ehartford commited on
Commit
7617837
1 Parent(s): f970491

Upload folder using huggingface_hub

Browse files
README.md CHANGED
@@ -1,3 +1,16 @@
1
  ---
2
  license: other
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: other
3
+ datasets:
4
+ - ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered
5
+ tags:
6
+ - uncensored
7
  ---
8
+ This is WizardLM trained with a subset of the dataset - responses that contained alignment / moralizing were removed. The intent is to train a WizardLM that doesn't have alignment built-in, so that alignment (of any sort) can be added separately with for example with a RLHF LoRA.
9
+
10
+ Shout out to the open source AI/ML community, and everyone who helped me out.
11
+
12
+ Note:
13
+ An uncensored model has no guardrails.
14
+ You are responsible for anything you do with the model, just as you are responsible for anything you do with any dangerous object such as a knife, gun, lighter, or car.
15
+ Publishing anything this model generates is the same as publishing it yourself.
16
+ You are responsible for the content you publish, and you cannot blame the model any more than you can blame the knife, gun, lighter, or car for what you do with it.
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "[PAD]": 32000
3
+ }
config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/workspace/WizardLM-30B-Uncensored",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 6656,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 17920,
12
+ "max_position_embeddings": 2048,
13
+ "max_sequence_length": 2048,
14
+ "model_type": "llama",
15
+ "num_attention_heads": 52,
16
+ "num_hidden_layers": 60,
17
+ "pad_token_id": 0,
18
+ "rms_norm_eps": 1e-06,
19
+ "tie_word_embeddings": false,
20
+ "torch_dtype": "float16",
21
+ "transformers_version": "4.30.0.dev0",
22
+ "use_cache": true,
23
+ "vocab_size": 32001
24
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "pad_token_id": 0,
6
+ "transformers_version": "4.30.0.dev0"
7
+ }
pytorch_model-00001-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e2f6fb41fab6377f1712be7c378779dfbc87d17d1506e49ec5e241d3edf94576
3
+ size 9818338003
pytorch_model-00002-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0d60e52eac10590d9a1165fbef61fb2ab7af9fc39aff32b5833ae95423e81f1
3
+ size 9958102871
pytorch_model-00003-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abb621a49ad16e90b05a42baed3e9006c29b663380eae1fa087acdeac8654388
3
+ size 9896734715
pytorch_model-00004-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef59d75d3cf5a88ee76fb3778026392a6b5328e0ee8c8c1aeebccb83097f0560
3
+ size 9869470609
pytorch_model-00005-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6218e4e60834ff2050b3ff6551042bb7207bdd4dc86f3fad5db872cee05f837a
3
+ size 9869470573
pytorch_model-00006-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80da71ae47a69a6701b38151c29237a96dcb28b053e53a719122af5a642fd950
3
+ size 9958102871
pytorch_model-00007-of-00007.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e29ab09c8ea4a30b4d14e276268f8358ebe1c8405f70b34fc35e4962c36cbf3
3
+ size 5687916593
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,610 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 65057921536
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "pytorch_model-00007-of-00007.bin",
7
+ "model.embed_tokens.weight": "pytorch_model-00001-of-00007.bin",
8
+ "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
9
+ "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
10
+ "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
11
+ "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
12
+ "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
13
+ "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
14
+ "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
15
+ "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
16
+ "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
17
+ "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
18
+ "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
19
+ "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
20
+ "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
21
+ "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
22
+ "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
23
+ "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
24
+ "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
25
+ "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
26
+ "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
27
+ "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
28
+ "model.layers.10.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
29
+ "model.layers.10.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
30
+ "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
31
+ "model.layers.10.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
32
+ "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
33
+ "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
34
+ "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
35
+ "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
36
+ "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
37
+ "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
38
+ "model.layers.11.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
39
+ "model.layers.11.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
40
+ "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
41
+ "model.layers.11.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
42
+ "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
43
+ "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
44
+ "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
45
+ "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
46
+ "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
47
+ "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
48
+ "model.layers.12.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
49
+ "model.layers.12.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
50
+ "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
51
+ "model.layers.12.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
52
+ "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
53
+ "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
54
+ "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
55
+ "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
56
+ "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
57
+ "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
58
+ "model.layers.13.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
59
+ "model.layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
60
+ "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
61
+ "model.layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
62
+ "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
63
+ "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
64
+ "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
65
+ "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
66
+ "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
67
+ "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
68
+ "model.layers.14.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
69
+ "model.layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
70
+ "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
71
+ "model.layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
72
+ "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
73
+ "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
74
+ "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
75
+ "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
76
+ "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
77
+ "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
78
+ "model.layers.15.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
79
+ "model.layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
80
+ "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
81
+ "model.layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
82
+ "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
83
+ "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
84
+ "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
85
+ "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
86
+ "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
87
+ "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
88
+ "model.layers.16.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
89
+ "model.layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
90
+ "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
91
+ "model.layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
92
+ "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
93
+ "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
94
+ "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
95
+ "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
96
+ "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
97
+ "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
98
+ "model.layers.17.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
99
+ "model.layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
100
+ "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
101
+ "model.layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
102
+ "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
103
+ "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
104
+ "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
105
+ "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
106
+ "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
107
+ "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
108
+ "model.layers.18.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
109
+ "model.layers.18.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
110
+ "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
111
+ "model.layers.18.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
112
+ "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
113
+ "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
114
+ "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
115
+ "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
116
+ "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
117
+ "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
118
+ "model.layers.19.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
119
+ "model.layers.19.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
120
+ "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
121
+ "model.layers.19.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
122
+ "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
123
+ "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
124
+ "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
125
+ "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
126
+ "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
127
+ "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
128
+ "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
129
+ "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
130
+ "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
131
+ "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
132
+ "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
133
+ "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
134
+ "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
135
+ "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
136
+ "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
137
+ "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
138
+ "model.layers.20.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
139
+ "model.layers.20.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
140
+ "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
141
+ "model.layers.20.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
142
+ "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
143
+ "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
144
+ "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
145
+ "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
146
+ "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
147
+ "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
148
+ "model.layers.21.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
149
+ "model.layers.21.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
150
+ "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
151
+ "model.layers.21.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
152
+ "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
153
+ "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
154
+ "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
155
+ "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
156
+ "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
157
+ "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
158
+ "model.layers.22.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
159
+ "model.layers.22.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
160
+ "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
161
+ "model.layers.22.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
162
+ "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
163
+ "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
164
+ "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
165
+ "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
166
+ "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
167
+ "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
168
+ "model.layers.23.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
169
+ "model.layers.23.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
170
+ "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
171
+ "model.layers.23.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
172
+ "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
173
+ "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
174
+ "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
175
+ "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
176
+ "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
177
+ "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
178
+ "model.layers.24.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
