe.experts.5.w3.weight": "model-00016-of-00019.safetensors",
615
+ "model.layers.26.block_sparse_moe.experts.6.w1.weight": "model-00016-of-00019.safetensors",
616
+ "model.layers.26.block_sparse_moe.experts.6.w2.weight": "model-00016-of-00019.safetensors",
617
+ "model.layers.26.block_sparse_moe.experts.6.w3.weight": "model-00016-of-00019.safetensors",
618
+ "model.layers.26.block_sparse_moe.experts.7.w1.weight": "model-00016-of-00019.safetensors",
619
+ "model.layers.26.block_sparse_moe.experts.7.w2.weight": "model-00016-of-00019.safetensors",
620
+ "model.layers.26.block_sparse_moe.experts.7.w3.weight": "model-00016-of-00019.safetensors",
621
+ "model.layers.26.block_sparse_moe.gate.weight": "model-00016-of-00019.safetensors",
622
+ "model.layers.26.input_layernorm.weight": "model-00016-of-00019.safetensors",
623
+ "model.layers.26.post_attention_layernorm.weight": "model-00016-of-00019.safetensors",
624
+ "model.layers.26.self_attn.k_proj.weight": "model-00016-of-00019.safetensors",
625
+ "model.layers.26.self_attn.o_proj.weight": "model-00016-of-00019.safetensors",
626
+ "model.layers.26.self_attn.q_proj.weight": "model-00016-of-00019.safetensors",
627
+ "model.layers.26.self_attn.v_proj.weight": "model-00016-of-00019.safetensors",
628
+ "model.layers.27.block_sparse_moe.experts.0.w1.weight": "model-00016-of-00019.safetensors",
629
+ "model.layers.27.block_sparse_moe.experts.0.w2.weight": "model-00016-of-00019.safetensors",
630
+ "model.layers.27.block_sparse_moe.experts.0.w3.weight": "model-00016-of-00019.safetensors",
631
+ "model.layers.27.block_sparse_moe.experts.1.w1.weight": "model-00016-of-00019.safetensors",
632
+ "model.layers.27.block_sparse_moe.experts.1.w2.weight": "model-00016-of-00019.safetensors",
633
+ "model.layers.27.block_sparse_moe.experts.1.w3.weight": "model-00017-of-00019.safetensors",
634
+ "model.layers.27.block_sparse_moe.experts.2.w1.weight": "model-00017-of-00019.safetensors",
635
+ "model.layers.27.block_sparse_moe.experts.2.w2.weight": "model-00017-of-00019.safetensors",
636
+ "model.layers.27.block_sparse_moe.experts.2.w3.weight": "model-00017-of-00019.safetensors",
637
+ "model.layers.27.block_sparse_moe.experts.3.w1.weight": "model-00017-of-00019.safetensors",
638
+ "model.layers.27.block_sparse_moe.experts.3.w2.weight": "model-00017-of-00019.safetensors",
639
+ "model.layers.27.block_sparse_moe.experts.3.w3.weight": "model-00017-of-00019.safetensors",
640
+ "model.layers.27.block_sparse_moe.experts.4.w1.weight": "model-00017-of-00019.safetensors",
641
+ "model.layers.27.block_sparse_moe.experts.4.w2.weight": "model-00017-of-00019.safetensors",
642
+ "model.layers.27.block_sparse_moe.experts.4.w3.weight": "model-00017-of-00019.safetensors",
643
+ "model.layers.27.block_sparse_moe.experts.5.w1.weight": "model-00017-of-00019.safetensors",
644
+ "model.layers.27.block_sparse_moe.experts.5.w2.weight": "model-00017-of-00019.safetensors",
645
+ "model.layers.27.block_sparse_moe.experts.5.w3.weight": "model-00017-of-00019.safetensors",
646
+ "model.layers.27.block_sparse_moe.experts.6.w1.weight": "model-00017-of-00019.safetensors",
647
+ "model.layers.27.block_sparse_moe.experts.6.w2.weight": "model-00017-of-00019.safetensors",
648
+ "model.layers.27.block_sparse_moe.experts.6.w3.weight": "model-00017-of-00019.safetensors",
649
+ "model.layers.27.block_sparse_moe.experts.7.w1.weight": "model-00017-of-00019.safetensors",
650
+ "model.layers.27.block_sparse_moe.experts.7.w2.weight": "model-00017-of-00019.safetensors",
651
+ "model.layers.27.block_sparse_moe.experts.7.w3.weight": "model-00017-of-00019.safetensors",
652
+ "model.layers.27.block_sparse_moe.gate.weight": "model-00016-of-00019.safetensors",
653
+ "model.layers.27.input_layernorm.weight": "model-00017-of-00019.safetensors",
654
+ "model.layers.27.post_attention_layernorm.weight": "model-00017-of-00019.safetensors",
655
+ "model.layers.27.self_attn.k_proj.weight": "model-00016-of-00019.safetensors",
656
+ "model.layers.27.self_attn.o_proj.weight": "model-00016-of-00019.safetensors",
657
+ "model.layers.27.self_attn.q_proj.weight": "model-00016-of-00019.safetensors",
658
+ "model.layers.27.self_attn.v_proj.weight": "model-00016-of-00019.safetensors",
659
+ "model.layers.28.block_sparse_moe.experts.0.w1.weight": "model-00017-of-00019.safetensors",
660
+ "model.layers.28.block_sparse_moe.experts.0.w2.weight": "model-00017-of-00019.safetensors",
661
+ "model.layers.28.block_sparse_moe.experts.0.w3.weight": "model-00017-of-00019.safetensors",
662
+ "model.layers.28.block_sparse_moe.experts.1.w1.weight": "model-00017-of-00019.safetensors",
663
+ "model.layers.28.block_sparse_moe.experts.1.w2.weight": "model-00017-of-00019.safetensors",
664
+ "model.layers.28.block_sparse_moe.experts.1.w3.weight": "model-00017-of-00019.safetensors",
665
+ "model.layers.28.block_sparse_moe.experts.2.w1.weight": "model-00017-of-00019.safetensors",
666
+ "model.layers.28.block_sparse_moe.experts.2.w2.weight": "model-00017-of-00019.safetensors",
667
+ "model.layers.28.block_sparse_moe.experts.2.w3.weight": "model-00017-of-00019.safetensors",
668
+ "model.layers.28.block_sparse_moe.experts.3.w1.weight": "model-00017-of-00019.safetensors",
669
+ "model.layers.28.block_sparse_moe.experts.3.w2.weight": "model-00017-of-00019.safetensors",
670
+ "model.layers.28.block_sparse_moe.experts.3.w3.weight": "model-00017-of-00019.safetensors",
671
+ "model.layers.28.block_sparse_moe.experts.4.w1.weight": "model-00017-of-00019.safetensors",
672
+ "model.layers.28.block_sparse_moe.experts.4.w2.weight": "model-00017-of-00019.safetensors",
673
+ "model.layers.28.block_sparse_moe.experts.4.w3.weight": "model-00017-of-00019.safetensors",