179
+ "model.layers.24.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
180
+ "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
181
+ "model.layers.24.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
182
+ "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
183
+ "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
184
+ "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
185
+ "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
186
+ "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
187
+ "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
188
+ "model.layers.25.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
189
+ "model.layers.25.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
190
+ "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
191
+ "model.layers.25.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
192
+ "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
193
+ "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
194
+ "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
195
+ "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
196
+ "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
197
+ "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
198
+ "model.layers.26.input_layernorm.weight": "pytorch_model-00003-of-00007.bin",
199
+ "model.layers.26.mlp.down_proj.weight": "pytorch_model-00003-of-00007.bin",
200
+ "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00003-of-00007.bin",
201
+ "model.layers.26.mlp.up_proj.weight": "pytorch_model-00003-of-00007.bin",
202
+ "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00003-of-00007.bin",
203
+ "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
204
+ "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
205
+ "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
206
+ "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
207
+ "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
208
+ "model.layers.27.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
209
+ "model.layers.27.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
210
+ "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
211
+ "model.layers.27.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
212
+ "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
213
+ "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00003-of-00007.bin",
214
+ "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00003-of-00007.bin",
215
+ "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00003-of-00007.bin",
216
+ "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00007.bin",
217
+ "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00003-of-00007.bin",
218
+ "model.layers.28.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
219
+ "model.layers.28.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
220
+ "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
221
+ "model.layers.28.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
222
+ "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
223
+ "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
224
+ "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
225
+ "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
226
+ "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
227
+ "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
228
+ "model.layers.29.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
229
+ "model.layers.29.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
230
+ "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
231
+ "model.layers.29.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
232
+ "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
233
+ "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
234
+ "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
235
+ "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
236
+ "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
237
+ "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
238
+ "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
239
+ "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
240
+ "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
241
+ "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
242
+ "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
243
+ "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
244
+ "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
245
+ "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
246
+ "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
247
+ "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
248
+ "model.layers.30.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
249
+ "model.layers.30.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
250
+ "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
251
+ "model.layers.30.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
252
+ "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
253
+ "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
254
+ "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
255
+ "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
256
+ "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
257
+ "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
258
+ "model.layers.31.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
259
+ "model.layers.31.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
260
+ "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
261
+ "model.layers.31.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
262
+ "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
263
+ "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
264
+ "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
265
+ "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
266
+ "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
267
+ "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
268
+ "model.layers.32.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
269
+ "model.layers.32.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
270
+ "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
271
+ "model.layers.32.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
272
+ "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
273
+ "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
274
+ "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
275
+ "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
276
+ "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
277
+ "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
278
+ "model.layers.33.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
279
+ "model.layers.33.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
280
+ "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
281
+ "model.layers.33.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
282
+ "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
283
+ "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
284
+ "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
285
+ "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
286
+ "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
287
+ "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
288
+ "model.layers.34.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
289
+ "model.layers.34.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
290
+ "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
291
+ "model.layers.34.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
292
+ "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
293
+ "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
294
+ "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
295
+ "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
296
+ "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
297
+ "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
298
+ "model.layers.35.input_layernorm.weight": "pytorch_model-00004-of-00007.bin",
299
+ "model.layers.35.mlp.down_proj.weight": "pytorch_model-00004-of-00007.bin",
300
+ "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
301
+ "model.layers.35.mlp.up_proj.weight": "pytorch_model-00004-of-00007.bin",
302
+ "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00004-of-00007.bin",
303
+ "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
304
+ "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
305
+ "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
306
+ "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
307
+ "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
308
+ "model.layers.36.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
309
+ "model.layers.36.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
310
+ "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00004-of-00007.bin",
311
+ "model.layers.36.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
312
+ "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
313
+ "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00004-of-00007.bin",
314
+ "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00004-of-00007.bin",
315
+ "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00004-of-00007.bin",
316
+ "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00007.bin",
317
+ "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00004-of-00007.bin",
318
+ "model.layers.37.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
319
+ "model.layers.37.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
320
+ "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
321
+ "model.layers.37.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
322
+ "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
323
+ "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
324
+ "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
325
+ "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
326
+ "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
327
+ "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
328
+ "model.layers.38.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
329
+ "model.layers.38.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
330
+ "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
331
+ "model.layers.38.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
332
+ "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
333
+ "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
334
+ "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
335
+ "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
336
+ "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
337
+ "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
338
+ "model.layers.39.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
339
+ "model.layers.39.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
340
+ "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
341
+ "model.layers.39.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
342
+ "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
343
+ "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
344
+ "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
345
+ "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
346
+ "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
347
+ "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
348
+ "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
349
+ "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
350
+ "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
351
+ "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
352
+ "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
353
+ "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
354
+ "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
355
+ "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
356
+ "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
357
+ "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
358
+ "model.layers.40.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
359
+ "model.layers.40.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
360
+ "model.layers.40.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
361
+ "model.layers.40.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
362
+ "model.layers.40.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
363
+ "model.layers.40.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
364
+ "model.layers.40.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
365
+ "model.layers.40.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
366
+ "model.layers.40.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
367
+ "model.layers.40.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
368
+ "model.layers.41.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
369
+ "model.layers.41.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
370
+ "model.layers.41.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
371
+ "model.layers.41.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
372
+ "model.layers.41.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
373
+ "model.layers.41.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
374
+ "model.layers.41.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
375
+ "model.layers.41.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
376
+ "model.layers.41.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
377
+ "model.layers.41.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
378
+ "model.layers.42.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
379
+ "model.layers.42.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
380
+ "model.layers.42.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
381
+ "model.layers.42.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
382
+ "model.layers.42.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
383
+ "model.layers.42.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
384
+ "model.layers.42.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
385
+ "model.layers.42.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
386
+ "model.layers.42.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
387
+ "model.layers.42.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
388
+ "model.layers.43.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
389
+ "model.layers.43.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
390
+ "model.layers.43.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
391
+ "model.layers.43.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
392
+ "model.layers.43.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
393
+ "model.layers.43.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
394
+ "model.layers.43.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
395
+ "model.layers.43.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
396
+ "model.layers.43.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
397
+ "model.layers.43.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
398
+ "model.layers.44.input_layernorm.weight": "pytorch_model-00005-of-00007.bin",
399
+ "model.layers.44.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
400
+ "model.layers.44.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
401
+ "model.layers.44.mlp.up_proj.weight": "pytorch_model-00005-of-00007.bin",
402
+ "model.layers.44.post_attention_layernorm.weight": "pytorch_model-00005-of-00007.bin",
403
+ "model.layers.44.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
404
+ "model.layers.44.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
405
+ "model.layers.44.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
406
+ "model.layers.44.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
407
+ "model.layers.44.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
408
+ "model.layers.45.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
409
+ "model.layers.45.mlp.down_proj.weight": "pytorch_model-00005-of-00007.bin",
410
+ "model.layers.45.mlp.gate_proj.weight": "pytorch_model-00005-of-00007.bin",
411
+ "model.layers.45.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
412
+ "model.layers.45.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
413
+ "model.layers.45.self_attn.k_proj.weight": "pytorch_model-00005-of-00007.bin",
414
+ "model.layers.45.self_attn.o_proj.weight": "pytorch_model-00005-of-00007.bin",
415
+ "model.layers.45.self_attn.q_proj.weight": "pytorch_model-00005-of-00007.bin",
416
+ "model.layers.45.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00007.bin",
417
+ "model.layers.45.self_attn.v_proj.weight": "pytorch_model-00005-of-00007.bin",
418
+ "model.layers.46.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
419
+ "model.layers.46.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
420
+ "model.layers.46.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
421
+ "model.layers.46.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
422
+ "model.layers.46.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
423
+ "model.layers.46.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
424
+ "model.layers.46.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
425
+ "model.layers.46.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
426
+ "model.layers.46.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
427
+ "model.layers.46.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
428
+ "model.layers.47.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
429
+ "model.layers.47.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
430
+ "model.layers.47.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
431
+ "model.layers.47.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
432
+ "model.layers.47.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
433
+ "model.layers.47.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
434
+ "model.layers.47.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
435
+ "model.layers.47.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
436
+ "model.layers.47.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
437
+ "model.layers.47.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
438
+ "model.layers.48.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
439
+ "model.layers.48.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
440
+ "model.layers.48.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
441
+ "model.layers.48.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
442
+ "model.layers.48.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
443
+ "model.layers.48.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
444
+ "model.layers.48.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
445
+ "model.layers.48.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
446
+ "model.layers.48.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
447
+ "model.layers.48.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
448
+ "model.layers.49.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
449
+ "model.layers.49.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
450
+ "model.layers.49.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
451
+ "model.layers.49.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
452
+ "model.layers.49.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
453
+ "model.layers.49.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
454
+ "model.layers.49.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
455
+ "model.layers.49.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
456
+ "model.layers.49.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
457
+ "model.layers.49.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
458
+ "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
459
+ "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
460
+ "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
461
+ "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
462
+ "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
463
+ "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
464
+ "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
465
+ "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
466
+ "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
467
+ "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
468
+ "model.layers.50.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
469
+ "model.layers.50.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
470
+ "model.layers.50.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
471
+ "model.layers.50.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
472
+ "model.layers.50.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
473
+ "model.layers.50.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
474
+ "model.layers.50.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
475
+ "model.layers.50.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
476
+ "model.layers.50.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
477
+ "model.layers.50.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
478
+ "model.layers.51.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
479
+ "model.layers.51.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
480
+ "model.layers.51.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
481
+ "model.layers.51.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
482
+ "model.layers.51.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
483
+ "model.layers.51.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
484
+ "model.layers.51.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
485
+ "model.layers.51.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
486
+ "model.layers.51.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
487
+ "model.layers.51.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
488
+ "model.layers.52.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
489
+ "model.layers.52.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
490
+ "model.layers.52.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
491
+ "model.layers.52.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
492
+ "model.layers.52.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
493
+ "model.layers.52.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
494
+ "model.layers.52.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
495
+ "model.layers.52.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
496
+ "model.layers.52.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
497
+ "model.layers.52.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
498
+ "model.layers.53.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
499
+ "model.layers.53.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
500
+ "model.layers.53.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
501
+ "model.layers.53.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
502
+ "model.layers.53.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
503
+ "model.layers.53.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
504
+ "model.layers.53.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
505
+ "model.layers.53.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
506
+ "model.layers.53.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
507
+ "model.layers.53.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
508
+ "model.layers.54.input_layernorm.weight": "pytorch_model-00006-of-00007.bin",
509
+ "model.layers.54.mlp.down_proj.weight": "pytorch_model-00006-of-00007.bin",
510
+ "model.layers.54.mlp.gate_proj.weight": "pytorch_model-00006-of-00007.bin",
511
+ "model.layers.54.mlp.up_proj.weight": "pytorch_model-00006-of-00007.bin",
512
+ "model.layers.54.post_attention_layernorm.weight": "pytorch_model-00006-of-00007.bin",
513
+ "model.layers.54.self_attn.k_proj.weight": "pytorch_model-00006-of-00007.bin",
514
+ "model.layers.54.self_attn.o_proj.weight": "pytorch_model-00006-of-00007.bin",
515
+ "model.layers.54.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
516
+ "model.layers.54.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00007.bin",
517
+ "model.layers.54.self_attn.v_proj.weight": "pytorch_model-00006-of-00007.bin",
518
+ "model.layers.55.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
519
+ "model.layers.55.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
520
+ "model.layers.55.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
521
+ "model.layers.55.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
522
+ "model.layers.55.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
523
+ "model.layers.55.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
524
+ "model.layers.55.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
525
+ "model.layers.55.self_attn.q_proj.weight": "pytorch_model-00006-of-00007.bin",
526
+ "model.layers.55.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
527
+ "model.layers.55.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
528
+ "model.layers.56.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
529
+ "model.layers.56.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
530
+ "model.layers.56.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
531
+ "model.layers.56.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
532
+ "model.layers.56.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
533
+ "model.layers.56.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
534
+ "model.layers.56.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
535
+ "model.layers.56.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
536
+ "model.layers.56.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
537
+ "model.layers.56.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
538
+ "model.layers.57.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
539
+ "model.layers.57.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
540
+ "model.layers.57.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
541
+ "model.layers.57.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
542
+ "model.layers.57.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
543
+ "model.layers.57.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
544
+ "model.layers.57.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
545
+ "model.layers.57.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
546
+ "model.layers.57.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
547
+ "model.layers.57.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
548
+ "model.layers.58.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
549
+ "model.layers.58.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
550
+ "model.layers.58.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
551
+ "model.layers.58.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
552
+ "model.layers.58.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
553
+ "model.layers.58.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
554
+ "model.layers.58.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
555
+ "model.layers.58.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
556
+ "model.layers.58.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
557
+ "model.layers.58.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
558
+ "model.layers.59.input_layernorm.weight": "pytorch_model-00007-of-00007.bin",
559
+ "model.layers.59.mlp.down_proj.weight": "pytorch_model-00007-of-00007.bin",
560
+ "model.layers.59.mlp.gate_proj.weight": "pytorch_model-00007-of-00007.bin",
561
+ "model.layers.59.mlp.up_proj.weight": "pytorch_model-00007-of-00007.bin",
562
+ "model.layers.59.post_attention_layernorm.weight": "pytorch_model-00007-of-00007.bin",
563
+ "model.layers.59.self_attn.k_proj.weight": "pytorch_model-00007-of-00007.bin",
564
+ "model.layers.59.self_attn.o_proj.weight": "pytorch_model-00007-of-00007.bin",
565
+ "model.layers.59.self_attn.q_proj.weight": "pytorch_model-00007-of-00007.bin",
566
+ "model.layers.59.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00007.bin",
567
+ "model.layers.59.self_attn.v_proj.weight": "pytorch_model-00007-of-00007.bin",
568
+ "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
569
+ "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
570
+ "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
571
+ "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
572
+ "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
573
+ "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
574
+ "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
575
+ "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