674
+ "model.layers.28.block_sparse_moe.experts.5.w1.weight": "model-00017-of-00019.safetensors",
675
+ "model.layers.28.block_sparse_moe.experts.5.w2.weight": "model-00017-of-00019.safetensors",
676
+ "model.layers.28.block_sparse_moe.experts.5.w3.weight": "model-00017-of-00019.safetensors",
677
+ "model.layers.28.block_sparse_moe.experts.6.w1.weight": "model-00017-of-00019.safetensors",
678
+ "model.layers.28.block_sparse_moe.experts.6.w2.weight": "model-00017-of-00019.safetensors",
679
+ "model.layers.28.block_sparse_moe.experts.6.w3.weight": "model-00017-of-00019.safetensors",
680
+ "model.layers.28.block_sparse_moe.experts.7.w1.weight": "model-00017-of-00019.safetensors",
681
+ "model.layers.28.block_sparse_moe.experts.7.w2.weight": "model-00018-of-00019.safetensors",
682
+ "model.layers.28.block_sparse_moe.experts.7.w3.weight": "model-00018-of-00019.safetensors",
683
+ "model.layers.28.block_sparse_moe.gate.weight": "model-00017-of-00019.safetensors",
684
+ "model.layers.28.input_layernorm.weight": "model-00018-of-00019.safetensors",
685
+ "model.layers.28.post_attention_layernorm.weight": "model-00018-of-00019.safetensors",
686
+ "model.layers.28.self_attn.k_proj.weight": "model-00017-of-00019.safetensors",
687
+ "model.layers.28.self_attn.o_proj.weight": "model-00017-of-00019.safetensors",
688
+ "model.layers.28.self_attn.q_proj.weight": "model-00017-of-00019.safetensors",
689
+ "model.layers.28.self_attn.v_proj.weight": "model-00017-of-00019.safetensors",
690
+ "model.layers.29.block_sparse_moe.experts.0.w1.weight": "model-00018-of-00019.safetensors",
691
+ "model.layers.29.block_sparse_moe.experts.0.w2.weight": "model-00018-of-00019.safetensors",
692
+ "model.layers.29.block_sparse_moe.experts.0.w3.weight": "model-00018-of-00019.safetensors",
693
+ "model.layers.29.block_sparse_moe.experts.1.w1.weight": "model-00018-of-00019.safetensors",
694
+ "model.layers.29.block_sparse_moe.experts.1.w2.weight": "model-00018-of-00019.safetensors",
695
+ "model.layers.29.block_sparse_moe.experts.1.w3.weight": "model-00018-of-00019.safetensors",
696
+ "model.layers.29.block_sparse_moe.experts.2.w1.weight": "model-00018-of-00019.safetensors",
697
+ "model.layers.29.block_sparse_moe.experts.2.w2.weight": "model-00018-of-00019.safetensors",
698
+ "model.layers.29.block_sparse_moe.experts.2.w3.weight": "model-00018-of-00019.safetensors",
699
+ "model.layers.29.block_sparse_moe.experts.3.w1.weight": "model-00018-of-00019.safetensors",
700
+ "model.layers.29.block_sparse_moe.experts.3.w2.weight": "model-00018-of-00019.safetensors",
701
+ "model.layers.29.block_sparse_moe.experts.3.w3.weight": "model-00018-of-00019.safetensors",
702
+ "model.layers.29.block_sparse_moe.experts.4.w1.weight": "model-00018-of-00019.safetensors",
703
+ "model.layers.29.block_sparse_moe.experts.4.w2.weight": "model-00018-of-00019.safetensors",
704
+ "model.layers.29.block_sparse_moe.experts.4.w3.weight": "model-00018-of-00019.safetensors",
705
+ "model.layers.29.block_sparse_moe.experts.5.w1.weight": "model-00018-of-00019.safetensors",
706
+ "model.layers.29.block_sparse_moe.experts.5.w2.weight": "model-00018-of-00019.safetensors",
707
+ "model.layers.29.block_sparse_moe.experts.5.w3.weight": "model-00018-of-00019.safetensors",
708
+ "model.layers.29.block_sparse_moe.experts.6.w1.weight": "model-00018-of-00019.safetensors",
709
+ "model.layers.29.block_sparse_moe.experts.6.w2.weight": "model-00018-of-00019.safetensors",
710
+ "model.layers.29.block_sparse_moe.experts.6.w3.weight": "model-00018-of-00019.safetensors",
711
+ "model.layers.29.block_sparse_moe.experts.7.w1.weight": "model-00018-of-00019.safetensors",
712
+ "model.layers.29.block_sparse_moe.experts.7.w2.weight": "model-00018-of-00019.safetensors",
713
+ "model.layers.29.block_sparse_moe.experts.7.w3.weight": "model-00018-of-00019.safetensors",
714
+ "model.layers.29.block_sparse_moe.gate.weight": "model-00018-of-00019.safetensors",
715
+ "model.layers.29.input_layernorm.weight": "model-00018-of-00019.safetensors",
716
+ "model.layers.29.post_attention_layernorm.weight": "model-00018-of-00019.safetensors",
717
+ "model.layers.29.self_attn.k_proj.weight": "model-00018-of-00019.safetensors",
718
+ "model.layers.29.self_attn.o_proj.weight": "model-00018-of-00019.safetensors",
719
+ "model.layers.29.self_attn.q_proj.weight": "model-00018-of-00019.safetensors",
720
+ "model.layers.29.self_attn.v_proj.weight": "model-00018-of-00019.safetensors",
721
+ "model.layers.3.block_sparse_moe.experts.0.w1.weight": "model-00002-of-00019.safetensors",
722
+ "model.layers.3.block_sparse_moe.experts.0.w2.weight": "model-00002-of-00019.safetensors",
723
+ "model.layers.3.block_sparse_moe.experts.0.w3.weight": "model-00002-of-00019.safetensors",
724
+ "model.layers.3.block_sparse_moe.experts.1.w1.weight": "model-00002-of-00019.safetensors",
725
+ "model.layers.3.block_sparse_moe.experts.1.w2.weight": "model-00002-of-00019.safetensors",
726
+ "model.layers.3.block_sparse_moe.experts.1.w3.weight": "model-00002-of-00019.safetensors",
727
+ "model.layers.3.block_sparse_moe.experts.2.w1.weight": "model-00002-of-00019.safetensors",
728
+ "model.layers.3.block_sparse_moe.experts.2.w2.weight": "model-00003-of-00019.safetensors",
729
+ "model.layers.3.block_sparse_moe.experts.2.w3.weight": "model-00003-of-00019.safetensors",
730
+ "model.layers.3.block_sparse_moe.experts.3.w1.weight": "model-00003-of-00019.safetensors",
731
+ "model.layers.3.block_sparse_moe.experts.3.w2.weight": "model-00003-of-00019.safetensors",
732
+ "model.layers.3.block_sparse_moe.experts.3.w3.weight": "model-00003-of-00019.safetensors",
733
+ "model.layers.3.block_sparse_moe.experts.4.w1.weight": "model-00003-of-00019.safetensors",