576
+ "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
577
+ "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
578
+ "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00007.bin",
579
+ "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
580
+ "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
581
+ "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00007.bin",
582
+ "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00007.bin",
583
+ "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
584
+ "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
585
+ "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
586
+ "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
587
+ "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
588
+ "model.layers.8.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
589
+ "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00007.bin",
590
+ "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00007.bin",
591
+ "model.layers.8.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
592
+ "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
593
+ "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00007.bin",
594
+ "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00007.bin",
595
+ "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00007.bin",
596
+ "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00007.bin",
597
+ "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00007.bin",
598
+ "model.layers.9.input_layernorm.weight": "pytorch_model-00002-of-00007.bin",
599
+ "model.layers.9.mlp.down_proj.weight": "pytorch_model-00002-of-00007.bin",
600
+ "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00002-of-00007.bin",
601
+ "model.layers.9.mlp.up_proj.weight": "pytorch_model-00002-of-00007.bin",
602
+ "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00002-of-00007.bin",
603
+ "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00002-of-00007.bin",
604
+ "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00002-of-00007.bin",
605
+ "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00002-of-00007.bin",
606
+ "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00007.bin",
607
+ "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00002-of-00007.bin",
608
+ "model.norm.weight": "pytorch_model-00007-of-00007.bin"
609
+ }
610
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "</s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "[PAD]",
5
+ "unk_token": "</s>"
6
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "model_max_length": 2048,
22
+ "pad_token": null,
23
+ "padding_side": "right",
24
+ "sp_model_kwargs": {},
25
+ "tokenizer_class": "LlamaTokenizer",
26
+ "unk_token": {
27
+ "__type": "AddedToken",
28
+ "content": "<unk>",
29
+ "lstrip": false,
30
+ "normalized": true,
31
+ "rstrip": false,
32
+ "single_word": false
33
+ }
34
+ }
trainer_state.json ADDED
@@ -0,0 +1,3874 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.9965075669383,
5
+ "global_step": 1287,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 2e-05,
13
+ "loss": 0.905,
14
+ "step": 2
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 1.9999880457421163e-05,
19
+ "loss": 0.6497,
20
+ "step": 4
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.9999521832542736e-05,
25
+ "loss": 0.6121,
26
+ "step": 6
27
+ },
28
+ {
29
+ "epoch": 0.02,
30
+ "learning_rate": 1.9998924133938902e-05,
31
+ "loss": 0.7236,
32
+ "step": 8
33
+ },
34
+ {
35
+ "epoch": 0.02,
36
+ "learning_rate": 1.9998087375899756e-05,
37
+ "loss": 0.7515,
38
+ "step": 10
39
+ },
40
+ {
41
+ "epoch": 0.03,
42
+ "learning_rate": 1.9997011578430938e-05,
43
+ "loss": 0.7073,
44
+ "step": 12
45
+ },
46
+ {
47
+ "epoch": 0.03,
48
+ "learning_rate": 1.9995696767253165e-05,
49
+ "loss": 0.6146,
50
+ "step": 14
51
+ },
52
+ {
53
+ "epoch": 0.04,
54
+ "learning_rate": 1.9994142973801627e-05,
55
+ "loss": 0.5923,
56
+ "step": 16
57
+ },
58
+ {
59
+ "epoch": 0.04,
60
+ "learning_rate": 1.9992350235225215e-05,
61
+ "loss": 0.5629,
62
+ "step": 18
63
+ },
64
+ {
65
+ "epoch": 0.05,
66
+ "learning_rate": 1.999031859438565e-05,
67
+ "loss": 0.5383,
68
+ "step": 20
69
+ },
70
+ {
71
+ "epoch": 0.05,
72
+ "learning_rate": 1.9988048099856443e-05,
73
+ "loss": 0.516,
74
+ "step": 22
75
+ },
76
+ {
77
+ "epoch": 0.06,
78
+ "learning_rate": 1.9985538805921757e-05,
79
+ "loss": 0.5035,
80
+ "step": 24
81
+ },
82
+ {
83
+ "epoch": 0.06,
84
+ "learning_rate": 1.998279077257508e-05,
85
+ "loss": 0.5244,
86
+ "step": 26
87
+ },
88
+ {
89
+ "epoch": 0.07,
90
+ "learning_rate": 1.9979804065517808e-05,
91
+ "loss": 0.486,
92
+ "step": 28
93
+ },
94
+ {
95
+ "epoch": 0.07,
96
+ "learning_rate": 1.9976578756157684e-05,
97
+ "loss": 0.4945,
98
+ "step": 30
99
+ },
100
+ {
101
+ "epoch": 0.07,
102
+ "learning_rate": 1.9973114921607055e-05,
103
+ "loss": 0.4966,
104
+ "step": 32
105
+ },
106
+ {
107
+ "epoch": 0.08,
108
+ "learning_rate": 1.9969412644681077e-05,
109
+ "loss": 0.4935,
110
+ "step": 34
111
+ },
112
+ {
113
+ "epoch": 0.08,
114
+ "learning_rate": 1.9965472013895685e-05,
115
+ "loss": 0.4739,
116
+ "step": 36
117
+ },
118
+ {
119
+ "epoch": 0.09,
120
+ "learning_rate": 1.996129312346552e-05,
121
+ "loss": 0.4913,
122
+ "step": 38
123
+ },
124
+ {
125
+ "epoch": 0.09,
126
+ "learning_rate": 1.9956876073301645e-05,
127
+ "loss": 0.4641,
128
+ "step": 40
129
+ },
130
+ {
131
+ "epoch": 0.1,
132
+ "learning_rate": 1.9952220969009175e-05,
133
+ "loss": 0.4691,
134
+ "step": 42
135
+ },
136
+ {
137
+ "epoch": 0.1,
138
+ "learning_rate": 1.9947327921884746e-05,
139
+ "loss": 0.4666,
140
+ "step": 44
141
+ },
142
+ {
143
+ "epoch": 0.11,
144
+ "learning_rate": 1.994219704891385e-05,
145
+ "loss": 0.4501,
146
+ "step": 46
147
+ },
148
+ {
149
+ "epoch": 0.11,
150
+ "learning_rate": 1.9936828472768043e-05,
151
+ "loss": 0.4558,
152
+ "step": 48
153
+ },
154
+ {
155
+ "epoch": 0.12,
156
+ "learning_rate": 1.9931222321802016e-05,
157
+ "loss": 0.4712,
158
+ "step": 50
159
+ },
160
+ {
161
+ "epoch": 0.12,
162
+ "learning_rate": 1.9925378730050518e-05,
163
+ "loss": 0.4661,
164
+ "step": 52
165
+ },
166
+ {
167
+ "epoch": 0.13,
168
+ "learning_rate": 1.9919297837225152e-05,
169
+ "loss": 0.4735,
170
+ "step": 54
171
+ },
172
+ {
173
+ "epoch": 0.13,
174
+ "learning_rate": 1.9912979788711042e-05,
175
+ "loss": 0.4526,
176
+ "step": 56
177
+ },
178
+ {
179
+ "epoch": 0.14,
180
+ "learning_rate": 1.990642473556335e-05,
181
+ "loss": 0.4453,
182
+ "step": 58
183
+ },
184
+ {
185
+ "epoch": 0.14,
186
+ "learning_rate": 1.9899632834503662e-05,
187
+ "loss": 0.4713,
188
+ "step": 60
189
+ },
190
+ {
191
+ "epoch": 0.14,
192
+ "learning_rate": 1.989260424791626e-05,
193
+ "loss": 0.4622,
194
+ "step": 62
195
+ },
196
+ {
197
+ "epoch": 0.15,
198
+ "learning_rate": 1.9885339143844217e-05,
199
+ "loss": 0.4585,
200
+ "step": 64
201
+ },
202
+ {
203
+ "epoch": 0.15,
204
+ "learning_rate": 1.987783769598538e-05,
205
+ "loss": 0.4576,
206
+ "step": 66
207
+ },
208
+ {
209
+ "epoch": 0.16,
210
+ "learning_rate": 1.9870100083688242e-05,
211
+ "loss": 0.4353,
212
+ "step": 68
213
+ },
214
+ {
215
+ "epoch": 0.16,
216
+ "learning_rate": 1.9862126491947624e-05,
217
+ "loss": 0.4509,
218
+ "step": 70
219
+ },
220
+ {
221
+ "epoch": 0.17,
222
+ "learning_rate": 1.985391711140027e-05,
223
+ "loss": 0.4402,
224
+ "step": 72
225
+ },
226
+ {
227
+ "epoch": 0.17,
228
+ "learning_rate": 1.9845472138320282e-05,
229
+ "loss": 0.437,
230
+ "step": 74
231
+ },
232
+ {
233
+ "epoch": 0.18,
234
+ "learning_rate": 1.9836791774614437e-05,
235
+ "loss": 0.4613,
236
+ "step": 76
237
+ },
238
+ {
239
+ "epoch": 0.18,
240
+ "learning_rate": 1.982787622781735e-05,
241
+ "loss": 0.4567,
242
+ "step": 78
243
+ },
244
+ {
245
+ "epoch": 0.19,
246
+ "learning_rate": 1.9818725711086506e-05,
247
+ "loss": 0.4541,
248
+ "step": 80
249
+ },
250
+ {
251
+ "epoch": 0.19,
252
+ "learning_rate": 1.980934044319718e-05,
253
+ "loss": 0.4398,
254
+ "step": 82
255
+ },
256
+ {
257
+ "epoch": 0.2,
258
+ "learning_rate": 1.9799720648537197e-05,
259
+ "loss": 0.4283,
260
+ "step": 84
261
+ },
262
+ {
263
+ "epoch": 0.2,
264
+ "learning_rate": 1.978986655710157e-05,
265
+ "loss": 0.4443,
266
+ "step": 86
267
+ },
268
+ {
269
+ "epoch": 0.2,
270
+ "learning_rate": 1.9779778404487e-05,
271
+ "loss": 0.4457,
272
+ "step": 88
273
+ },
274
+ {
275
+ "epoch": 0.21,
276
+ "learning_rate": 1.9769456431886244e-05,
277
+ "loss": 0.4326,
278
+ "step": 90
279
+ },
280
+ {
281
+ "epoch": 0.21,
282
+ "learning_rate": 1.9758900886082343e-05,
283
+ "loss": 0.4557,
284
+ "step": 92
285
+ },
286
+ {
287
+ "epoch": 0.22,
288
+ "learning_rate": 1.9748112019442734e-05,
289
+ "loss": 0.4402,
290
+ "step": 94
291
+ },
292
+ {
293
+ "epoch": 0.22,
294
+ "learning_rate": 1.9737090089913205e-05,
295
+ "loss": 0.465,
296
+ "step": 96
297
+ },
298
+ {
299
+ "epoch": 0.23,
300
+ "learning_rate": 1.9725835361011726e-05,
301
+ "loss": 0.4387,
302
+ "step": 98
303
+ },
304
+ {
305
+ "epoch": 0.23,
306
+ "learning_rate": 1.971434810182217e-05,
307
+ "loss": 0.4479,
308
+ "step": 100
309
+ },
310
+ {
311
+ "epoch": 0.24,
312
+ "learning_rate": 1.9702628586987846e-05,
313
+ "loss": 0.4344,
314
+ "step": 102
315
+ },
316
+ {
317
+ "epoch": 0.24,
318
+ "learning_rate": 1.9690677096704964e-05,
319
+ "loss": 0.4302,
320
+ "step": 104
321
+ },
322
+ {
323
+ "epoch": 0.25,
324
+ "learning_rate": 1.9678493916715914e-05,
325
+ "loss": 0.4331,
326
+ "step": 106
327
+ },
328
+ {
329
+ "epoch": 0.25,
330
+ "learning_rate": 1.966607933830245e-05,
331
+ "loss": 0.4224,
332
+ "step": 108
333
+ },
334
+ {
335
+ "epoch": 0.26,
336
+ "learning_rate": 1.9653433658278717e-05,
337
+ "loss": 0.4225,
338
+ "step": 110
339
+ },
340
+ {
341
+ "epoch": 0.26,
342
+ "learning_rate": 1.9640557178984152e-05,
343
+ "loss": 0.4177,
344
+ "step": 112
345
+ },
346
+ {
347
+ "epoch": 0.27,
348
+ "learning_rate": 1.9627450208276265e-05,
349
+ "loss": 0.4546,
350
+ "step": 114
351
+ },
352
+ {
353
+ "epoch": 0.27,
354
+ "learning_rate": 1.9614113059523273e-05,
355
+ "loss": 0.4257,
356
+ "step": 116
357
+ },
358
+ {
359
+ "epoch": 0.27,
360
+ "learning_rate": 1.9600546051596604e-05,
361
+ "loss": 0.4453,
362
+ "step": 118
363
+ },
364
+ {
365
+ "epoch": 0.28,
366
+ "learning_rate": 1.9586749508863284e-05,
367
+ "loss": 0.458,
368
+ "step": 120
369
+ },
370
+ {
371
+ "epoch": 0.28,
372
+ "learning_rate": 1.9572723761178168e-05,
373
+ "loss": 0.4287,
374
+ "step": 122
375
+ },
376
+ {
377
+ "epoch": 0.29,
378
+ "learning_rate": 1.955846914387607e-05,
379
+ "loss": 0.4581,
380
+ "step": 124
381
+ },
382
+ {
383
+ "epoch": 0.29,
384
+ "learning_rate": 1.954398599776373e-05,
385
+ "loss": 0.4343,
386
+ "step": 126
387
+ },
388
+ {
389
+ "epoch": 0.3,
390
+ "learning_rate": 1.952927466911168e-05,
391
+ "loss": 0.4431,
392
+ "step": 128
393
+ },
394
+ {
395
+ "epoch": 0.3,
396
+ "learning_rate": 1.9514335509645948e-05,
397
+ "loss": 0.4332,
398
+ "step": 130
399
+ },
400
+ {
401
+ "epoch": 0.31,
402
+ "learning_rate": 1.9499168876539666e-05,
403
+ "loss": 0.4315,
404
+ "step": 132
405
+ },
406
+ {
407
+ "epoch": 0.31,
408
+ "learning_rate": 1.9483775132404517e-05,
409
+ "loss": 0.4403,
410
+ "step": 134
411
+ },
412
+ {
413
+ "epoch": 0.32,
414
+ "learning_rate": 1.946815464528208e-05,
415
+ "loss": 0.4618,
416
+ "step": 136
417
+ },
418
+ {
419
+ "epoch": 0.32,
420
+ "learning_rate": 1.9452307788635015e-05,
421
+ "loss": 0.4292,
422
+ "step": 138
423
+ },
424
+ {
425
+ "epoch": 0.33,
426
+ "learning_rate": 1.9436234941338145e-05,
427
+ "loss": 0.4333,
428
+ "step": 140
429
+ },
430
+ {
431
+ "epoch": 0.33,
432
+ "learning_rate": 1.9419936487669396e-05,
433
+ "loss": 0.4557,
434
+ "step": 142
435
+ },
436
+ {
437
+ "epoch": 0.34,
438
+ "learning_rate": 1.94034128173006e-05,
439
+ "loss": 0.4575,
440
+ "step": 144
441
+ },
442
+ {
443
+ "epoch": 0.34,
444
+ "learning_rate": 1.938666432528819e-05,
445
+ "loss": 0.4012,
446
+ "step": 146
447
+ },
448
+ {
449
+ "epoch": 0.34,
450
+ "learning_rate": 1.9369691412063755e-05,
451
+ "loss": 0.4579,
452
+ "step": 148
453
+ },
454
+ {
455
+ "epoch": 0.35,
456
+ "learning_rate": 1.9352494483424456e-05,
457
+ "loss": 0.4337,
458
+ "step": 150
459
+ },
460
+ {
461
+ "epoch": 0.35,
462
+ "learning_rate": 1.9335073950523335e-05,
463
+ "loss": 0.4142,
464
+ "step": 152
465
+ },
466
+ {
467
+ "epoch": 0.36,
468
+ "learning_rate": 1.9317430229859474e-05,
469
+ "loss": 0.4545,
470
+ "step": 154
471
+ },
472
+ {
473
+ "epoch": 0.36,
474
+ "learning_rate": 1.929956374326805e-05,
475
+ "loss": 0.4679,
476
+ "step": 156
477
+ },
478
+ {
479
+ "epoch": 0.37,
480
+ "learning_rate": 1.928147491791024e-05,
481
+ "loss": 0.4178,
482
+ "step": 158
483
+ },
484
+ {
485
+ "epoch": 0.37,
486
+ "learning_rate": 1.9263164186263003e-05,
487
+ "loss": 0.4474,
488
+ "step": 160
489
+ },
490
+ {
491
+ "epoch": 0.38,
492
+ "learning_rate": 1.9244631986108768e-05,
493
+ "loss": 0.4237,
494
+ "step": 162
495
+ },
496
+ {
497
+ "epoch": 0.38,
498
+ "learning_rate": 1.922587876052492e-05,
499
+ "loss": 0.4456,
500
+ "step": 164
501
+ },
502
+ {
503
+ "epoch": 0.39,
504
+ "learning_rate": 1.920690495787326e-05,
505
+ "loss": 0.412,
506
+ "step": 166
507
+ },
508
+ {
509
+ "epoch": 0.39,
510
+ "learning_rate": 1.918771103178924e-05,
511
+ "loss": 0.4279,
512
+ "step": 168
513
+ },
514
+ {
515
+ "epoch": 0.4,
516
+ "learning_rate": 1.916829744117115e-05,
517
+ "loss": 0.413,
518
+ "step": 170
519
+ },
520
+ {
521
+ "epoch": 0.4,
522
+ "learning_rate": 1.9148664650169128e-05,
523
+ "loss": 0.4508,
524
+ "step": 172
525
+ },
526
+ {
527
+ "epoch": 0.41,
528
+ "learning_rate": 1.9128813128174063e-05,
529
+ "loss": 0.4054,
530
+ "step": 174
531
+ },
532
+ {
533
+ "epoch": 0.41,
534
+ "learning_rate": 1.9108743349806382e-05,
535
+ "loss": 0.4021,
536
+ "step": 176
537
+ },
538
+ {
539
+ "epoch": 0.41,
540
+ "learning_rate": 1.90884557949047e-05,
541
+ "loss": 0.4392,
542
+ "step": 178
543
+ },
544
+ {
545
+ "epoch": 0.42,
546
+ "learning_rate": 1.9067950948514343e-05,
547
+ "loss": 0.4414,
548
+ "step": 180
549
+ },
550
+ {
551
+ "epoch": 0.42,
552
+ "learning_rate": 1.904722930087575e-05,
553
+ "loss": 0.4327,
554
+ "step": 182
555
+ },
556
+ {
557
+ "epoch": 0.43,
558
+ "learning_rate": 1.9026291347412765e-05,
559
+ "loss": 0.4081,
560
+ "step": 184
561
+ },
562
+ {