734
+ "model.layers.3.block_sparse_moe.experts.4.w2.weight": "model-00003-of-00019.safetensors",
735
+ "model.layers.3.block_sparse_moe.experts.4.w3.weight": "model-00003-of-00019.safetensors",
736
+ "model.layers.3.block_sparse_moe.experts.5.w1.weight": "model-00003-of-00019.safetensors",
737
+ "model.layers.3.block_sparse_moe.experts.5.w2.weight": "model-00003-of-00019.safetensors",
738
+ "model.layers.3.block_sparse_moe.experts.5.w3.weight": "model-00003-of-00019.safetensors",
739
+ "model.layers.3.block_sparse_moe.experts.6.w1.weight": "model-00003-of-00019.safetensors",
740
+ "model.layers.3.block_sparse_moe.experts.6.w2.weight": "model-00003-of-00019.safetensors",
741
+ "model.layers.3.block_sparse_moe.experts.6.w3.weight": "model-00003-of-00019.safetensors",
742
+ "model.layers.3.block_sparse_moe.experts.7.w1.weight": "model-00003-of-00019.safetensors",
743
+ "model.layers.3.block_sparse_moe.experts.7.w2.weight": "model-00003-of-00019.safetensors",
744
+ "model.layers.3.block_sparse_moe.experts.7.w3.weight": "model-00003-of-00019.safetensors",
745
+ "model.layers.3.block_sparse_moe.gate.weight": "model-00002-of-00019.safetensors",
746
+ "model.layers.3.input_layernorm.weight": "model-00003-of-00019.safetensors",
747
+ "model.layers.3.post_attention_layernorm.weight": "model-00003-of-00019.safetensors",
748
+ "model.layers.3.self_attn.k_proj.weight": "model-00002-of-00019.safetensors",
749
+ "model.layers.3.self_attn.o_proj.weight": "model-00002-of-00019.safetensors",
750
+ "model.layers.3.self_attn.q_proj.weight": "model-00002-of-00019.safetensors",
751
+ "model.layers.3.self_attn.v_proj.weight": "model-00002-of-00019.safetensors",
752
+ "model.layers.30.block_sparse_moe.experts.0.w1.weight": "model-00018-of-00019.safetensors",
753
+ "model.layers.30.block_sparse_moe.experts.0.w2.weight": "model-00018-of-00019.safetensors",
754
+ "model.layers.30.block_sparse_moe.experts.0.w3.weight": "model-00018-of-00019.safetensors",
755
+ "model.layers.30.block_sparse_moe.experts.1.w1.weight": "model-00018-of-00019.safetensors",
756
+ "model.layers.30.block_sparse_moe.experts.1.w2.weight": "model-00018-of-00019.safetensors",
757
+ "model.layers.30.block_sparse_moe.experts.1.w3.weight": "model-00018-of-00019.safetensors",
758
+ "model.layers.30.block_sparse_moe.experts.2.w1.weight": "model-00018-of-00019.safetensors",
759
+ "model.layers.30.block_sparse_moe.experts.2.w2.weight": "model-00018-of-00019.safetensors",
760
+ "model.layers.30.block_sparse_moe.experts.2.w3.weight": "model-00018-of-00019.safetensors",
761
+ "model.layers.30.block_sparse_moe.experts.3.w1.weight": "model-00018-of-00019.safetensors",
762
+ "model.layers.30.block_sparse_moe.experts.3.w2.weight": "model-00018-of-00019.safetensors",
763
+ "model.layers.30.block_sparse_moe.experts.3.w3.weight": "model-00018-of-00019.safetensors",
764
+ "model.layers.30.block_sparse_moe.experts.4.w1.weight": "model-00018-of-00019.safetensors",
765
+ "model.layers.30.block_sparse_moe.experts.4.w2.weight": "model-00018-of-00019.safetensors",
766
+ "model.layers.30.block_sparse_moe.experts.4.w3.weight": "model-00018-of-00019.safetensors",
767
+ "model.layers.30.block_sparse_moe.experts.5.w1.weight": "model-00019-of-00019.safetensors",
768
+ "model.layers.30.block_sparse_moe.experts.5.w2.weight": "model-00019-of-00019.safetensors",
769
+ "model.layers.30.block_sparse_moe.experts.5.w3.weight": "model-00019-of-00019.safetensors",
770
+ "model.layers.30.block_sparse_moe.experts.6.w1.weight": "model-00019-of-00019.safetensors",
771
+ "model.layers.30.block_sparse_moe.experts.6.w2.weight": "model-00019-of-00019.safetensors",
772
+ "model.layers.30.block_sparse_moe.experts.6.w3.weight": "model-00019-of-00019.safetensors",
773
+ "model.layers.30.block_sparse_moe.experts.7.w1.weight": "model-00019-of-00019.safetensors",
774
+ "model.layers.30.block_sparse_moe.experts.7.w2.weight": "model-00019-of-00019.safetensors",
775
+ "model.layers.30.block_sparse_moe.experts.7.w3.weight": "model-00019-of-00019.safetensors",
776
+ "model.layers.30.block_sparse_moe.gate.weight": "model-00018-of-00019.safetensors",
777
+ "model.layers.30.input_layernorm.weight": "model-00019-of-00019.safetensors",
778
+ "model.layers.30.post_attention_layernorm.weight": "model-00019-of-00019.safetensors",
779
+ "model.layers.30.self_attn.k_proj.weight": "model-00018-of-00019.safetensors",
780
+ "model.layers.30.self_attn.o_proj.weight": "model-00018-of-00019.safetensors",
781
+ "model.layers.30.self_attn.q_proj.weight": "model-00018-of-00019.safetensors",
782
+ "model.layers.30.self_attn.v_proj.weight": "model-00018-of-00019.safetensors",
783
+ "model.layers.31.block_sparse_moe.experts.0.w1.weight": "model-00019-of-00019.safetensors",
784
+ "model.layers.31.block_sparse_moe.experts.0.w2.weight": "model-00019-of-00019.safetensors",
785
+ "model.layers.31.block_sparse_moe.experts.0.w3.weight": "model-00019-of-00019.safetensors",
786
+ "model.layers.31.block_sparse_moe.experts.1.w1.weight": "model-00019-of-00019.safetensors",
787
+ "model.layers.31.block_sparse_moe.experts.1.w2.weight": "model-00019-of-00019.safetensors",
788
+ "model.layers.31.block_sparse_moe.experts.1.w3.weight": "model-00019-of-00019.safetensors",
789
+ "model.layers.31.block_sparse_moe.experts.2.w1.weight": "model-00019-of-00019.safetensors",
790
+ "model.layers.31.block_sparse_moe.experts.2.w2.weight": "model-00019-of-00019.safetensors",
791
+ "model.layers.31.block_sparse_moe.experts.2.w3.weight": "model-00019-of-00019.safetensors",
792
+ "model.layers.31.block_sparse_moe.experts.3.w1.weight": "model-00019-of-00019.safetensors",
793
+ "model.layers.31.block_sparse_moe.experts.3.w2.weight": "model-00019-of-00019.safetensors",
794
+ "model.layers.31.block_sparse_moe.experts.3.w3.weight": "model-00019-of-00019.safetensors",
795
+ "model.layers.31.block_sparse_moe.experts.4.w1.weight": "model-00019-of-00019.safetensors",
796
+ "model.layers.31.block_sparse_moe.experts.4.w2.weight": "model-00019-of-00019.safetensors",
797
+ "model.layers.31.block_sparse_moe.experts.4.w3.weight": "model-00019-of-00019.safetensors",
798
+ "model.layers.31.block_sparse_moe.experts.5.w1.weight": "model-00019-of-00019.safetensors",
799
+ "model.layers.31.block_sparse_moe.experts.5.w2.weight": "model-00019-of-00019.safetensors",
800
+ "model.layers.31.block_sparse_moe.experts.5.w3.weight": "model-00019-of-00019.safetensors",
801
+ "model.layers.31.block_sparse_moe.experts.6.w1.weight": "model-00019-of-00019.safetensors",
802
+ "model.layers.31.block_sparse_moe.experts.6.w2.weight": "model-00019-of-00019.safetensors",
803
+ "model.layers.31.block_sparse_moe.experts.6.w3.weight": "model-00019-of-00019.safetensors",
804
+ "model.layers.31.block_sparse_moe.experts.7.w1.weight": "model-00019-of-00019.safetensors",
805
+ "model.layers.31.block_sparse_moe.experts.7.w2.weight": "model-00019-of-00019.safetensors",
806
+ "model.layers.31.block_sparse_moe.experts.7.w3.weight": "model-00019-of-00019.safetensors",
807
+ "model.layers.31.block_sparse_moe.gate.weight": "model-00019-of-00019.safetensors",
808
+ "model.layers.31.input_layernorm.weight": "model-00019-of-00019.safetensors",
809
+ "model.layers.31.post_attention_layernorm.weight": "model-00019-of-00019.safetensors",
810
+ "model.layers.31.self_attn.k_proj.weight": "model-00019-of-00019.safetensors",
811
+ "model.layers.31.self_attn.o_proj.weight": "model-00019-of-00019.safetensors",
812
+ "model.layers.31.self_attn.q_proj.weight": "model-00019-of-00019.safetensors",
813
+ "model.layers.31.self_attn.v_proj.weight": "model-00019-of-00019.safetensors",
814
+ "model.layers.4.block_sparse_moe.experts.0.w1.weight": "model-00003-of-00019.safetensors",
815
+ "model.layers.4.block_sparse_moe.experts.0.w2.weight": "model-00003-of-00019.safetensors",
816
+ "model.layers.4.block_sparse_moe.experts.0.w3.weight": "model-00003-of-00019.safetensors",
817
+ "model.layers.4.block_sparse_moe.experts.1.w1.weight": "model-00003-of-00019.safetensors",
818
+ "model.layers.4.block_sparse_moe.experts.1.w2.weight": "model-00003-of-00019.safetensors",
819
+ "model.layers.4.block_sparse_moe.experts.1.w3.weight": "model-00003-of-00019.safetensors",
820
+ "model.layers.4.block_sparse_moe.experts.2.w1.weight": "model-00003-of-00019.safetensors",
821
+ "model.layers.4.block_sparse_moe.experts.2.w2.weight": "model-00003-of-00019.safetensors",
822
+ "model.layers.4.block_sparse_moe.experts.2.w3.weight": "model-00003-of-00019.safetensors",
823
+ "model.layers.4.block_sparse_moe.experts.3.w1.weight": "model-00003-of-00019.safetensors",
824
+ "model.layers.4.block_sparse_moe.experts.3.w2.weight": "model-00003-of-00019.safetensors",
825
+ "model.layers.4.block_sparse_moe.experts.3.w3.weight": "model-00003-of-00019.safetensors",
826
+ "model.layers.4.block_sparse_moe.experts.4.w1.weight": "model-00003-of-00019.safetensors",
827
+ "model.layers.4.block_sparse_moe.experts.4.w2.weight": "model-00003-of-00019.safetensors",
828
+ "model.layers.4.block_sparse_moe.experts.4.w3.weight": "model-00003-of-00019.safetensors",
829
+ "model.layers.4.block_sparse_moe.experts.5.w1.weight": "model-00003-of-00019.safetensors",
830
+ "model.layers.4.block_sparse_moe.experts.5.w2.weight": "model-00003-of-00019.safetensors",
831
+ "model.layers.4.block_sparse_moe.experts.5.w3.weight": "model-00003-of-00019.safetensors",
832
+ "model.layers.4.block_sparse_moe.experts.6.w1.weight": "model-00003-of-00019.safetensors",
833
+ "model.layers.4.block_sparse_moe.experts.6.w2.weight": "model-00003-of-00019.safetensors",
834
+ "model.layers.4.block_sparse_moe.experts.6.w3.weight": "model-00003-of-00019.safetensors",
835
+ "model.layers.4.block_sparse_moe.experts.7.w1.weight": "model-00003-of-00019.safetensors",
836
+ "model.layers.4.block_sparse_moe.experts.7.w2.weight": "model-00003-of-00019.safetensors",
837
+ "model.layers.4.block_sparse_moe.experts.7.w3.weight": "model-00003-of-00019.safetensors",
838
+ "model.layers.4.block_sparse_moe.gate.weight": "model-00003-of-00019.safetensors",
839
+ "model.layers.4.input_layernorm.weight": "model-00003-of-00019.safetensors",
840
+ "model.layers.4.post_attention_layernorm.weight": "model-00003-of-00019.safetensors",
841
+ "model.layers.4.self_attn.k_proj.weight": "model-00003-of-00019.safetensors",
842
+ "model.layers.4.self_attn.o_proj.weight": "model-00003-of-00019.safetensors",
843
+ "model.layers.4.self_attn.q_proj.weight": "model-00003-of-00019.safetensors",
844
+ "model.layers.4.self_attn.v_proj.weight": "model-00003-of-00019.safetensors",
845
+ "model.layers.5.block_sparse_moe.experts.0.w1.weight": "model-00004-of-00019.safetensors",
846
+ "model.layers.5.block_sparse_moe.experts.0.w2.weight": "model-00004-of-00019.safetensors",
847
+ "model.layers.5.block_sparse_moe.experts.0.w3.weight": "model-00004-of-00019.safetensors",
848
+ "model.layers.5.block_sparse_moe.experts.1.w1.weight": "model-00004-of-00019.safetensors",
849
+ "model.layers.5.block_sparse_moe.experts.1.w2.weight": "model-00004-of-00019.safetensors",
850
+ "model.layers.5.block_sparse_moe.experts.1.w3.weight": "model-00004-of-00019.safetensors",
851
+ "model.layers.5.block_sparse_moe.experts.2.w1.weight": "model-00004-of-00019.safetensors",
852
+ "model.layers.5.block_sparse_moe.experts.2.w2.weight": "model-00004-of-00019.safetensors",
853