563
+ "epoch": 0.43,
564
+ "learning_rate": 1.900513758872078e-05,
565
+ "loss": 0.4432,
566
+ "step": 186
567
+ },
568
+ {
569
+ "epoch": 0.44,
570
+ "learning_rate": 1.8983768530554765e-05,
571
+ "loss": 0.4355,
572
+ "step": 188
573
+ },
574
+ {
575
+ "epoch": 0.44,
576
+ "learning_rate": 1.8962184683817182e-05,
577
+ "loss": 0.4292,
578
+ "step": 190
579
+ },
580
+ {
581
+ "epoch": 0.45,
582
+ "learning_rate": 1.8940386564545773e-05,
583
+ "loss": 0.4182,
584
+ "step": 192
585
+ },
586
+ {
587
+ "epoch": 0.45,
588
+ "learning_rate": 1.891837469390122e-05,
589
+ "loss": 0.4402,
590
+ "step": 194
591
+ },
592
+ {
593
+ "epoch": 0.46,
594
+ "learning_rate": 1.8896149598154675e-05,
595
+ "loss": 0.4377,
596
+ "step": 196
597
+ },
598
+ {
599
+ "epoch": 0.46,
600
+ "learning_rate": 1.887371180867519e-05,
601
+ "loss": 0.4236,
602
+ "step": 198
603
+ },
604
+ {
605
+ "epoch": 0.47,
606
+ "learning_rate": 1.8851061861917013e-05,
607
+ "loss": 0.4399,
608
+ "step": 200
609
+ },
610
+ {
611
+ "epoch": 0.47,
612
+ "learning_rate": 1.8828200299406747e-05,
613
+ "loss": 0.4285,
614
+ "step": 202
615
+ },
616
+ {
617
+ "epoch": 0.47,
618
+ "learning_rate": 1.8805127667730426e-05,
619
+ "loss": 0.4465,
620
+ "step": 204
621
+ },
622
+ {
623
+ "epoch": 0.48,
624
+ "learning_rate": 1.878184451852042e-05,
625
+ "loss": 0.4264,
626
+ "step": 206
627
+ },
628
+ {
629
+ "epoch": 0.48,
630
+ "learning_rate": 1.8758351408442278e-05,
631
+ "loss": 0.4196,
632
+ "step": 208
633
+ },
634
+ {
635
+ "epoch": 0.49,
636
+ "learning_rate": 1.8734648899181388e-05,
637
+ "loss": 0.4104,
638
+ "step": 210
639
+ },
640
+ {
641
+ "epoch": 0.49,
642
+ "learning_rate": 1.871073755742957e-05,
643
+ "loss": 0.4188,
644
+ "step": 212
645
+ },
646
+ {
647
+ "epoch": 0.5,
648
+ "learning_rate": 1.868661795487151e-05,
649
+ "loss": 0.4418,
650
+ "step": 214
651
+ },
652
+ {
653
+ "epoch": 0.5,
654
+ "learning_rate": 1.8662290668171107e-05,
655
+ "loss": 0.4183,
656
+ "step": 216
657
+ },
658
+ {
659
+ "epoch": 0.51,
660
+ "learning_rate": 1.8637756278957683e-05,
661
+ "loss": 0.4076,
662
+ "step": 218
663
+ },
664
+ {
665
+ "epoch": 0.51,
666
+ "learning_rate": 1.8613015373812066e-05,
667
+ "loss": 0.4105,
668
+ "step": 220
669
+ },
670
+ {
671
+ "epoch": 0.52,
672
+ "learning_rate": 1.8588068544252572e-05,
673
+ "loss": 0.4478,
674
+ "step": 222
675
+ },
676
+ {
677
+ "epoch": 0.52,
678
+ "learning_rate": 1.8562916386720883e-05,
679
+ "loss": 0.4312,
680
+ "step": 224
681
+ },
682
+ {
683
+ "epoch": 0.53,
684
+ "learning_rate": 1.853755950256774e-05,
685
+ "loss": 0.4044,
686
+ "step": 226
687
+ },
688
+ {
689
+ "epoch": 0.53,
690
+ "learning_rate": 1.8511998498038615e-05,
691
+ "loss": 0.4069,
692
+ "step": 228
693
+ },
694
+ {
695
+ "epoch": 0.54,
696
+ "learning_rate": 1.8486233984259186e-05,
697
+ "loss": 0.4349,
698
+ "step": 230
699
+ },
700
+ {
701
+ "epoch": 0.54,
702
+ "learning_rate": 1.8460266577220733e-05,
703
+ "loss": 0.4039,
704
+ "step": 232
705
+ },
706
+ {
707
+ "epoch": 0.54,
708
+ "learning_rate": 1.8434096897765422e-05,
709
+ "loss": 0.4153,
710
+ "step": 234
711
+ },
712
+ {
713
+ "epoch": 0.55,
714
+ "learning_rate": 1.8407725571571448e-05,
715
+ "loss": 0.4188,
716
+ "step": 236
717
+ },
718
+ {
719
+ "epoch": 0.55,
720
+ "learning_rate": 1.838115322913807e-05,
721
+ "loss": 0.4409,
722
+ "step": 238
723
+ },
724
+ {
725
+ "epoch": 0.56,
726
+ "learning_rate": 1.835438050577057e-05,
727
+ "loss": 0.4109,
728
+ "step": 240
729
+ },
730
+ {
731
+ "epoch": 0.56,
732
+ "learning_rate": 1.8327408041565013e-05,
733
+ "loss": 0.4247,
734
+ "step": 242
735
+ },
736
+ {
737
+ "epoch": 0.57,
738
+ "learning_rate": 1.8300236481392995e-05,
739
+ "loss": 0.4451,
740
+ "step": 244
741
+ },
742
+ {
743
+ "epoch": 0.57,
744
+ "learning_rate": 1.8272866474886185e-05,
745
+ "loss": 0.4127,
746
+ "step": 246
747
+ },
748
+ {
749
+ "epoch": 0.58,
750
+ "learning_rate": 1.8245298676420814e-05,
751
+ "loss": 0.4346,
752
+ "step": 248
753
+ },
754
+ {
755
+ "epoch": 0.58,
756
+ "learning_rate": 1.8217533745102032e-05,
757
+ "loss": 0.4078,
758
+ "step": 250
759
+ },
760
+ {
761
+ "epoch": 0.59,
762
+ "learning_rate": 1.818957234474813e-05,
763
+ "loss": 0.4034,
764
+ "step": 252
765
+ },
766
+ {
767
+ "epoch": 0.59,
768
+ "learning_rate": 1.81614151438747e-05,
769
+ "loss": 0.4355,
770
+ "step": 254
771
+ },
772
+ {
773
+ "epoch": 0.6,
774
+ "learning_rate": 1.8133062815678614e-05,
775
+ "loss": 0.446,
776
+ "step": 256
777
+ },
778
+ {
779
+ "epoch": 0.6,
780
+ "learning_rate": 1.810451603802196e-05,
781
+ "loss": 0.4329,
782
+ "step": 258
783
+ },
784
+ {
785
+ "epoch": 0.61,
786
+ "learning_rate": 1.807577549341582e-05,
787
+ "loss": 0.4387,
788
+ "step": 260
789
+ },
790
+ {
791
+ "epoch": 0.61,
792
+ "learning_rate": 1.8046841869003962e-05,
793
+ "loss": 0.4001,
794
+ "step": 262
795
+ },
796
+ {
797
+ "epoch": 0.61,
798
+ "learning_rate": 1.8017715856546397e-05,
799
+ "loss": 0.4109,
800
+ "step": 264
801
+ },
802
+ {
803
+ "epoch": 0.62,
804
+ "learning_rate": 1.7988398152402857e-05,
805
+ "loss": 0.4156,
806
+ "step": 266
807
+ },
808
+ {
809
+ "epoch": 0.62,
810
+ "learning_rate": 1.7958889457516134e-05,
811
+ "loss": 0.4121,
812
+ "step": 268
813
+ },
814
+ {
815
+ "epoch": 0.63,
816
+ "learning_rate": 1.7929190477395318e-05,
817
+ "loss": 0.4187,
818
+ "step": 270
819
+ },
820
+ {
821
+ "epoch": 0.63,
822
+ "learning_rate": 1.7899301922098958e-05,
823
+ "loss": 0.4072,
824
+ "step": 272
825
+ },
826
+ {
827
+ "epoch": 0.64,
828
+ "learning_rate": 1.7869224506218034e-05,
829
+ "loss": 0.4556,
830
+ "step": 274
831
+ },
832
+ {
833
+ "epoch": 0.64,
834
+ "learning_rate": 1.7838958948858923e-05,
835
+ "loss": 0.4135,
836
+ "step": 276
837
+ },
838
+ {
839
+ "epoch": 0.65,
840
+ "learning_rate": 1.7808505973626183e-05,
841
+ "loss": 0.4384,
842
+ "step": 278
843
+ },
844
+ {
845
+ "epoch": 0.65,
846
+ "learning_rate": 1.777786630860525e-05,
847
+ "loss": 0.4226,
848
+ "step": 280
849
+ },
850
+ {
851
+ "epoch": 0.66,
852
+ "learning_rate": 1.774704068634504e-05,
853
+ "loss": 0.4362,
854
+ "step": 282
855
+ },
856
+ {
857
+ "epoch": 0.66,
858
+ "learning_rate": 1.771602984384043e-05,
859
+ "loss": 0.4243,
860
+ "step": 284
861
+ },
862
+ {
863
+ "epoch": 0.67,
864
+ "learning_rate": 1.7684834522514632e-05,
865
+ "loss": 0.4622,
866
+ "step": 286
867
+ },
868
+ {
869
+ "epoch": 0.67,
870
+ "learning_rate": 1.7653455468201483e-05,
871
+ "loss": 0.448,
872
+ "step": 288
873
+ },
874
+ {
875
+ "epoch": 0.68,
876
+ "learning_rate": 1.7621893431127596e-05,
877
+ "loss": 0.4385,
878
+ "step": 290
879
+ },
880
+ {
881
+ "epoch": 0.68,
882
+ "learning_rate": 1.759014916589443e-05,
883
+ "loss": 0.4149,
884
+ "step": 292
885
+ },
886
+ {
887
+ "epoch": 0.68,
888
+ "learning_rate": 1.7558223431460254e-05,
889
+ "loss": 0.4229,
890
+ "step": 294
891
+ },
892
+ {
893
+ "epoch": 0.69,
894
+ "learning_rate": 1.7526116991121988e-05,
895
+ "loss": 0.4115,
896
+ "step": 296
897
+ },
898
+ {
899
+ "epoch": 0.69,
900
+ "learning_rate": 1.7493830612496975e-05,
901
+ "loss": 0.4204,
902
+ "step": 298
903
+ },
904
+ {
905
+ "epoch": 0.7,
906
+ "learning_rate": 1.7461365067504602e-05,
907
+ "loss": 0.4171,
908
+ "step": 300
909
+ },
910
+ {
911
+ "epoch": 0.7,
912
+ "learning_rate": 1.7428721132347863e-05,
913
+ "loss": 0.4161,
914
+ "step": 302
915
+ },
916
+ {
917
+ "epoch": 0.71,
918
+ "learning_rate": 1.73958995874948e-05,
919
+ "loss": 0.4214,
920
+ "step": 304
921
+ },
922
+ {
923
+ "epoch": 0.71,
924
+ "learning_rate": 1.7362901217659833e-05,
925
+ "loss": 0.4175,
926
+ "step": 306
927
+ },
928
+ {
929
+ "epoch": 0.72,
930
+ "learning_rate": 1.7329726811785012e-05,
931
+ "loss": 0.4105,
932
+ "step": 308
933
+ },
934
+ {
935
+ "epoch": 0.72,
936
+ "learning_rate": 1.7296377163021133e-05,
937
+ "loss": 0.4354,
938
+ "step": 310
939
+ },
940
+ {
941
+ "epoch": 0.73,
942
+ "learning_rate": 1.7262853068708807e-05,
943
+ "loss": 0.4113,
944
+ "step": 312
945
+ },
946
+ {
947
+ "epoch": 0.73,
948
+ "learning_rate": 1.7229155330359368e-05,
949
+ "loss": 0.452,
950
+ "step": 314
951
+ },
952
+ {
953
+ "epoch": 0.74,
954
+ "learning_rate": 1.719528475363573e-05,
955
+ "loss": 0.4154,
956
+ "step": 316
957
+ },
958
+ {
959
+ "epoch": 0.74,
960
+ "learning_rate": 1.7161242148333107e-05,
961
+ "loss": 0.4236,
962
+ "step": 318
963
+ },
964
+ {
965
+ "epoch": 0.75,
966
+ "learning_rate": 1.712702832835966e-05,
967
+ "loss": 0.4146,
968
+ "step": 320
969
+ },
970
+ {
971
+ "epoch": 0.75,
972
+ "learning_rate": 1.7092644111717052e-05,
973
+ "loss": 0.4183,
974
+ "step": 322
975
+ },
976
+ {
977
+ "epoch": 0.75,
978
+ "learning_rate": 1.7058090320480866e-05,
979
+ "loss": 0.4038,
980
+ "step": 324
981
+ },
982
+ {
983
+ "epoch": 0.76,
984
+ "learning_rate": 1.702336778078096e-05,
985
+ "loss": 0.4135,
986
+ "step": 326
987
+ },
988
+ {
989
+ "epoch": 0.76,
990
+ "learning_rate": 1.698847732278173e-05,
991
+ "loss": 0.408,
992
+ "step": 328
993
+ },
994
+ {
995
+ "epoch": 0.77,
996
+ "learning_rate": 1.6953419780662232e-05,
997
+ "loss": 0.4003,
998
+ "step": 330
999
+ },
1000
+ {
1001
+ "epoch": 0.77,
1002
+ "learning_rate": 1.6918195992596274e-05,
1003
+ "loss": 0.4065,
1004
+ "step": 332
1005
+ },
1006
+ {
1007
+ "epoch": 0.78,
1008
+ "learning_rate": 1.6882806800732338e-05,
1009
+ "loss": 0.4205,
1010
+ "step": 334
1011
+ },
1012
+ {
1013
+ "epoch": 0.78,
1014
+ "learning_rate": 1.6847253051173487e-05,
1015
+ "loss": 0.4135,
1016
+ "step": 336
1017
+ },
1018
+ {
1019
+ "epoch": 0.79,
1020
+ "learning_rate": 1.6811535593957093e-05,
1021
+ "loss": 0.3965,
1022
+ "step": 338
1023
+ },
1024
+ {
1025
+ "epoch": 0.79,
1026
+ "learning_rate": 1.6775655283034548e-05,
1027
+ "loss": 0.4028,
1028
+ "step": 340
1029
+ },
1030
+ {
1031
+ "epoch": 0.8,
1032
+ "learning_rate": 1.6739612976250836e-05,
1033
+ "loss": 0.4578,
1034
+ "step": 342
1035
+ },
1036
+ {
1037
+ "epoch": 0.8,
1038
+ "learning_rate": 1.670340953532401e-05,
1039
+ "loss": 0.4298,
1040
+ "step": 344
1041
+ },
1042
+ {
1043
+ "epoch": 0.81,
1044
+ "learning_rate": 1.6667045825824616e-05,
1045
+ "loss": 0.4221,
1046
+ "step": 346
1047
+ },
1048
+ {
1049
+ "epoch": 0.81,
1050
+ "learning_rate": 1.663052271715497e-05,
1051
+ "loss": 0.4062,
1052
+ "step": 348
1053
+ },
1054
+ {
1055
+ "epoch": 0.81,
1056
+ "learning_rate": 1.6593841082528394e-05,
1057
+ "loss": 0.3934,
1058
+ "step": 350
1059
+ },
1060
+ {
1061
+ "epoch": 0.82,
1062
+ "learning_rate": 1.6557001798948324e-05,
1063
+ "loss": 0.4279,
1064
+ "step": 352
1065
+ },
1066
+ {
1067
+ "epoch": 0.82,
1068
+ "learning_rate": 1.6520005747187358e-05,
1069
+ "loss": 0.3993,
1070
+ "step": 354
1071
+ },
1072
+ {
1073
+ "epoch": 0.83,
1074
+ "learning_rate": 1.648285381176618e-05,
1075
+ "loss": 0.4191,
1076
+ "step": 356
1077
+ },
1078
+ {
1079
+ "epoch": 0.83,
1080
+ "learning_rate": 1.6445546880932425e-05,
1081
+ "loss": 0.4198,
1082
+ "step": 358
1083
+ },
1084
+ {
1085
+ "epoch": 0.84,
1086
+ "learning_rate": 1.6408085846639435e-05,
1087
+ "loss": 0.4,
1088
+ "step": 360
1089
+ },
1090
+ {
1091
+ "epoch": 0.84,
1092
+ "learning_rate": 1.637047160452494e-05,
1093
+ "loss": 0.4347,
1094
+ "step": 362
1095
+ },
1096
+ {
1097
+ "epoch": 0.85,
1098
+ "learning_rate": 1.6332705053889643e-05,
1099
+ "loss": 0.4188,
1100
+ "step": 364
1101
+ },
1102
+ {
1103
+ "epoch": 0.85,
1104
+ "learning_rate": 1.6294787097675712e-05,
1105
+ "loss": 0.4052,
1106
+ "step": 366
1107
+ },
1108
+ {
1109
+ "epoch": 0.86,
1110
+ "learning_rate": 1.6256718642445202e-05,
1111
+ "loss": 0.4214,
1112
+ "step": 368
1113
+ },
1114
+ {
1115
+ "epoch": 0.86,
1116
+ "learning_rate": 1.6218500598358376e-05,
1117
+ "loss": 0.4283,
1118
+ "step": 370
1119
+ },
1120
+ {
1121
+ "epoch": 0.87,
1122
+ "learning_rate": 1.6180133879151943e-05,
1123
+ "loss": 0.4188,
1124
+ "step": 372
1125
+ },
1126
+ {
1127
+ "epoch": 0.87,
1128
+ "learning_rate": 1.6141619402117213e-05,
1129
+ "loss": 0.3989,
1130
+ "step": 374
1131
+ },
1132
+ {
1133
+ "epoch": 0.88,
1134
+ "learning_rate": 1.6102958088078172e-05,
1135
+ "loss": 0.4126,
1136
+ "step": 376
1137
+ },
1138
+ {
1139
+ "epoch": 0.88,
1140
+ "learning_rate": 1.606415086136945e-05,
1141
+ "loss": 0.4148,
1142
+ "step": 378
1143
+ },
1144
+ {
1145
+ "epoch": 0.88,
1146
+ "learning_rate": 1.6025198649814243e-05,
1147
+ "loss": 0.42,
1148
+ "step": 380
1149
+ },
1150
+ {
1151
+ "epoch": 0.89,
1152
+ "learning_rate": 1.5986102384702112e-05,
1153
+ "loss": 0.4398,
1154
+ "step": 382
1155
+ },
1156
+ {
1157
+ "epoch": 0.89,
1158
+ "learning_rate": 1.594686300076673e-05,
1159
+ "loss": 0.3987,
1160
+ "step": 384
1161
+ },
1162
+ {
1163
+ "epoch": 0.9,
1164
+ "learning_rate": 1.590748143616353e-05,
1165
+ "loss": 0.4313,
1166
+ "step": 386
1167
+ },
1168
+ {
1169
+ "epoch": 0.9,
1170
+ "learning_rate": 1.5867958632447263e-05,
1171
+ "loss": 0.4214,
1172
+ "step": 388
1173
+ },
1174
+ {
1175
+ "epoch": 0.91,
1176
+ "learning_rate": 1.582829553454951e-05,
1177
+ "loss": 0.4066,
1178
+ "step": 390
1179
+ },
1180
+ {
1181
+ "epoch": 0.91,
1182
+ "learning_rate": 1.5788493090756074e-05,
1183
+ "loss": 0.4064,
1184
+ "step": 392
1185
+ },
1186
+ {
1187
+ "epoch": 0.92,
1188
+ "learning_rate": 1.5748552252684303e-05,
1189
+ "loss": 0.4109,
1190
+ "step": 394
1191
+ },
1192
+ {
1193
+ "epoch": 0.92,
1194
+ "learning_rate": 1.5708473975260356e-05,
1195
+ "loss": 0.4282,
1196
+ "step": 396
1197
+ },
1198
+ {
1199
+ "epoch": 0.93,
1200
+ "learning_rate": 1.5668259216696366e-05,
1201
+ "loss": 0.4358,
1202
+ "step": 398
1203
+ },
1204
+ {
1205
+ "epoch": 0.93,
1206
+ "learning_rate": 1.5627908938467516e-05,
1207
+ "loss": 0.4303,
1208
+ "step": 400
1209
+ },
1210
+ {
1211
+ "epoch": 0.94,
1212
+ "learning_rate": 1.558742410528907e-05,
1213
+ "loss": 0.4082,
1214
+ "step": 402
1215
+ },
1216
+ {
1217
+ "epoch": 0.94,
1218
+ "learning_rate": 1.5546805685093308e-05,
1219
+ "loss": 0.4041,
1220
+ "step": 404
1221
+ },
1222
+ {
1223
+ "epoch": 0.95,
1224
+ "learning_rate": 1.550605464900636e-05,
1225
+ "loss": 0.4148,
1226
+ "step": 406
1227
+ },
1228
+ {
1229
+ "epoch": 0.95,
1230
+ "learning_rate": 1.546517197132502e-05,
1231
+ "loss": 0.386,
1232
+ "step": 408
1233
+ },
1234
+ {
1235
+ "epoch": 0.95,
1236
+ "learning_rate": 1.542415862949343e-05,
1237
+ "loss": 0.4227,
1238
+ "step": 410
1239
+ },
1240
+ {
1241
+ "epoch": 0.96,
1242