+ "model.layers.5.block_sparse_moe.experts.2.w3.weight": "model-00004-of-00019.safetensors",
854
+ "model.layers.5.block_sparse_moe.experts.3.w1.weight": "model-00004-of-00019.safetensors",
855
+ "model.layers.5.block_sparse_moe.experts.3.w2.weight": "model-00004-of-00019.safetensors",
856
+ "model.layers.5.block_sparse_moe.experts.3.w3.weight": "model-00004-of-00019.safetensors",
857
+ "model.layers.5.block_sparse_moe.experts.4.w1.weight": "model-00004-of-00019.safetensors",
858
+ "model.layers.5.block_sparse_moe.experts.4.w2.weight": "model-00004-of-00019.safetensors",
859
+ "model.layers.5.block_sparse_moe.experts.4.w3.weight": "model-00004-of-00019.safetensors",
860
+ "model.layers.5.block_sparse_moe.experts.5.w1.weight": "model-00004-of-00019.safetensors",
861
+ "model.layers.5.block_sparse_moe.experts.5.w2.weight": "model-00004-of-00019.safetensors",
862
+ "model.layers.5.block_sparse_moe.experts.5.w3.weight": "model-00004-of-00019.safetensors",
863
+ "model.layers.5.block_sparse_moe.experts.6.w1.weight": "model-00004-of-00019.safetensors",
864
+ "model.layers.5.block_sparse_moe.experts.6.w2.weight": "model-00004-of-00019.safetensors",
865
+ "model.layers.5.block_sparse_moe.experts.6.w3.weight": "model-00004-of-00019.safetensors",
866
+ "model.layers.5.block_sparse_moe.experts.7.w1.weight": "model-00004-of-00019.safetensors",
867
+ "model.layers.5.block_sparse_moe.experts.7.w2.weight": "model-00004-of-00019.safetensors",
868
+ "model.layers.5.block_sparse_moe.experts.7.w3.weight": "model-00004-of-00019.safetensors",
869
+ "model.layers.5.block_sparse_moe.gate.weight": "model-00003-of-00019.safetensors",
870
+ "model.layers.5.input_layernorm.weight": "model-00004-of-00019.safetensors",
871
+ "model.layers.5.post_attention_layernorm.weight": "model-00004-of-00019.safetensors",
872
+ "model.layers.5.self_attn.k_proj.weight": "model-00003-of-00019.safetensors",
873
+ "model.layers.5.self_attn.o_proj.weight": "model-00003-of-00019.safetensors",
874
+ "model.layers.5.self_attn.q_proj.weight": "model-00003-of-00019.safetensors",
875
+ "model.layers.5.self_attn.v_proj.weight": "model-00003-of-00019.safetensors",
876
+ "model.layers.6.block_sparse_moe.experts.0.w1.weight": "model-00004-of-00019.safetensors",
877
+ "model.layers.6.block_sparse_moe.experts.0.w2.weight": "model-00004-of-00019.safetensors",
878
+ "model.layers.6.block_sparse_moe.experts.0.w3.weight": "model-00004-of-00019.safetensors",
879
+ "model.layers.6.block_sparse_moe.experts.1.w1.weight": "model-00004-of-00019.safetensors",
880
+ "model.layers.6.block_sparse_moe.experts.1.w2.weight": "model-00004-of-00019.safetensors",
881
+ "model.layers.6.block_sparse_moe.experts.1.w3.weight": "model-00004-of-00019.safetensors",
882
+ "model.layers.6.block_sparse_moe.experts.2.w1.weight": "model-00004-of-00019.safetensors",
883
+ "model.layers.6.block_sparse_moe.experts.2.w2.weight": "model-00004-of-00019.safetensors",
884
+ "model.layers.6.block_sparse_moe.experts.2.w3.weight": "model-00004-of-00019.safetensors",
885
+ "model.layers.6.block_sparse_moe.experts.3.w1.weight": "model-00004-of-00019.safetensors",
886
+ "model.layers.6.block_sparse_moe.experts.3.w2.weight": "model-00004-of-00019.safetensors",
887
+ "model.layers.6.block_sparse_moe.experts.3.w3.weight": "model-00004-of-00019.safetensors",
888
+ "model.layers.6.block_sparse_moe.experts.4.w1.weight": "model-00004-of-00019.safetensors",
889
+ "model.layers.6.block_sparse_moe.experts.4.w2.weight": "model-00004-of-00019.safetensors",
890
+ "model.layers.6.block_sparse_moe.experts.4.w3.weight": "model-00004-of-00019.safetensors",
891
+ "model.layers.6.block_sparse_moe.experts.5.w1.weight": "model-00004-of-00019.safetensors",
892
+ "model.layers.6.block_sparse_moe.experts.5.w2.weight": "model-00004-of-00019.safetensors",
893
+ "model.layers.6.block_sparse_moe.experts.5.w3.weight": "model-00005-of-00019.safetensors",
894
+ "model.layers.6.block_sparse_moe.experts.6.w1.weight": "model-00005-of-00019.safetensors",
895
+ "model.layers.6.block_sparse_moe.experts.6.w2.weight": "model-00005-of-00019.safetensors",
896
+ "model.layers.6.block_sparse_moe.experts.6.w3.weight": "model-00005-of-00019.safetensors",
897
+ "model.layers.6.block_sparse_moe.experts.7.w1.weight": "model-00005-of-00019.safetensors",
898
+ "model.layers.6.block_sparse_moe.experts.7.w2.weight": "model-00005-of-00019.safetensors",
899
+ "model.layers.6.block_sparse_moe.experts.7.w3.weight": "model-00005-of-00019.safetensors",
900
+ "model.layers.6.block_sparse_moe.gate.weight": "model-00004-of-00019.safetensors",
901
+ "model.layers.6.input_layernorm.weight": "model-00005-of-00019.safetensors",
902
+ "model.layers.6.post_attention_layernorm.weight": "model-00005-of-00019.safetensors",
903
+ "model.layers.6.self_attn.k_proj.weight": "model-00004-of-00019.safetensors",
904
+ "model.layers.6.self_attn.o_proj.weight": "model-00004-of-00019.safetensors",
905
+ "model.layers.6.self_attn.q_proj.weight": "model-00004-of-00019.safetensors",
906
+ "model.layers.6.self_attn.v_proj.weight": "model-00004-of-00019.safetensors",
907
+ "model.layers.7.block_sparse_moe.experts.0.w1.weight": "model-00005-of-00019.safetensors",
908
+ "model.layers.7.block_sparse_moe.experts.0.w2.weight": "model-00005-of-00019.safetensors",
909
+ "model.layers.7.block_sparse_moe.experts.0.w3.weight": "model-00005-of-00019.safetensors",
910
+ "model.layers.7.block_sparse_moe.experts.1.w1.weight": "model-00005-of-00019.safetensors",
911
+ "model.layers.7.block_sparse_moe.experts.1.w2.weight": "model-00005-of-00019.safetensors",
912
+ "model.layers.7.block_sparse_moe.experts.1.w3.weight": "model-00005-of-00019.safetensors",
913