+ "learning_rate": 1.5383015604079723e-05,
1243
+ "loss": 0.4174,
1244
+ "step": 412
1245
+ },
1246
+ {
1247
+ "epoch": 0.96,
1248
+ "learning_rate": 1.5341743878752563e-05,
1249
+ "loss": 0.4302,
1250
+ "step": 414
1251
+ },
1252
+ {
1253
+ "epoch": 0.97,
1254
+ "learning_rate": 1.5300344440257657e-05,
1255
+ "loss": 0.4076,
1256
+ "step": 416
1257
+ },
1258
+ {
1259
+ "epoch": 0.97,
1260
+ "learning_rate": 1.5258818278394125e-05,
1261
+ "loss": 0.4047,
1262
+ "step": 418
1263
+ },
1264
+ {
1265
+ "epoch": 0.98,
1266
+ "learning_rate": 1.5217166385990865e-05,
1267
+ "loss": 0.4242,
1268
+ "step": 420
1269
+ },
1270
+ {
1271
+ "epoch": 0.98,
1272
+ "learning_rate": 1.5175389758882803e-05,
1273
+ "loss": 0.4032,
1274
+ "step": 422
1275
+ },
1276
+ {
1277
+ "epoch": 0.99,
1278
+ "learning_rate": 1.5133489395887089e-05,
1279
+ "loss": 0.4268,
1280
+ "step": 424
1281
+ },
1282
+ {
1283
+ "epoch": 0.99,
1284
+ "learning_rate": 1.509146629877921e-05,
1285
+ "loss": 0.4132,
1286
+ "step": 426
1287
+ },
1288
+ {
1289
+ "epoch": 1.0,
1290
+ "learning_rate": 1.5049321472269043e-05,
1291
+ "loss": 0.4031,
1292
+ "step": 428
1293
+ },
1294
+ {
1295
+ "epoch": 1.0,
1296
+ "learning_rate": 1.5007055923976843e-05,
1297
+ "loss": 0.3714,
1298
+ "step": 430
1299
+ },
1300
+ {
1301
+ "epoch": 1.01,
1302
+ "learning_rate": 1.4964670664409136e-05,
1303
+ "loss": 0.235,
1304
+ "step": 432
1305
+ },
1306
+ {
1307
+ "epoch": 1.01,
1308
+ "learning_rate": 1.4922166706934566e-05,
1309
+ "loss": 0.2015,
1310
+ "step": 434
1311
+ },
1312
+ {
1313
+ "epoch": 1.02,
1314
+ "learning_rate": 1.4879545067759673e-05,
1315
+ "loss": 0.2057,
1316
+ "step": 436
1317
+ },
1318
+ {
1319
+ "epoch": 1.02,
1320
+ "learning_rate": 1.4836806765904587e-05,
1321
+ "loss": 0.1876,
1322
+ "step": 438
1323
+ },
1324
+ {
1325
+ "epoch": 1.02,
1326
+ "learning_rate": 1.4793952823178676e-05,
1327
+ "loss": 0.1879,
1328
+ "step": 440
1329
+ },
1330
+ {
1331
+ "epoch": 1.03,
1332
+ "learning_rate": 1.4750984264156103e-05,
1333
+ "loss": 0.1897,
1334
+ "step": 442
1335
+ },
1336
+ {
1337
+ "epoch": 1.03,
1338
+ "learning_rate": 1.4707902116151338e-05,
1339
+ "loss": 0.2166,
1340
+ "step": 444
1341
+ },
1342
+ {
1343
+ "epoch": 1.04,
1344
+ "learning_rate": 1.4664707409194598e-05,
1345
+ "loss": 0.1852,
1346
+ "step": 446
1347
+ },
1348
+ {
1349
+ "epoch": 1.04,
1350
+ "learning_rate": 1.462140117600721e-05,
1351
+ "loss": 0.1909,
1352
+ "step": 448
1353
+ },
1354
+ {
1355
+ "epoch": 1.05,
1356
+ "learning_rate": 1.457798445197694e-05,
1357
+ "loss": 0.1845,
1358
+ "step": 450
1359
+ },
1360
+ {
1361
+ "epoch": 1.05,
1362
+ "learning_rate": 1.4534458275133214e-05,
1363
+ "loss": 0.1772,
1364
+ "step": 452
1365
+ },
1366
+ {
1367
+ "epoch": 1.06,
1368
+ "learning_rate": 1.449082368612232e-05,
1369
+ "loss": 0.1873,
1370
+ "step": 454
1371
+ },
1372
+ {
1373
+ "epoch": 1.06,
1374
+ "learning_rate": 1.4447081728182518e-05,
1375
+ "loss": 0.1983,
1376
+ "step": 456
1377
+ },
1378
+ {
1379
+ "epoch": 1.07,
1380
+ "learning_rate": 1.4403233447119096e-05,
1381
+ "loss": 0.192,
1382
+ "step": 458
1383
+ },
1384
+ {
1385
+ "epoch": 1.07,
1386
+ "learning_rate": 1.4359279891279376e-05,
1387
+ "loss": 0.1808,
1388
+ "step": 460
1389
+ },
1390
+ {
1391
+ "epoch": 1.08,
1392
+ "learning_rate": 1.431522211152764e-05,
1393
+ "loss": 0.1893,
1394
+ "step": 462
1395
+ },
1396
+ {
1397
+ "epoch": 1.08,
1398
+ "learning_rate": 1.4271061161220007e-05,
1399
+ "loss": 0.186,
1400
+ "step": 464
1401
+ },
1402
+ {
1403
+ "epoch": 1.08,
1404
+ "learning_rate": 1.4226798096179262e-05,
1405
+ "loss": 0.1854,
1406
+ "step": 466
1407
+ },
1408
+ {
1409
+ "epoch": 1.09,
1410
+ "learning_rate": 1.4182433974669584e-05,
1411
+ "loss": 0.1736,
1412
+ "step": 468
1413
+ },
1414
+ {
1415
+ "epoch": 1.09,
1416
+ "learning_rate": 1.4137969857371277e-05,
1417
+ "loss": 0.1876,
1418
+ "step": 470
1419
+ },
1420
+ {
1421
+ "epoch": 1.1,
1422
+ "learning_rate": 1.4093406807355389e-05,
1423
+ "loss": 0.1904,
1424
+ "step": 472
1425
+ },
1426
+ {
1427
+ "epoch": 1.1,
1428
+ "learning_rate": 1.4048745890058304e-05,
1429
+ "loss": 0.1829,
1430
+ "step": 474
1431
+ },
1432
+ {
1433
+ "epoch": 1.11,
1434
+ "learning_rate": 1.4003988173256267e-05,
1435
+ "loss": 0.1835,
1436
+ "step": 476
1437
+ },
1438
+ {
1439
+ "epoch": 1.11,
1440
+ "learning_rate": 1.3959134727039854e-05,
1441
+ "loss": 0.1829,
1442
+ "step": 478
1443
+ },
1444
+ {
1445
+ "epoch": 1.12,
1446
+ "learning_rate": 1.3914186623788398e-05,
1447
+ "loss": 0.1907,
1448
+ "step": 480
1449
+ },
1450
+ {
1451
+ "epoch": 1.12,
1452
+ "learning_rate": 1.3869144938144325e-05,
1453
+ "loss": 0.1842,
1454
+ "step": 482
1455
+ },
1456
+ {
1457
+ "epoch": 1.13,
1458
+ "learning_rate": 1.3824010746987495e-05,
1459
+ "loss": 0.1929,
1460
+ "step": 484
1461
+ },
1462
+ {
1463
+ "epoch": 1.13,
1464
+ "learning_rate": 1.3778785129409424e-05,
1465
+ "loss": 0.1824,
1466
+ "step": 486
1467
+ },
1468
+ {
1469
+ "epoch": 1.14,
1470
+ "learning_rate": 1.3733469166687505e-05,
1471
+ "loss": 0.1867,
1472
+ "step": 488
1473
+ },
1474
+ {
1475
+ "epoch": 1.14,
1476
+ "learning_rate": 1.3688063942259141e-05,
1477
+ "loss": 0.1842,
1478
+ "step": 490
1479
+ },
1480
+ {
1481
+ "epoch": 1.15,
1482
+ "learning_rate": 1.3642570541695867e-05,
1483
+ "loss": 0.1874,
1484
+ "step": 492
1485
+ },
1486
+ {
1487
+ "epoch": 1.15,
1488
+ "learning_rate": 1.359699005267736e-05,
1489
+ "loss": 0.1985,
1490
+ "step": 494
1491
+ },
1492
+ {
1493
+ "epoch": 1.15,
1494
+ "learning_rate": 1.3551323564965465e-05,
1495
+ "loss": 0.1671,
1496
+ "step": 496
1497
+ },
1498
+ {
1499
+ "epoch": 1.16,
1500
+ "learning_rate": 1.3505572170378118e-05,
1501
+ "loss": 0.1861,
1502
+ "step": 498
1503
+ },
1504
+ {
1505
+ "epoch": 1.16,
1506
+ "learning_rate": 1.3459736962763263e-05,
1507
+ "loss": 0.1873,
1508
+ "step": 500
1509
+ },
1510
+ {
1511
+ "epoch": 1.17,
1512
+ "learning_rate": 1.3413819037972682e-05,
1513
+ "loss": 0.1946,
1514
+ "step": 502
1515
+ },
1516
+ {
1517
+ "epoch": 1.17,
1518
+ "learning_rate": 1.33678194938358e-05,
1519
+ "loss": 0.1769,
1520
+ "step": 504
1521
+ },
1522
+ {
1523
+ "epoch": 1.18,
1524
+ "learning_rate": 1.332173943013345e-05,
1525
+ "loss": 0.205,
1526
+ "step": 506
1527
+ },
1528
+ {
1529
+ "epoch": 1.18,
1530
+ "learning_rate": 1.327557994857156e-05,
1531
+ "loss": 0.1899,
1532
+ "step": 508
1533
+ },
1534
+ {
1535
+ "epoch": 1.19,
1536
+ "learning_rate": 1.322934215275482e-05,
1537
+ "loss": 0.1746,
1538
+ "step": 510
1539
+ },
1540
+ {
1541
+ "epoch": 1.19,
1542
+ "learning_rate": 1.3183027148160304e-05,
1543
+ "loss": 0.1843,
1544
+ "step": 512
1545
+ },
1546
+ {
1547
+ "epoch": 1.2,
1548
+ "learning_rate": 1.3136636042111025e-05,
1549
+ "loss": 0.165,
1550
+ "step": 514
1551
+ },
1552
+ {
1553
+ "epoch": 1.2,
1554
+ "learning_rate": 1.3090169943749475e-05,
1555
+ "loss": 0.189,
1556
+ "step": 516
1557
+ },
1558
+ {
1559
+ "epoch": 1.21,
1560
+ "learning_rate": 1.3043629964011104e-05,
1561
+ "loss": 0.1745,
1562
+ "step": 518
1563
+ },
1564
+ {
1565
+ "epoch": 1.21,
1566
+ "learning_rate": 1.2997017215597743e-05,
1567
+ "loss": 0.1829,
1568
+ "step": 520
1569
+ },
1570
+ {
1571
+ "epoch": 1.22,
1572
+ "learning_rate": 1.295033281295103e-05,
1573
+ "loss": 0.1871,
1574
+ "step": 522
1575
+ },
1576
+ {
1577
+ "epoch": 1.22,
1578
+ "learning_rate": 1.2903577872225737e-05,
1579
+ "loss": 0.1786,
1580
+ "step": 524
1581
+ },
1582
+ {
1583
+ "epoch": 1.22,
1584
+ "learning_rate": 1.2856753511263105e-05,
1585
+ "loss": 0.1759,
1586
+ "step": 526
1587
+ },
1588
+ {
1589
+ "epoch": 1.23,
1590
+ "learning_rate": 1.2809860849564103e-05,
1591
+ "loss": 0.2027,
1592
+ "step": 528
1593
+ },
1594
+ {
1595
+ "epoch": 1.23,
1596
+ "learning_rate": 1.2762901008262678e-05,
1597
+ "loss": 0.1824,
1598
+ "step": 530
1599
+ },
1600
+ {
1601
+ "epoch": 1.24,
1602
+ "learning_rate": 1.271587511009893e-05,
1603
+ "loss": 0.1805,
1604
+ "step": 532
1605
+ },
1606
+ {
1607
+ "epoch": 1.24,
1608
+ "learning_rate": 1.2668784279392287e-05,
1609
+ "loss": 0.1777,
1610
+ "step": 534
1611
+ },
1612
+ {
1613
+ "epoch": 1.25,
1614
+ "learning_rate": 1.2621629642014623e-05,
1615
+ "loss": 0.1873,
1616
+ "step": 536
1617
+ },
1618
+ {
1619
+ "epoch": 1.25,
1620
+ "learning_rate": 1.2574412325363326e-05,
1621
+ "loss": 0.184,
1622
+ "step": 538
1623
+ },
1624
+ {
1625
+ "epoch": 1.26,
1626
+ "learning_rate": 1.2527133458334353e-05,
1627
+ "loss": 0.1932,
1628
+ "step": 540
1629
+ },
1630
+ {
1631
+ "epoch": 1.26,
1632
+ "learning_rate": 1.2479794171295248e-05,
1633
+ "loss": 0.1875,
1634
+ "step": 542
1635
+ },
1636
+ {
1637
+ "epoch": 1.27,
1638
+ "learning_rate": 1.2432395596058097e-05,
1639
+ "loss": 0.1853,
1640
+ "step": 544
1641
+ },
1642
+ {
1643
+ "epoch": 1.27,
1644
+ "learning_rate": 1.2384938865852482e-05,
1645
+ "loss": 0.1815,
1646
+ "step": 546
1647
+ },
1648
+ {
1649
+ "epoch": 1.28,
1650
+ "learning_rate": 1.2337425115298389e-05,
1651
+ "loss": 0.1845,
1652
+ "step": 548
1653
+ },
1654
+ {
1655
+ "epoch": 1.28,
1656
+ "learning_rate": 1.2289855480379074e-05,
1657
+ "loss": 0.1953,
1658
+ "step": 550
1659
+ },
1660
+ {
1661
+ "epoch": 1.29,
1662
+ "learning_rate": 1.22422310984139e-05,
1663
+ "loss": 0.1886,
1664
+ "step": 552
1665
+ },
1666
+ {
1667
+ "epoch": 1.29,
1668
+ "learning_rate": 1.2194553108031153e-05,
1669
+ "loss": 0.1875,
1670
+ "step": 554
1671
+ },
1672
+ {
1673
+ "epoch": 1.29,
1674
+ "learning_rate": 1.214682264914082e-05,
1675
+ "loss": 0.1829,
1676
+ "step": 556
1677
+ },
1678
+ {
1679
+ "epoch": 1.3,
1680
+ "learning_rate": 1.2099040862907332e-05,
1681
+ "loss": 0.1935,
1682
+ "step": 558
1683
+ },
1684
+ {
1685
+ "epoch": 1.3,
1686
+ "learning_rate": 1.2051208891722274e-05,
1687
+ "loss": 0.1851,
1688
+ "step": 560
1689
+ },
1690
+ {
1691
+ "epoch": 1.31,
1692
+ "learning_rate": 1.2003327879177085e-05,
1693
+ "loss": 0.1991,
1694
+ "step": 562
1695
+ },
1696
+ {
1697
+ "epoch": 1.31,
1698
+ "learning_rate": 1.195539897003571e-05,
1699
+ "loss": 0.1927,
1700
+ "step": 564
1701
+ },
1702
+ {
1703
+ "epoch": 1.32,
1704
+ "learning_rate": 1.190742331020723e-05,
1705
+ "loss": 0.183,
1706
+ "step": 566
1707
+ },
1708
+ {
1709
+ "epoch": 1.32,
1710
+ "learning_rate": 1.185940204671846e-05,
1711
+ "loss": 0.201,
1712
+ "step": 568
1713
+ },
1714
+ {
1715
+ "epoch": 1.33,
1716
+ "learning_rate": 1.1811336327686537e-05,
1717
+ "loss": 0.198,
1718
+ "step": 570
1719
+ },
1720
+ {
1721
+ "epoch": 1.33,
1722
+ "learning_rate": 1.1763227302291464e-05,
1723
+ "loss": 0.185,
1724
+ "step": 572
1725
+ },
1726
+ {
1727
+ "epoch": 1.34,
1728
+ "learning_rate": 1.1715076120748631e-05,
1729
+ "loss": 0.179,
1730
+ "step": 574
1731
+ },
1732
+ {
1733
+ "epoch": 1.34,
1734
+ "learning_rate": 1.1666883934281324e-05,
1735
+ "loss": 0.1934,
1736
+ "step": 576
1737
+ },
1738
+ {
1739
+ "epoch": 1.35,
1740
+ "learning_rate": 1.1618651895093192e-05,
1741
+ "loss": 0.1996,
1742
+ "step": 578
1743
+ },
1744
+ {
1745
+ "epoch": 1.35,
1746
+ "learning_rate": 1.1570381156340701e-05,
1747
+ "loss": 0.1813,
1748
+ "step": 580
1749
+ },
1750
+ {
1751
+ "epoch": 1.36,
1752
+ "learning_rate": 1.1522072872105576e-05,
1753
+ "loss": 0.1874,
1754
+ "step": 582
1755
+ },
1756
+ {
1757
+ "epoch": 1.36,
1758
+ "learning_rate": 1.147372819736719e-05,
1759
+ "loss": 0.1773,
1760
+ "step": 584
1761
+ },
1762
+ {
1763
+ "epoch": 1.36,
1764
+ "learning_rate": 1.1425348287974956e-05,
1765
+ "loss": 0.1912,
1766
+ "step": 586
1767
+ },
1768
+ {
1769
+ "epoch": 1.37,
1770
+ "learning_rate": 1.1376934300620706e-05,
1771
+ "loss": 0.1949,
1772
+ "step": 588
1773
+ },
1774
+ {
1775
+ "epoch": 1.37,
1776
+ "learning_rate": 1.1328487392811019e-05,
1777
+ "loss": 0.1883,
1778
+ "step": 590
1779
+ },
1780
+ {
1781
+ "epoch": 1.38,
1782
+ "learning_rate": 1.1280008722839552e-05,
1783
+ "loss": 0.1766,
1784
+ "step": 592
1785
+ },
1786
+ {
1787
+ "epoch": 1.38,
1788
+ "learning_rate": 1.1231499449759355e-05,
1789
+ "loss": 0.1987,
1790
+ "step": 594
1791
+ },
1792
+ {
1793
+ "epoch": 1.39,
1794
+ "learning_rate": 1.1182960733355142e-05,
1795
+ "loss": 0.1785,
1796
+ "step": 596
1797
+ },
1798
+ {
1799
+ "epoch": 1.39,
1800
+ "learning_rate": 1.1134393734115587e-05,
1801
+ "loss": 0.1961,
1802
+ "step": 598
1803
+ },
1804
+ {
1805
+ "epoch": 1.4,
1806
+ "learning_rate": 1.1085799613205552e-05,
1807
+ "loss": 0.1805,
1808
+ "step": 600
1809
+ },
1810
+ {
1811
+ "epoch": 1.4,
1812
+ "learning_rate": 1.1037179532438345e-05,
1813
+ "loss": 0.1745,
1814
+ "step": 602
1815
+ },
1816
+ {
1817
+ "epoch": 1.41,
1818
+ "learning_rate": 1.098853465424793e-05,
1819
+ "loss": 0.1904,
1820
+ "step": 604
1821
+ },
1822
+ {
1823
+ "epoch": 1.41,
1824
+ "learning_rate": 1.0939866141661148e-05,
1825
+ "loss": 0.1858,
1826
+ "step": 606
1827
+ },
1828
+ {
1829
+ "epoch": 1.42,
1830
+ "learning_rate": 1.08911751582699e-05,
1831
+ "loss": 0.2095,
1832
+ "step": 608
1833
+ },
1834
+ {
1835
+ "epoch": 1.42,
1836
+ "learning_rate": 1.0842462868203329e-05,
1837
+ "loss": 0.1935,
1838
+ "step": 610
1839
+ },
1840
+ {
1841
+ "epoch": 1.42,
1842
+ "learning_rate": 1.079373043609999e-05,
1843
+ "loss": 0.1807,
1844
+ "step": 612
1845
+ },
1846
+ {
1847
+ "epoch": 1.43,
1848
+ "learning_rate": 1.0744979027080003e-05,
1849
+ "loss": 0.194,
1850
+ "step": 614
1851
+ },
1852
+ {
1853
+ "epoch": 1.43,
1854
+ "learning_rate": 1.06962098067172e-05,
1855
+ "loss": 0.196,
1856
+ "step": 616
1857
+ },
1858
+ {
1859
+ "epoch": 1.44,
1860
+ "learning_rate": 1.0647423941011255e-05,
1861
+ "loss": 0.1916,
1862
+ "step": 618
1863
+ },
1864
+ {
1865
+ "epoch": 1.44,
1866
+ "learning_rate": 1.0598622596359808e-05,
1867
+ "loss": 0.1904,
1868
+ "step": 620
1869
+ },
1870
+ {
1871
+ "epoch": 1.45,
1872
+ "learning_rate": 1.054980693953058e-05,
1873
+ "loss": 0.1766,
1874
+ "step": 622
1875
+ },
1876
+ {
1877
+ "epoch": 1.45,
1878
+ "learning_rate": 1.0500978137633469e-05,
1879
+ "loss": 0.1946,
1880
+ "step": 624
1881
+ },
1882
+ {
1883
+ "epoch": 1.46,
1884
+ "learning_rate": 1.0452137358092654e-05,
1885
+ "loss": 0.1918,
1886
+ "step": 626
1887
+ },
1888
+ {
1889
+ "epoch": 1.46,
1890
+ "learning_rate": 1.0403285768618682e-05,
1891
+ "loss": 0.1813,
1892
+ "step": 628
1893
+ },
1894
+ {
1895
+ "epoch": 1.47,
1896
+ "learning_rate": 1.0354424537180554e-05,
1897
+ "loss": 0.1879,
1898
+ "step": 630
1899
+ },
1900
+ {
1901