+ "model.layers.7.block_sparse_moe.experts.2.w1.weight": "model-00005-of-00019.safetensors",
914
+ "model.layers.7.block_sparse_moe.experts.2.w2.weight": "model-00005-of-00019.safetensors",
915
+ "model.layers.7.block_sparse_moe.experts.2.w3.weight": "model-00005-of-00019.safetensors",
916
+ "model.layers.7.block_sparse_moe.experts.3.w1.weight": "model-00005-of-00019.safetensors",
917
+ "model.layers.7.block_sparse_moe.experts.3.w2.weight": "model-00005-of-00019.safetensors",
918
+ "model.layers.7.block_sparse_moe.experts.3.w3.weight": "model-00005-of-00019.safetensors",
919
+ "model.layers.7.block_sparse_moe.experts.4.w1.weight": "model-00005-of-00019.safetensors",
920
+ "model.layers.7.block_sparse_moe.experts.4.w2.weight": "model-00005-of-00019.safetensors",
921
+ "model.layers.7.block_sparse_moe.experts.4.w3.weight": "model-00005-of-00019.safetensors",
922
+ "model.layers.7.block_sparse_moe.experts.5.w1.weight": "model-00005-of-00019.safetensors",
923
+ "model.layers.7.block_sparse_moe.experts.5.w2.weight": "model-00005-of-00019.safetensors",
924
+ "model.layers.7.block_sparse_moe.experts.5.w3.weight": "model-00005-of-00019.safetensors",
925
+ "model.layers.7.block_sparse_moe.experts.6.w1.weight": "model-00005-of-00019.safetensors",
926
+ "model.layers.7.block_sparse_moe.experts.6.w2.weight": "model-00005-of-00019.safetensors",
927
+ "model.layers.7.block_sparse_moe.experts.6.w3.weight": "model-00005-of-00019.safetensors",
928
+ "model.layers.7.block_sparse_moe.experts.7.w1.weight": "model-00005-of-00019.safetensors",
929
+ "model.layers.7.block_sparse_moe.experts.7.w2.weight": "model-00005-of-00019.safetensors",
930
+ "model.layers.7.block_sparse_moe.experts.7.w3.weight": "model-00005-of-00019.safetensors",
931
+ "model.layers.7.block_sparse_moe.gate.weight": "model-00005-of-00019.safetensors",
932
+ "model.layers.7.input_layernorm.weight": "model-00005-of-00019.safetensors",
933
+ "model.layers.7.post_attention_layernorm.weight": "model-00005-of-00019.safetensors",
934
+ "model.layers.7.self_attn.k_proj.weight": "model-00005-of-00019.safetensors",
935
+ "model.layers.7.self_attn.o_proj.weight": "model-00005-of-00019.safetensors",
936
+ "model.layers.7.self_attn.q_proj.weight": "model-00005-of-00019.safetensors",
937
+ "model.layers.7.self_attn.v_proj.weight": "model-00005-of-00019.safetensors",
938
+ "model.layers.8.block_sparse_moe.experts.0.w1.weight": "model-00005-of-00019.safetensors",
939
+ "model.layers.8.block_sparse_moe.experts.0.w2.weight": "model-00005-of-00019.safetensors",
940
+ "model.layers.8.block_sparse_moe.experts.0.w3.weight": "model-00005-of-00019.safetensors",
941
+ "model.layers.8.block_sparse_moe.experts.1.w1.weight": "model-00005-of-00019.safetensors",
942
+ "model.layers.8.block_sparse_moe.experts.1.w2.weight": "model-00005-of-00019.safetensors",
943
+ "model.layers.8.block_sparse_moe.experts.1.w3.weight": "model-00005-of-00019.safetensors",
944
+ "model.layers.8.block_sparse_moe.experts.2.w1.weight": "model-00005-of-00019.safetensors",
945
+ "model.layers.8.block_sparse_moe.experts.2.w2.weight": "model-00005-of-00019.safetensors",
946
+ "model.layers.8.block_sparse_moe.experts.2.w3.weight": "model-00005-of-00019.safetensors",
947
+ "model.layers.8.block_sparse_moe.experts.3.w1.weight": "model-00005-of-00019.safetensors",
948
+ "model.layers.8.block_sparse_moe.experts.3.w2.weight": "model-00006-of-00019.safetensors",
949
+ "model.layers.8.block_sparse_moe.experts.3.w3.weight": "model-00006-of-00019.safetensors",
950
+ "model.layers.8.block_sparse_moe.experts.4.w1.weight": "model-00006-of-00019.safetensors",
951
+ "model.layers.8.block_sparse_moe.experts.4.w2.weight": "model-00006-of-00019.safetensors",
952
+ "model.layers.8.block_sparse_moe.experts.4.w3.weight": "model-00006-of-00019.safetensors",
953
+ "model.layers.8.block_sparse_moe.experts.5.w1.weight": "model-00006-of-00019.safetensors",
954
+ "model.layers.8.block_sparse_moe.experts.5.w2.weight": "model-00006-of-00019.safetensors",
955
+ "model.layers.8.block_sparse_moe.experts.5.w3.weight": "model-00006-of-00019.safetensors",
956
+ "model.layers.8.block_sparse_moe.experts.6.w1.weight": "model-00006-of-00019.safetensors",
957
+ "model.layers.8.block_sparse_moe.experts.6.w2.weight": "model-00006-of-00019.safetensors",
958
+ "model.layers.8.block_sparse_moe.experts.6.w3.weight": "model-00006-of-00019.safetensors",
959
+ "model.layers.8.block_sparse_moe.experts.7.w1.weight": "model-00006-of-00019.safetensors",
960
+ "model.layers.8.block_sparse_moe.experts.7.w2.weight": "model-00006-of-00019.safetensors",
961
+ "model.layers.8.block_sparse_moe.experts.7.w3.weight": "model-00006-of-00019.safetensors",
962
+ "model.layers.8.block_sparse_moe.gate.weight": "model-00005-of-00019.safetensors",
963
+ "model.layers.8.input_layernorm.weight": "model-00006-of-00019.safetensors",
964
+ "model.layers.8.post_attention_layernorm.weight": "model-00006-of-00019.safetensors",
965
+ "model.layers.8.self_attn.k_proj.weight": "model-00005-of-00019.safetensors",
966
+ "model.layers.8.self_attn.o_proj.weight": "model-00005-of-00019.safetensors",
967
+ "model.layers.8.self_attn.q_proj.weight": "model-00005-of-00019.safetensors",
968
+ "model.layers.8.self_attn.v_proj.weight": "model-00005-of-00019.safetensors",
969
+ "model.layers.9.block_sparse_moe.experts.0.w1.weight": "model-00006-of-00019.safetensors",
970
+ "model.layers.9.block_sparse_moe.experts.0.w2.weight": "model-00006-of-00019.safetensors",
971
+ "model.layers.9.block_sparse_moe.experts.0.w3.weight": "model-00006-of-00019.safetensors",
972
+ "model.layers.9.block_sparse_moe.experts.1.w1.weight": "model-00006-of-00019.safetensors",
973
+ "model.layers.9.block_sparse_moe.experts.1.w2.weight": "model-00006-of-00019.safetensors",