+ "epoch": 1.47,
1902
+ "learning_rate": 1.0305554831977788e-05,
1903
+ "loss": 0.1857,
1904
+ "step": 632
1905
+ },
1906
+ {
1907
+ "epoch": 1.48,
1908
+ "learning_rate": 1.0256677821412508e-05,
1909
+ "loss": 0.1949,
1910
+ "step": 634
1911
+ },
1912
+ {
1913
+ "epoch": 1.48,
1914
+ "learning_rate": 1.0207794674061483e-05,
1915
+ "loss": 0.209,
1916
+ "step": 636
1917
+ },
1918
+ {
1919
+ "epoch": 1.49,
1920
+ "learning_rate": 1.015890655864822e-05,
1921
+ "loss": 0.2652,
1922
+ "step": 638
1923
+ },
1924
+ {
1925
+ "epoch": 1.49,
1926
+ "learning_rate": 1.0110014644014994e-05,
1927
+ "loss": 0.263,
1928
+ "step": 640
1929
+ },
1930
+ {
1931
+ "epoch": 1.49,
1932
+ "learning_rate": 1.0061120099094917e-05,
1933
+ "loss": 0.2231,
1934
+ "step": 642
1935
+ },
1936
+ {
1937
+ "epoch": 1.5,
1938
+ "learning_rate": 1.0012224092883986e-05,
1939
+ "loss": 0.2141,
1940
+ "step": 644
1941
+ },
1942
+ {
1943
+ "epoch": 1.5,
1944
+ "learning_rate": 9.963327794413137e-06,
1945
+ "loss": 0.2057,
1946
+ "step": 646
1947
+ },
1948
+ {
1949
+ "epoch": 1.51,
1950
+ "learning_rate": 9.914432372720294e-06,
1951
+ "loss": 0.2352,
1952
+ "step": 648
1953
+ },
1954
+ {
1955
+ "epoch": 1.51,
1956
+ "learning_rate": 9.865538996822418e-06,
1957
+ "loss": 0.2138,
1958
+ "step": 650
1959
+ },
1960
+ {
1961
+ "epoch": 1.52,
1962
+ "learning_rate": 9.816648835687557e-06,
1963
+ "loss": 0.2054,
1964
+ "step": 652
1965
+ },
1966
+ {
1967
+ "epoch": 1.52,
1968
+ "learning_rate": 9.767763058206897e-06,
1969
+ "loss": 0.2073,
1970
+ "step": 654
1971
+ },
1972
+ {
1973
+ "epoch": 1.53,
1974
+ "learning_rate": 9.718882833166823e-06,
1975
+ "loss": 0.2001,
1976
+ "step": 656
1977
+ },
1978
+ {
1979
+ "epoch": 1.53,
1980
+ "learning_rate": 9.670009329220963e-06,
1981
+ "loss": 0.1985,
1982
+ "step": 658
1983
+ },
1984
+ {
1985
+ "epoch": 1.54,
1986
+ "learning_rate": 9.62114371486226e-06,
1987
+ "loss": 0.2006,
1988
+ "step": 660
1989
+ },
1990
+ {
1991
+ "epoch": 1.54,
1992
+ "learning_rate": 9.572287158395025e-06,
1993
+ "loss": 0.2005,
1994
+ "step": 662
1995
+ },
1996
+ {
1997
+ "epoch": 1.55,
1998
+ "learning_rate": 9.523440827907006e-06,
1999
+ "loss": 0.1974,
2000
+ "step": 664
2001
+ },
2002
+ {
2003
+ "epoch": 1.55,
2004
+ "learning_rate": 9.474605891241465e-06,
2005
+ "loss": 0.207,
2006
+ "step": 666
2007
+ },
2008
+ {
2009
+ "epoch": 1.56,
2010
+ "learning_rate": 9.425783515969258e-06,
2011
+ "loss": 0.1863,
2012
+ "step": 668
2013
+ },
2014
+ {
2015
+ "epoch": 1.56,
2016
+ "learning_rate": 9.376974869360918e-06,
2017
+ "loss": 0.2004,
2018
+ "step": 670
2019
+ },
2020
+ {
2021
+ "epoch": 1.56,
2022
+ "learning_rate": 9.328181118358734e-06,
2023
+ "loss": 0.1884,
2024
+ "step": 672
2025
+ },
2026
+ {
2027
+ "epoch": 1.57,
2028
+ "learning_rate": 9.279403429548877e-06,
2029
+ "loss": 0.1884,
2030
+ "step": 674
2031
+ },
2032
+ {
2033
+ "epoch": 1.57,
2034
+ "learning_rate": 9.230642969133483e-06,
2035
+ "loss": 0.1939,
2036
+ "step": 676
2037
+ },
2038
+ {
2039
+ "epoch": 1.58,
2040
+ "learning_rate": 9.181900902902794e-06,
2041
+ "loss": 0.2122,
2042
+ "step": 678
2043
+ },
2044
+ {
2045
+ "epoch": 1.58,
2046
+ "learning_rate": 9.13317839620727e-06,
2047
+ "loss": 0.197,
2048
+ "step": 680
2049
+ },
2050
+ {
2051
+ "epoch": 1.59,
2052
+ "learning_rate": 9.084476613929726e-06,
2053
+ "loss": 0.1765,
2054
+ "step": 682
2055
+ },
2056
+ {
2057
+ "epoch": 1.59,
2058
+ "learning_rate": 9.035796720457495e-06,
2059
+ "loss": 0.1879,
2060
+ "step": 684
2061
+ },
2062
+ {
2063
+ "epoch": 1.6,
2064
+ "learning_rate": 8.987139879654575e-06,
2065
+ "loss": 0.189,
2066
+ "step": 686
2067
+ },
2068
+ {
2069
+ "epoch": 1.6,
2070
+ "learning_rate": 8.938507254833811e-06,
2071
+ "loss": 0.1925,
2072
+ "step": 688
2073
+ },
2074
+ {
2075
+ "epoch": 1.61,
2076
+ "learning_rate": 8.889900008729084e-06,
2077
+ "loss": 0.197,
2078
+ "step": 690
2079
+ },
2080
+ {
2081
+ "epoch": 1.61,
2082
+ "learning_rate": 8.841319303467502e-06,
2083
+ "loss": 0.1954,
2084
+ "step": 692
2085
+ },
2086
+ {
2087
+ "epoch": 1.62,
2088
+ "learning_rate": 8.792766300541622e-06,
2089
+ "loss": 0.1815,
2090
+ "step": 694
2091
+ },
2092
+ {
2093
+ "epoch": 1.62,
2094
+ "learning_rate": 8.744242160781682e-06,
2095
+ "loss": 0.1914,
2096
+ "step": 696
2097
+ },
2098
+ {
2099
+ "epoch": 1.63,
2100
+ "learning_rate": 8.69574804432784e-06,
2101
+ "loss": 0.186,
2102
+ "step": 698
2103
+ },
2104
+ {
2105
+ "epoch": 1.63,
2106
+ "learning_rate": 8.647285110602443e-06,
2107
+ "loss": 0.1937,
2108
+ "step": 700
2109
+ },
2110
+ {
2111
+ "epoch": 1.63,
2112
+ "learning_rate": 8.59885451828231e-06,
2113
+ "loss": 0.198,
2114
+ "step": 702
2115
+ },
2116
+ {
2117
+ "epoch": 1.64,
2118
+ "learning_rate": 8.550457425271022e-06,
2119
+ "loss": 0.1819,
2120
+ "step": 704
2121
+ },
2122
+ {
2123
+ "epoch": 1.64,
2124
+ "learning_rate": 8.502094988671232e-06,
2125
+ "loss": 0.2001,
2126
+ "step": 706
2127
+ },
2128
+ {
2129
+ "epoch": 1.65,
2130
+ "learning_rate": 8.453768364757027e-06,
2131
+ "loss": 0.1704,
2132
+ "step": 708
2133
+ },
2134
+ {
2135
+ "epoch": 1.65,
2136
+ "learning_rate": 8.405478708946254e-06,
2137
+ "loss": 0.1873,
2138
+ "step": 710
2139
+ },
2140
+ {
2141
+ "epoch": 1.66,
2142
+ "learning_rate": 8.35722717577291e-06,
2143
+ "loss": 0.1771,
2144
+ "step": 712
2145
+ },
2146
+ {
2147
+ "epoch": 1.66,
2148
+ "learning_rate": 8.309014918859538e-06,
2149
+ "loss": 0.1843,
2150
+ "step": 714
2151
+ },
2152
+ {
2153
+ "epoch": 1.67,
2154
+ "learning_rate": 8.26084309088964e-06,
2155
+ "loss": 0.1808,
2156
+ "step": 716
2157
+ },
2158
+ {
2159
+ "epoch": 1.67,
2160
+ "learning_rate": 8.212712843580124e-06,
2161
+ "loss": 0.2045,
2162
+ "step": 718
2163
+ },
2164
+ {
2165
+ "epoch": 1.68,
2166
+ "learning_rate": 8.164625327653772e-06,
2167
+ "loss": 0.1799,
2168
+ "step": 720
2169
+ },
2170
+ {
2171
+ "epoch": 1.68,
2172
+ "learning_rate": 8.116581692811711e-06,
2173
+ "loss": 0.1838,
2174
+ "step": 722
2175
+ },
2176
+ {
2177
+ "epoch": 1.69,
2178
+ "learning_rate": 8.068583087705946e-06,
2179
+ "loss": 0.1923,
2180
+ "step": 724
2181
+ },
2182
+ {
2183
+ "epoch": 1.69,
2184
+ "learning_rate": 8.020630659911881e-06,
2185
+ "loss": 0.1827,
2186
+ "step": 726
2187
+ },
2188
+ {
2189
+ "epoch": 1.69,
2190
+ "learning_rate": 7.972725555900895e-06,
2191
+ "loss": 0.1819,
2192
+ "step": 728
2193
+ },
2194
+ {
2195
+ "epoch": 1.7,
2196
+ "learning_rate": 7.924868921012918e-06,
2197
+ "loss": 0.1824,
2198
+ "step": 730
2199
+ },
2200
+ {
2201
+ "epoch": 1.7,
2202
+ "learning_rate": 7.877061899429067e-06,
2203
+ "loss": 0.1973,
2204
+ "step": 732
2205
+ },
2206
+ {
2207
+ "epoch": 1.71,
2208
+ "learning_rate": 7.829305634144264e-06,
2209
+ "loss": 0.183,
2210
+ "step": 734
2211
+ },
2212
+ {
2213
+ "epoch": 1.71,
2214
+ "learning_rate": 7.781601266939936e-06,
2215
+ "loss": 0.1652,
2216
+ "step": 736
2217
+ },
2218
+ {
2219
+ "epoch": 1.72,
2220
+ "learning_rate": 7.733949938356695e-06,
2221
+ "loss": 0.1895,
2222
+ "step": 738
2223
+ },
2224
+ {
2225
+ "epoch": 1.72,
2226
+ "learning_rate": 7.686352787667083e-06,
2227
+ "loss": 0.1845,
2228
+ "step": 740
2229
+ },
2230
+ {
2231
+ "epoch": 1.73,
2232
+ "learning_rate": 7.638810952848328e-06,
2233
+ "loss": 0.1894,
2234
+ "step": 742
2235
+ },
2236
+ {
2237
+ "epoch": 1.73,
2238
+ "learning_rate": 7.591325570555136e-06,
2239
+ "loss": 0.1707,
2240
+ "step": 744
2241
+ },
2242
+ {
2243
+ "epoch": 1.74,
2244
+ "learning_rate": 7.543897776092519e-06,
2245
+ "loss": 0.1776,
2246
+ "step": 746
2247
+ },
2248
+ {
2249
+ "epoch": 1.74,
2250
+ "learning_rate": 7.496528703388648e-06,
2251
+ "loss": 0.1788,
2252
+ "step": 748
2253
+ },
2254
+ {
2255
+ "epoch": 1.75,
2256
+ "learning_rate": 7.449219484967749e-06,
2257
+ "loss": 0.1777,
2258
+ "step": 750
2259
+ },
2260
+ {
2261
+ "epoch": 1.75,
2262
+ "learning_rate": 7.401971251923015e-06,
2263
+ "loss": 0.183,
2264
+ "step": 752
2265
+ },
2266
+ {
2267
+ "epoch": 1.76,
2268
+ "learning_rate": 7.354785133889566e-06,
2269
+ "loss": 0.1857,
2270
+ "step": 754
2271
+ },
2272
+ {
2273
+ "epoch": 1.76,
2274
+ "learning_rate": 7.307662259017454e-06,
2275
+ "loss": 0.1892,
2276
+ "step": 756
2277
+ },
2278
+ {
2279
+ "epoch": 1.76,
2280
+ "learning_rate": 7.260603753944674e-06,
2281
+ "loss": 0.1785,
2282
+ "step": 758
2283
+ },
2284
+ {
2285
+ "epoch": 1.77,
2286
+ "learning_rate": 7.213610743770234e-06,
2287
+ "loss": 0.1884,
2288
+ "step": 760
2289
+ },
2290
+ {
2291
+ "epoch": 1.77,
2292
+ "learning_rate": 7.166684352027265e-06,
2293
+ "loss": 0.1773,
2294
+ "step": 762
2295
+ },
2296
+ {
2297
+ "epoch": 1.78,
2298
+ "learning_rate": 7.119825700656138e-06,
2299
+ "loss": 0.1862,
2300
+ "step": 764
2301
+ },
2302
+ {
2303
+ "epoch": 1.78,
2304
+ "learning_rate": 7.073035909977661e-06,
2305
+ "loss": 0.1872,
2306
+ "step": 766
2307
+ },
2308
+ {
2309
+ "epoch": 1.79,
2310
+ "learning_rate": 7.026316098666282e-06,
2311
+ "loss": 0.1917,
2312
+ "step": 768
2313
+ },
2314
+ {
2315
+ "epoch": 1.79,
2316
+ "learning_rate": 6.979667383723345e-06,
2317
+ "loss": 0.1823,
2318
+ "step": 770
2319
+ },
2320
+ {
2321
+ "epoch": 1.8,
2322
+ "learning_rate": 6.9330908804503874e-06,
2323
+ "loss": 0.179,
2324
+ "step": 772
2325
+ },
2326
+ {
2327
+ "epoch": 1.8,
2328
+ "learning_rate": 6.886587702422474e-06,
2329
+ "loss": 0.1731,
2330
+ "step": 774
2331
+ },
2332
+ {
2333
+ "epoch": 1.81,
2334
+ "learning_rate": 6.840158961461567e-06,
2335
+ "loss": 0.1843,
2336
+ "step": 776
2337
+ },
2338
+ {
2339
+ "epoch": 1.81,
2340
+ "learning_rate": 6.793805767609953e-06,
2341
+ "loss": 0.1789,
2342
+ "step": 778
2343
+ },
2344
+ {
2345
+ "epoch": 1.82,
2346
+ "learning_rate": 6.7475292291037e-06,
2347
+ "loss": 0.1851,
2348
+ "step": 780
2349
+ },
2350
+ {
2351
+ "epoch": 1.82,
2352
+ "learning_rate": 6.701330452346156e-06,
2353
+ "loss": 0.1795,
2354
+ "step": 782
2355
+ },
2356
+ {
2357
+ "epoch": 1.83,
2358
+ "learning_rate": 6.655210541881502e-06,
2359
+ "loss": 0.1907,
2360
+ "step": 784
2361
+ },
2362
+ {
2363
+ "epoch": 1.83,
2364
+ "learning_rate": 6.609170600368346e-06,
2365
+ "loss": 0.1885,
2366
+ "step": 786
2367
+ },
2368
+ {
2369
+ "epoch": 1.83,
2370
+ "learning_rate": 6.56321172855336e-06,
2371
+ "loss": 0.1804,
2372
+ "step": 788
2373
+ },
2374
+ {
2375
+ "epoch": 1.84,
2376
+ "learning_rate": 6.51733502524495e-06,
2377
+ "loss": 0.184,
2378
+ "step": 790
2379
+ },
2380
+ {
2381
+ "epoch": 1.84,
2382
+ "learning_rate": 6.471541587287003e-06,
2383
+ "loss": 0.186,
2384
+ "step": 792
2385
+ },
2386
+ {
2387
+ "epoch": 1.85,
2388
+ "learning_rate": 6.425832509532652e-06,
2389
+ "loss": 0.167,
2390
+ "step": 794
2391
+ },
2392
+ {
2393
+ "epoch": 1.85,
2394
+ "learning_rate": 6.380208884818104e-06,
2395
+ "loss": 0.1728,
2396
+ "step": 796
2397
+ },
2398
+ {
2399
+ "epoch": 1.86,
2400
+ "learning_rate": 6.3346718039365076e-06,
2401
+ "loss": 0.1765,
2402
+ "step": 798
2403
+ },
2404
+ {
2405
+ "epoch": 1.86,
2406
+ "learning_rate": 6.289222355611881e-06,
2407
+ "loss": 0.1813,
2408
+ "step": 800
2409
+ },
2410
+ {
2411
+ "epoch": 1.87,
2412
+ "learning_rate": 6.243861626473073e-06,
2413
+ "loss": 0.1875,
2414
+ "step": 802
2415
+ },
2416
+ {
2417
+ "epoch": 1.87,
2418
+ "learning_rate": 6.198590701027796e-06,
2419
+ "loss": 0.1829,
2420
+ "step": 804
2421
+ },
2422
+ {
2423
+ "epoch": 1.88,
2424
+ "learning_rate": 6.153410661636683e-06,
2425
+ "loss": 0.1803,
2426
+ "step": 806
2427
+ },
2428
+ {
2429
+ "epoch": 1.88,
2430
+ "learning_rate": 6.108322588487419e-06,
2431
+ "loss": 0.1768,
2432
+ "step": 808
2433
+ },
2434
+ {
2435
+ "epoch": 1.89,
2436
+ "learning_rate": 6.063327559568908e-06,
2437
+ "loss": 0.1764,
2438
+ "step": 810
2439
+ },
2440
+ {
2441
+ "epoch": 1.89,
2442
+ "learning_rate": 6.0184266506455125e-06,
2443
+ "loss": 0.1818,
2444
+ "step": 812
2445
+ },
2446
+ {
2447
+ "epoch": 1.9,
2448
+ "learning_rate": 5.973620935231318e-06,
2449
+ "loss": 0.1834,
2450
+ "step": 814
2451
+ },
2452
+ {
2453
+ "epoch": 1.9,
2454
+ "learning_rate": 5.928911484564481e-06,
2455
+ "loss": 0.1682,
2456
+ "step": 816
2457
+ },
2458
+ {
2459
+ "epoch": 1.9,
2460
+ "learning_rate": 5.884299367581607e-06,
2461
+ "loss": 0.1828,
2462
+ "step": 818
2463
+ },
2464
+ {
2465
+ "epoch": 1.91,
2466
+ "learning_rate": 5.8397856508922e-06,
2467
+ "loss": 0.1802,
2468
+ "step": 820
2469
+ },
2470
+ {
2471
+ "epoch": 1.91,
2472
+ "learning_rate": 5.795371398753153e-06,
2473
+ "loss": 0.1949,
2474
+ "step": 822
2475
+ },
2476
+ {
2477
+ "epoch": 1.92,
2478
+ "learning_rate": 5.751057673043316e-06,
2479
+ "loss": 0.1777,
2480
+ "step": 824
2481
+ },
2482
+ {
2483
+ "epoch": 1.92,
2484
+ "learning_rate": 5.706845533238097e-06,
2485
+ "loss": 0.1728,
2486
+ "step": 826
2487
+ },
2488
+ {
2489
+ "epoch": 1.93,
2490
+ "learning_rate": 5.662736036384142e-06,
2491
+ "loss": 0.1701,
2492
+ "step": 828
2493
+ },
2494
+ {
2495
+ "epoch": 1.93,
2496
+ "learning_rate": 5.618730237074048e-06,
2497
+ "loss": 0.1667,
2498
+ "step": 830
2499
+ },
2500
+ {
2501
+ "epoch": 1.94,
2502
+ "learning_rate": 5.574829187421166e-06,
2503
+ "loss": 0.1746,
2504
+ "step": 832
2505
+ },
2506
+ {
2507
+ "epoch": 1.94,
2508
+ "learning_rate": 5.531033937034429e-06,
2509
+ "loss": 0.1827,
2510
+ "step": 834
2511
+ },
2512
+ {
2513
+ "epoch": 1.95,
2514
+ "learning_rate": 5.4873455329932736e-06,
2515
+ "loss": 0.1769,
2516
+ "step": 836
2517
+ },
2518
+ {
2519
+ "epoch": 1.95,
2520
+ "learning_rate": 5.443765019822593e-06,
2521
+ "loss": 0.1854,
2522
+ "step": 838
2523
+ },
2524
+ {
2525
+ "epoch": 1.96,
2526
+ "learning_rate": 5.400293439467781e-06,
2527
+ "loss": 0.1921,
2528
+ "step": 840
2529
+ },
2530
+ {
2531
+ "epoch": 1.96,
2532
+ "learning_rate": 5.356931831269798e-06,
2533
+ "loss": 0.1815,
2534
+ "step": 842
2535
+ },
2536
+ {
2537
+ "epoch": 1.97,
2538
+ "learning_rate": 5.313681231940338e-06,
2539
+ "loss": 0.1781,
2540
+ "step": 844
2541
+ },
2542
+ {
2543
+ "epoch": 1.97,
2544
+ "learning_rate": 5.270542675537034e-06,
2545
+ "loss": 0.2022,
2546
+ "step": 846
2547
+ },
2548
+ {
2549
+ "epoch": 1.97,
2550
+ "learning_rate": 5.227517193438746e-06,
2551
+ "loss": 0.1866,
2552
+ "step": 848
2553
+ },
2554
+ {
2555
+ "epoch": 1.98,
2556
+ "learning_rate": 5.184605814320889e-06,
2557
+ "loss": 0.1754,
2558
+ "step": 850
2559
+ },
2560
+ {
2561
+ "epoch": 1.98,
2562
+ "learning_rate": 5.141809564130847e-06,
2563