974
+ "model.layers.9.block_sparse_moe.experts.1.w3.weight": "model-00006-of-00019.safetensors",
975
+ "model.layers.9.block_sparse_moe.experts.2.w1.weight": "model-00006-of-00019.safetensors",
976
+ "model.layers.9.block_sparse_moe.experts.2.w2.weight": "model-00006-of-00019.safetensors",
977
+ "model.layers.9.block_sparse_moe.experts.2.w3.weight": "model-00006-of-00019.safetensors",
978
+ "model.layers.9.block_sparse_moe.experts.3.w1.weight": "model-00006-of-00019.safetensors",
979
+ "model.layers.9.block_sparse_moe.experts.3.w2.weight": "model-00006-of-00019.safetensors",
980
+ "model.layers.9.block_sparse_moe.experts.3.w3.weight": "model-00006-of-00019.safetensors",
981
+ "model.layers.9.block_sparse_moe.experts.4.w1.weight": "model-00006-of-00019.safetensors",
982
+ "model.layers.9.block_sparse_moe.experts.4.w2.weight": "model-00006-of-00019.safetensors",
983
+ "model.layers.9.block_sparse_moe.experts.4.w3.weight": "model-00006-of-00019.safetensors",
984
+ "model.layers.9.block_sparse_moe.experts.5.w1.weight": "model-00006-of-00019.safetensors",
985
+ "model.layers.9.block_sparse_moe.experts.5.w2.weight": "model-00006-of-00019.safetensors",
986
+ "model.layers.9.block_sparse_moe.experts.5.w3.weight": "model-00006-of-00019.safetensors",
987
+ "model.layers.9.block_sparse_moe.experts.6.w1.weight": "model-00006-of-00019.safetensors",
988
+ "model.layers.9.block_sparse_moe.experts.6.w2.weight": "model-00006-of-00019.safetensors",
989
+ "model.layers.9.block_sparse_moe.experts.6.w3.weight": "model-00006-of-00019.safetensors",
990
+ "model.layers.9.block_sparse_moe.experts.7.w1.weight": "model-00006-of-00019.safetensors",
991
+ "model.layers.9.block_sparse_moe.experts.7.w2.weight": "model-00006-of-00019.safetensors",
992
+ "model.layers.9.block_sparse_moe.experts.7.w3.weight": "model-00006-of-00019.safetensors",
993
+ "model.layers.9.block_sparse_moe.gate.weight": "model-00006-of-00019.safetensors",
994
+ "model.layers.9.input_layernorm.weight": "model-00006-of-00019.safetensors",
995
+ "model.layers.9.post_attention_layernorm.weight": "model-00006-of-00019.safetensors",
996
+ "model.layers.9.self_attn.k_proj.weight": "model-00006-of-00019.safetensors",
997
+ "model.layers.9.self_attn.o_proj.weight": "model-00006-of-00019.safetensors",
998
+ "model.layers.9.self_attn.q_proj.weight": "model-00006-of-00019.safetensors",
999
+ "model.layers.9.self_attn.v_proj.weight": "model-00006-of-00019.safetensors",
1000
+ "model.norm.weight": "model-00019-of-00019.safetensors"
1001
+ }
1002
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "</s>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
tokenizer_config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "additional_special_tokens": [],
32
+ "bos_token": "<s>",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "legacy": true,
36
+ "model_max_length": 1000000000000000019884624838656,
37
+ "pad_token": "</s>",
38
+ "sp_model_kwargs": {},
39
+ "spaces_between_special_tokens": false,
40
+ "tokenizer_class": "LlamaTokenizer",
41
+ "unk_token": "<unk>",
42
+ "use_default_system_prompt": false,
43
+ "use_fast": true
44
+ }
zero_to_fp32.py ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
215
+ exclude_frozen_parameters)
216
+ elif zero_stage == 3:
217
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
218
+ exclude_frozen_parameters)
219
+
220
+
221
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
222
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
223
+ return
224
+
225
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
226
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
227
+
228
+ if debug:
229
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
230
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
231
+
232
+ wanted_params = len(frozen_param_shapes)
233
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
234
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
235
+ print(f'Frozen params: Have {avail_numel} numels to process.')
236
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
237
+
238
+ total_params = 0
239
+ total_numel = 0
240
+ for name, shape in frozen_param_shapes.items():
241
+ total_params += 1
242
+ unpartitioned_numel = shape.numel()
243
+ total_numel += unpartitioned_numel
244
+
245
+ state_dict[name] = frozen_param_fragments[name]
246
+
247
+ if debug:
248
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
249
+
250
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
251
+
252
+
253
+ def _has_callable(obj, fn):
254
+ attr = getattr(obj, fn, None)
255
+ return callable(attr)
256
+
257
+
258
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
259
+ param_shapes = zero_model_states[0].param_shapes
260
+
261
+ # Reconstruction protocol:
262
+ #
263
+ # XXX: document this
264
+
265
+ if debug:
266
+ for i in range(world_size):
267
+ for j in range(len(fp32_flat_groups[0])):
268
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
269
+
270
+ # XXX: memory usage doubles here (zero2)
271
+ num_param_groups = len(fp32_flat_groups[0])
272
+ merged_single_partition_of_fp32_groups = []
273
+ for i in range(num_param_groups):
274
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
275
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
276
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
277
+ avail_numel = sum(
278
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
279
+
280
+ if debug:
281
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
282
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
283
+ # not asserting if there is a mismatch due to possible padding
284
+ print(f"Have {avail_numel} numels to process.")
285
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
286
+
287
+ # params
288
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
289
+ # out-of-core computing solution
290
+ total_numel = 0
291
+ total_params = 0
292
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
293
+ offset = 0
294
+ avail_numel = full_single_fp32_vector.numel()
295
+ for name, shape in shapes.items():
296
+
297
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
298
+ total_numel += unpartitioned_numel
299
+ total_params += 1
300
+
301
+ if debug:
302
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
303
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
304
+ offset += unpartitioned_numel
305
+
306
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
307
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
308
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
309
+ # live optimizer object, so we are checking that the numbers are within the right range
310
+ align_to = 2 * world_size
311
+
312
+ def zero2_align(x):
313
+ return align_to * math.ceil(x / align_to)
314
+
315
+ if debug:
316
+ print(f"original offset={offset}, avail_numel={avail_numel}")
317
+
318
+ offset = zero2_align(offset)
319
+ avail_numel = zero2_align(avail_numel)
320
+
321
+ if debug:
322
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
323
+
324
+ # Sanity check
325
+ if offset != avail_numel:
326
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
327
+
328
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
329
+
330
+
331
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
332
+ exclude_frozen_parameters):
333
+ state_dict = OrderedDict()
334
+
335
+ # buffers
336
+ buffers = zero_model_states[0].buffers
337
+ state_dict.update(buffers)
338
+ if debug:
339
+ print(f"added {len(buffers)} buffers")
340
+
341
+ if not exclude_frozen_parameters:
342
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
343
+
344
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
345
+
346
+ # recover shared parameters
347
+ for pair in zero_model_states[0].shared_params:
348
+ if pair[1] in state_dict:
349
+ state_dict[pair[0]] = state_dict[pair[1]]
350
+
351
+ return state_dict
352
+
353
+
354
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
355
+ remainder = unpartitioned_numel % world_size
356
+ padding_numel = (world_size - remainder) if remainder else 0
357
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
358
+ return partitioned_numel, padding_numel
359
+
360
+
361
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
362
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
363
+ return
364
+
365
+ if debug:
366
+ for i in range(world_size):
367
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
368
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
369
+
370
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
371
+ wanted_params = len(frozen_param_shapes)
372
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
373
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
374
+ print(f'Frozen params: Have {avail_numel} numels to process.')
375
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
376
+
377
+ total_params = 0
378
+ total_numel = 0
379
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
380
+ total_params += 1
381
+ unpartitioned_numel = shape.numel()
382
+ total_numel += unpartitioned_numel
383
+
384
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
385
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
386
+
387
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
388
+
389
+ if debug:
390
+ print(
391
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
392
+ )
393
+
394
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
395
+
396
+
397
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
398
+ param_shapes = zero_model_states[0].param_shapes
399
+ avail_numel = fp32_flat_groups[0].numel() * world_size
400
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
401
+ # param, re-consolidating each param, while dealing with padding if any
402
+
403
+ # merge list of dicts, preserving order
404
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
405
+
406
+ if debug:
407
+ for i in range(world_size):
408
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
409
+
410
+ wanted_params = len(param_shapes)
411
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
412
+ # not asserting if there is a mismatch due to possible padding
413
+ avail_numel = fp32_flat_groups[0].numel() * world_size
414
+ print(f"Trainable params: Have {avail_numel} numels to process.")
415
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
416
+
417
+ # params
418
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
419
+ # out-of-core computing solution
420
+ offset = 0
421
+ total_numel = 0
422
+ total_params = 0
423
+ for name, shape in param_shapes.items():
424
+
425
+ unpartitioned_numel = shape.numel()
426
+ total_numel += unpartitioned_numel
427
+ total_params += 1
428
+
429
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
430
+
431
+ if debug:
432
+ print(
433
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
434
+ )
435
+
436
+ # XXX: memory usage doubles here
437
+ state_dict[name] = torch.cat(
438
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
439
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
440
+ offset += partitioned_numel
441
+
442
+ offset *= world_size
443
+
444
+ # Sanity check
445
+ if offset != avail_numel:
446
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
447
+
448
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
449
+
450
+
451
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
452
+ exclude_frozen_parameters):
453
+ state_dict = OrderedDict()
454
+
455
+ # buffers
456
+ buffers = zero_model_states[0].buffers
457
+ state_dict.update(buffers)
458
+ if debug:
459
+ print(f"added {len(buffers)} buffers")
460
+
461
+ if not exclude_frozen_parameters:
462
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
463
+
464
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
465
+
466
+ # recover shared parameters
467
+ for pair in zero_model_states[0].shared_params:
468
+ if pair[1] in state_dict:
469
+ state_dict[pair[0]] = state_dict[pair[1]]
470
+
471
+ return state_dict
472
+
473
+
474
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
475
+ """
476
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
477
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
478
+ via a model hub.
479
+
480
+ Args:
481
+ - ``checkpoint_dir``: path to the desired checkpoint folder
482
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
483
+ - ``exclude_frozen_parameters``: exclude frozen parameters
484
+
485
+ Returns:
486
+ - pytorch ``state_dict``
487
+
488
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
489
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
490
+ the checkpoint.
491
+
492
+ A typical usage might be ::
493
+
494
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
495
+ # do the training and checkpoint saving
496
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
497
+ model = model.cpu() # move to cpu
498
+ model.load_state_dict(state_dict)
499
+ # submit to model hub or save the model to share with others
500
+
501
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
502
+ application. i.e. you will need to re-initialize the deepspeed engine, since
503
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
504
+
505
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
506
+
507
+ """
508
+ if tag is None:
509
+ latest_path = os.path.join(checkpoint_dir, 'latest')
510
+ if os.path.isfile(latest_path):
511
+ with open(latest_path, 'r') as fd:
512
+ tag = fd.read().strip()
513
+ else:
514
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
515
+
516
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
517
+
518
+ if not os.path.isdir(ds_checkpoint_dir):
519
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
520
+
521
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
522
+
523
+
524
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
525
+ """
526
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
527
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
528
+
529
+ Args:
530
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
531
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
532
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
533
+ - ``exclude_frozen_parameters``: exclude frozen parameters
534
+ """
535
+
536
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
537
+ print(f"Saving fp32 state dict to {output_file}")
538
+ torch.save(state_dict, output_file)
539
+
540
+
541
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
542
+ """
543
+ 1. Put the provided model to cpu
544
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
545
+ 3. Load it into the provided model
546
+
547
+ Args:
548
+ - ``model``: the model object to update
549
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
550
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
551
+
552
+ Returns:
553
+ - ``model`: modified model
554
+
555
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
556
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
557
+ conveniently placed for you in the checkpoint folder.
558
+
559
+ A typical usage might be ::
560
+
561
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
562
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
563
+ # submit to model hub or save the model to share with others
564
+
565
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
566
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
567
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
568
+
569
+ """
570
+ logger.info(f"Extracting fp32 weights")
571
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
572
+
573
+ logger.info(f"Overwriting model with fp32 weights")
574
+ model = model.cpu()
575
+ model.load_state_dict(state_dict, strict=False)
576
+
577
+ return model
578
+
579
+
580
+ if __name__ == "__main__":
581
+
582
+ parser = argparse.ArgumentParser()
583
+ parser.add_argument("checkpoint_dir",
584
+ type=str,
585
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
586
+ parser.add_argument(
587
+ "output_file",
588
+ type=str,
589
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
590
+ parser.add_argument("-t",
591
+ "--tag",
592
+ type=str,
593
+ default=None,
594
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
595
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
596
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
597
+ args = parser.parse_args()
598
+
599
+ debug = args.debug
600
+
601
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
602
+ args.output_file,
603
+ tag=args.tag,
604
+ exclude_frozen_parameters=args.exclude_frozen_parameters)