+ "loss": 0.1745,
2564
+ "step": 852
2565
+ },
2566
+ {
2567
+ "epoch": 1.99,
2568
+ "learning_rate": 5.099129466063444e-06,
2569
+ "loss": 0.1803,
2570
+ "step": 854
2571
+ },
2572
+ {
2573
+ "epoch": 1.99,
2574
+ "learning_rate": 5.056566540536476e-06,
2575
+ "loss": 0.1678,
2576
+ "step": 856
2577
+ },
2578
+ {
2579
+ "epoch": 2.0,
2580
+ "learning_rate": 5.014121805166321e-06,
2581
+ "loss": 0.1702,
2582
+ "step": 858
2583
+ },
2584
+ {
2585
+ "epoch": 2.0,
2586
+ "learning_rate": 4.971796274743601e-06,
2587
+ "loss": 0.1313,
2588
+ "step": 860
2589
+ },
2590
+ {
2591
+ "epoch": 2.01,
2592
+ "learning_rate": 4.9295909612089265e-06,
2593
+ "loss": 0.0643,
2594
+ "step": 862
2595
+ },
2596
+ {
2597
+ "epoch": 2.01,
2598
+ "learning_rate": 4.887506873628708e-06,
2599
+ "loss": 0.0624,
2600
+ "step": 864
2601
+ },
2602
+ {
2603
+ "epoch": 2.02,
2604
+ "learning_rate": 4.845545018171013e-06,
2605
+ "loss": 0.0604,
2606
+ "step": 866
2607
+ },
2608
+ {
2609
+ "epoch": 2.02,
2610
+ "learning_rate": 4.80370639808152e-06,
2611
+ "loss": 0.0648,
2612
+ "step": 868
2613
+ },
2614
+ {
2615
+ "epoch": 2.03,
2616
+ "learning_rate": 4.7619920136595465e-06,
2617
+ "loss": 0.0731,
2618
+ "step": 870
2619
+ },
2620
+ {
2621
+ "epoch": 2.03,
2622
+ "learning_rate": 4.720402862234105e-06,
2623
+ "loss": 0.0582,
2624
+ "step": 872
2625
+ },
2626
+ {
2627
+ "epoch": 2.03,
2628
+ "learning_rate": 4.678939938140079e-06,
2629
+ "loss": 0.0601,
2630
+ "step": 874
2631
+ },
2632
+ {
2633
+ "epoch": 2.04,
2634
+ "learning_rate": 4.637604232694441e-06,
2635
+ "loss": 0.0527,
2636
+ "step": 876
2637
+ },
2638
+ {
2639
+ "epoch": 2.04,
2640
+ "learning_rate": 4.596396734172559e-06,
2641
+ "loss": 0.0575,
2642
+ "step": 878
2643
+ },
2644
+ {
2645
+ "epoch": 2.05,
2646
+ "learning_rate": 4.555318427784561e-06,
2647
+ "loss": 0.0578,
2648
+ "step": 880
2649
+ },
2650
+ {
2651
+ "epoch": 2.05,
2652
+ "learning_rate": 4.514370295651781e-06,
2653
+ "loss": 0.0543,
2654
+ "step": 882
2655
+ },
2656
+ {
2657
+ "epoch": 2.06,
2658
+ "learning_rate": 4.473553316783282e-06,
2659
+ "loss": 0.0547,
2660
+ "step": 884
2661
+ },
2662
+ {
2663
+ "epoch": 2.06,
2664
+ "learning_rate": 4.432868467052449e-06,
2665
+ "loss": 0.053,
2666
+ "step": 886
2667
+ },
2668
+ {
2669
+ "epoch": 2.07,
2670
+ "learning_rate": 4.392316719173651e-06,
2671
+ "loss": 0.0587,
2672
+ "step": 888
2673
+ },
2674
+ {
2675
+ "epoch": 2.07,
2676
+ "learning_rate": 4.351899042678993e-06,
2677
+ "loss": 0.0628,
2678
+ "step": 890
2679
+ },
2680
+ {
2681
+ "epoch": 2.08,
2682
+ "learning_rate": 4.311616403895126e-06,
2683
+ "loss": 0.0582,
2684
+ "step": 892
2685
+ },
2686
+ {
2687
+ "epoch": 2.08,
2688
+ "learning_rate": 4.271469765920163e-06,
2689
+ "loss": 0.0578,
2690
+ "step": 894
2691
+ },
2692
+ {
2693
+ "epoch": 2.09,
2694
+ "learning_rate": 4.231460088600626e-06,
2695
+ "loss": 0.064,
2696
+ "step": 896
2697
+ },
2698
+ {
2699
+ "epoch": 2.09,
2700
+ "learning_rate": 4.191588328508518e-06,
2701
+ "loss": 0.0525,
2702
+ "step": 898
2703
+ },
2704
+ {
2705
+ "epoch": 2.1,
2706
+ "learning_rate": 4.1518554389184416e-06,
2707
+ "loss": 0.0584,
2708
+ "step": 900
2709
+ },
2710
+ {
2711
+ "epoch": 2.1,
2712
+ "learning_rate": 4.1122623697848164e-06,
2713
+ "loss": 0.0621,
2714
+ "step": 902
2715
+ },
2716
+ {
2717
+ "epoch": 2.1,
2718
+ "learning_rate": 4.0728100677191585e-06,
2719
+ "loss": 0.0563,
2720
+ "step": 904
2721
+ },
2722
+ {
2723
+ "epoch": 2.11,
2724
+ "learning_rate": 4.033499475967451e-06,
2725
+ "loss": 0.0598,
2726
+ "step": 906
2727
+ },
2728
+ {
2729
+ "epoch": 2.11,
2730
+ "learning_rate": 3.994331534387602e-06,
2731
+ "loss": 0.0528,
2732
+ "step": 908
2733
+ },
2734
+ {
2735
+ "epoch": 2.12,
2736
+ "learning_rate": 3.95530717942696e-06,
2737
+ "loss": 0.0588,
2738
+ "step": 910
2739
+ },
2740
+ {
2741
+ "epoch": 2.12,
2742
+ "learning_rate": 3.916427344099928e-06,
2743
+ "loss": 0.0668,
2744
+ "step": 912
2745
+ },
2746
+ {
2747
+ "epoch": 2.13,
2748
+ "learning_rate": 3.877692957965663e-06,
2749
+ "loss": 0.0569,
2750
+ "step": 914
2751
+ },
2752
+ {
2753
+ "epoch": 2.13,
2754
+ "learning_rate": 3.839104947105847e-06,
2755
+ "loss": 0.0588,
2756
+ "step": 916
2757
+ },
2758
+ {
2759
+ "epoch": 2.14,
2760
+ "learning_rate": 3.8006642341025456e-06,
2761
+ "loss": 0.0594,
2762
+ "step": 918
2763
+ },
2764
+ {
2765
+ "epoch": 2.14,
2766
+ "learning_rate": 3.762371738016153e-06,
2767
+ "loss": 0.059,
2768
+ "step": 920
2769
+ },
2770
+ {
2771
+ "epoch": 2.15,
2772
+ "learning_rate": 3.72422837436341e-06,
2773
+ "loss": 0.0523,
2774
+ "step": 922
2775
+ },
2776
+ {
2777
+ "epoch": 2.15,
2778
+ "learning_rate": 3.686235055095536e-06,
2779
+ "loss": 0.0538,
2780
+ "step": 924
2781
+ },
2782
+ {
2783
+ "epoch": 2.16,
2784
+ "learning_rate": 3.648392688576401e-06,
2785
+ "loss": 0.0586,
2786
+ "step": 926
2787
+ },
2788
+ {
2789
+ "epoch": 2.16,
2790
+ "learning_rate": 3.610702179560821e-06,
2791
+ "loss": 0.055,
2792
+ "step": 928
2793
+ },
2794
+ {
2795
+ "epoch": 2.17,
2796
+ "learning_rate": 3.573164429172924e-06,
2797
+ "loss": 0.0524,
2798
+ "step": 930
2799
+ },
2800
+ {
2801
+ "epoch": 2.17,
2802
+ "learning_rate": 3.5357803348846087e-06,
2803
+ "loss": 0.0618,
2804
+ "step": 932
2805
+ },
2806
+ {
2807
+ "epoch": 2.17,
2808
+ "learning_rate": 3.498550790494083e-06,
2809
+ "loss": 0.0527,
2810
+ "step": 934
2811
+ },
2812
+ {
2813
+ "epoch": 2.18,
2814
+ "learning_rate": 3.461476686104495e-06,
2815
+ "loss": 0.0541,
2816
+ "step": 936
2817
+ },
2818
+ {
2819
+ "epoch": 2.18,
2820
+ "learning_rate": 3.424558908102653e-06,
2821
+ "loss": 0.0579,
2822
+ "step": 938
2823
+ },
2824
+ {
2825
+ "epoch": 2.19,
2826
+ "learning_rate": 3.387798339137837e-06,
2827
+ "loss": 0.0567,
2828
+ "step": 940
2829
+ },
2830
+ {
2831
+ "epoch": 2.19,
2832
+ "learning_rate": 3.3511958581006874e-06,
2833
+ "loss": 0.0519,
2834
+ "step": 942
2835
+ },
2836
+ {
2837
+ "epoch": 2.2,
2838
+ "learning_rate": 3.314752340102201e-06,
2839
+ "loss": 0.0573,
2840
+ "step": 944
2841
+ },
2842
+ {
2843
+ "epoch": 2.2,
2844
+ "learning_rate": 3.278468656452798e-06,
2845
+ "loss": 0.061,
2846
+ "step": 946
2847
+ },
2848
+ {
2849
+ "epoch": 2.21,
2850
+ "learning_rate": 3.242345674641508e-06,
2851
+ "loss": 0.0611,
2852
+ "step": 948
2853
+ },
2854
+ {
2855
+ "epoch": 2.21,
2856
+ "learning_rate": 3.2063842583152095e-06,
2857
+ "loss": 0.0604,
2858
+ "step": 950
2859
+ },
2860
+ {
2861
+ "epoch": 2.22,
2862
+ "learning_rate": 3.1705852672579853e-06,
2863
+ "loss": 0.0556,
2864
+ "step": 952
2865
+ },
2866
+ {
2867
+ "epoch": 2.22,
2868
+ "learning_rate": 3.134949557370587e-06,
2869
+ "loss": 0.0557,
2870
+ "step": 954
2871
+ },
2872
+ {
2873
+ "epoch": 2.23,
2874
+ "learning_rate": 3.099477980649941e-06,
2875
+ "loss": 0.0539,
2876
+ "step": 956
2877
+ },
2878
+ {
2879
+ "epoch": 2.23,
2880
+ "learning_rate": 3.0641713851687994e-06,
2881
+ "loss": 0.061,
2882
+ "step": 958
2883
+ },
2884
+ {
2885
+ "epoch": 2.24,
2886
+ "learning_rate": 3.0290306150554573e-06,
2887
+ "loss": 0.0566,
2888
+ "step": 960
2889
+ },
2890
+ {
2891
+ "epoch": 2.24,
2892
+ "learning_rate": 2.994056510473571e-06,
2893
+ "loss": 0.0631,
2894
+ "step": 962
2895
+ },
2896
+ {
2897
+ "epoch": 2.24,
2898
+ "learning_rate": 2.959249907602071e-06,
2899
+ "loss": 0.052,
2900
+ "step": 964
2901
+ },
2902
+ {
2903
+ "epoch": 2.25,
2904
+ "learning_rate": 2.9246116386151704e-06,
2905
+ "loss": 0.0553,
2906
+ "step": 966
2907
+ },
2908
+ {
2909
+ "epoch": 2.25,
2910
+ "learning_rate": 2.890142531662471e-06,
2911
+ "loss": 0.0578,
2912
+ "step": 968
2913
+ },
2914
+ {
2915
+ "epoch": 2.26,
2916
+ "learning_rate": 2.8558434108491585e-06,
2917
+ "loss": 0.0522,
2918
+ "step": 970
2919
+ },
2920
+ {
2921
+ "epoch": 2.26,
2922
+ "learning_rate": 2.8217150962163044e-06,
2923
+ "loss": 0.0575,
2924
+ "step": 972
2925
+ },
2926
+ {
2927
+ "epoch": 2.27,
2928
+ "learning_rate": 2.7877584037212555e-06,
2929
+ "loss": 0.0615,
2930
+ "step": 974
2931
+ },
2932
+ {
2933
+ "epoch": 2.27,
2934
+ "learning_rate": 2.75397414521813e-06,
2935
+ "loss": 0.0512,
2936
+ "step": 976
2937
+ },
2938
+ {
2939
+ "epoch": 2.28,
2940
+ "learning_rate": 2.720363128438408e-06,
2941
+ "loss": 0.0595,
2942
+ "step": 978
2943
+ },
2944
+ {
2945
+ "epoch": 2.28,
2946
+ "learning_rate": 2.6869261569716134e-06,
2947
+ "loss": 0.0557,
2948
+ "step": 980
2949
+ },
2950
+ {
2951
+ "epoch": 2.29,
2952
+ "learning_rate": 2.6536640302461036e-06,
2953
+ "loss": 0.0605,
2954
+ "step": 982
2955
+ },
2956
+ {
2957
+ "epoch": 2.29,
2958
+ "learning_rate": 2.6205775435099624e-06,
2959
+ "loss": 0.0548,
2960
+ "step": 984
2961
+ },
2962
+ {
2963
+ "epoch": 2.3,
2964
+ "learning_rate": 2.5876674878119735e-06,
2965
+ "loss": 0.0544,
2966
+ "step": 986
2967
+ },
2968
+ {
2969
+ "epoch": 2.3,
2970
+ "learning_rate": 2.554934649982731e-06,
2971
+ "loss": 0.0566,
2972
+ "step": 988
2973
+ },
2974
+ {
2975
+ "epoch": 2.31,
2976
+ "learning_rate": 2.5223798126158004e-06,
2977
+ "loss": 0.055,
2978
+ "step": 990
2979
+ },
2980
+ {
2981
+ "epoch": 2.31,
2982
+ "learning_rate": 2.490003754049024e-06,
2983
+ "loss": 0.0545,
2984
+ "step": 992
2985
+ },
2986
+ {
2987
+ "epoch": 2.31,
2988
+ "learning_rate": 2.457807248345908e-06,
2989
+ "loss": 0.0611,
2990
+ "step": 994
2991
+ },
2992
+ {
2993
+ "epoch": 2.32,
2994
+ "learning_rate": 2.425791065277119e-06,
2995
+ "loss": 0.0558,
2996
+ "step": 996
2997
+ },
2998
+ {
2999
+ "epoch": 2.32,
3000
+ "learning_rate": 2.393955970302072e-06,
3001
+ "loss": 0.0699,
3002
+ "step": 998
3003
+ },
3004
+ {
3005
+ "epoch": 2.33,
3006
+ "learning_rate": 2.362302724550639e-06,
3007
+ "loss": 0.0589,
3008
+ "step": 1000
3009
+ },
3010
+ {
3011
+ "epoch": 2.33,
3012
+ "learning_rate": 2.3308320848049436e-06,
3013
+ "loss": 0.0584,
3014
+ "step": 1002
3015
+ },
3016
+ {
3017
+ "epoch": 2.34,
3018
+ "learning_rate": 2.299544803481274e-06,
3019
+ "loss": 0.0524,
3020
+ "step": 1004
3021
+ },
3022
+ {
3023
+ "epoch": 2.34,
3024
+ "learning_rate": 2.2684416286120846e-06,
3025
+ "loss": 0.0595,
3026
+ "step": 1006
3027
+ },
3028
+ {
3029
+ "epoch": 2.35,
3030
+ "learning_rate": 2.23752330382813e-06,
3031
+ "loss": 0.0551,
3032
+ "step": 1008
3033
+ },
3034
+ {
3035
+ "epoch": 2.35,
3036
+ "learning_rate": 2.20679056834066e-06,
3037
+ "loss": 0.0523,
3038
+ "step": 1010
3039
+ },
3040
+ {
3041
+ "epoch": 2.36,
3042
+ "learning_rate": 2.176244156923768e-06,
3043
+ "loss": 0.0913,
3044
+ "step": 1012
3045
+ },
3046
+ {
3047
+ "epoch": 2.36,
3048
+ "learning_rate": 2.1458847998968123e-06,
3049
+ "loss": 0.0561,
3050
+ "step": 1014
3051
+ },
3052
+ {
3053
+ "epoch": 2.37,
3054
+ "learning_rate": 2.115713223106959e-06,
3055
+ "loss": 0.0562,
3056
+ "step": 1016
3057
+ },
3058
+ {
3059
+ "epoch": 2.37,
3060
+ "learning_rate": 2.0857301479118276e-06,
3061
+ "loss": 0.0574,
3062
+ "step": 1018
3063
+ },
3064
+ {
3065
+ "epoch": 2.37,
3066
+ "learning_rate": 2.0559362911622438e-06,
3067
+ "loss": 0.0641,
3068
+ "step": 1020
3069
+ },
3070
+ {
3071
+ "epoch": 2.38,
3072
+ "learning_rate": 2.026332365185102e-06,
3073
+ "loss": 0.0569,
3074
+ "step": 1022
3075
+ },
3076
+ {
3077
+ "epoch": 2.38,
3078
+ "learning_rate": 1.996919077766334e-06,
3079
+ "loss": 0.0656,
3080
+ "step": 1024
3081
+ },
3082
+ {
3083
+ "epoch": 2.39,
3084
+ "learning_rate": 1.967697132133981e-06,
3085
+ "loss": 0.0508,
3086
+ "step": 1026
3087
+ },
3088
+ {
3089
+ "epoch": 2.39,
3090
+ "learning_rate": 1.9386672269413976e-06,
3091
+ "loss": 0.0533,
3092
+ "step": 1028
3093
+ },
3094
+ {
3095
+ "epoch": 2.4,
3096
+ "learning_rate": 1.9098300562505266e-06,
3097
+ "loss": 0.0555,
3098
+ "step": 1030
3099
+ },
3100
+ {
3101
+ "epoch": 2.4,
3102
+ "learning_rate": 1.8811863095153182e-06,
3103
+ "loss": 0.0522,
3104
+ "step": 1032
3105
+ },
3106
+ {
3107
+ "epoch": 2.41,
3108
+ "learning_rate": 1.852736671565244e-06,
3109
+ "loss": 0.0541,
3110
+ "step": 1034
3111
+ },
3112
+ {
3113
+ "epoch": 2.41,
3114
+ "learning_rate": 1.8244818225889183e-06,
3115
+ "loss": 0.053,
3116
+ "step": 1036
3117
+ },
3118
+ {
3119
+ "epoch": 2.42,
3120
+ "learning_rate": 1.7964224381178474e-06,
3121
+ "loss": 0.0514,
3122
+ "step": 1038
3123
+ },
3124
+ {
3125
+ "epoch": 2.42,
3126
+ "learning_rate": 1.768559189010267e-06,
3127
+ "loss": 0.0513,
3128
+ "step": 1040
3129
+ },
3130
+ {
3131
+ "epoch": 2.43,
3132
+ "learning_rate": 1.7408927414351051e-06,
3133
+ "loss": 0.0546,
3134
+ "step": 1042
3135
+ },
3136
+ {
3137
+ "epoch": 2.43,
3138
+ "learning_rate": 1.7134237568560619e-06,
3139
+ "loss": 0.0515,
3140
+ "step": 1044
3141
+ },
3142
+ {
3143
+ "epoch": 2.44,
3144
+ "learning_rate": 1.6861528920157877e-06,
3145
+ "loss": 0.0559,
3146
+ "step": 1046
3147
+ },
3148
+ {
3149
+ "epoch": 2.44,
3150
+ "learning_rate": 1.6590807989201841e-06,
3151
+ "loss": 0.0594,
3152
+ "step": 1048
3153
+ },
3154
+ {
3155
+ "epoch": 2.44,
3156
+ "learning_rate": 1.632208124822815e-06,
3157
+ "loss": 0.0527,
3158
+ "step": 1050
3159
+ },
3160
+ {
3161
+ "epoch": 2.45,
3162
+ "learning_rate": 1.6055355122094352e-06,
3163
+ "loss": 0.0503,
3164
+ "step": 1052
3165
+ },
3166
+ {
3167
+ "epoch": 2.45,
3168
+ "learning_rate": 1.579063598782622e-06,
3169
+ "loss": 0.0534,
3170
+ "step": 1054
3171
+ },
3172
+ {
3173
+ "epoch": 2.46,
3174
+ "learning_rate": 1.5527930174465356e-06,
3175
+ "loss": 0.0639,
3176
+ "step": 1056
3177
+ },
3178
+ {
3179
+ "epoch": 2.46,
3180
+ "learning_rate": 1.5267243962917833e-06,
3181
+ "loss": 0.0575,
3182
+ "step": 1058
3183
+ },
3184
+ {
3185
+ "epoch": 2.47,
3186
+ "learning_rate": 1.5008583585804048e-06,
3187
+ "loss": 0.052,
3188
+ "step": 1060
3189
+ },
3190
+ {
3191
+ "epoch": 2.47,
3192
+ "learning_rate": 1.4751955227309722e-06,
3193
+ "loss": 0.0532,
3194
+ "step": 1062
3195
+ },
3196
+ {
3197
+ "epoch": 2.48,
3198
+ "learning_rate": 1.4497365023038012e-06,
3199
+ "loss": 0.0542,
3200
+ "step": 1064
3201
+ },
3202
+ {
3203
+ "epoch": 2.48,
3204
+ "learning_rate": 1.4244819059862824e-06,
3205
+ "loss": 0.0525,
3206
+ "step": 1066
3207
+ },
3208
+ {
3209
+ "epoch": 2.49,
3210
+ "learning_rate": 1.399432337578327e-06,
3211
+ "loss": 0.0588,
3212
+ "step": 1068
3213
+ },
3214
+ {
3215
+ "epoch": 2.49,
3216
+ "learning_rate": 1.3745883959779415e-06,
3217
+ "loss": 0.0552,
3218
+ "step": 1070
3219
+ },
3220
+ {
3221
+ "epoch": 2.5,
3222
+ "learning_rate": 1.3499506751668933e-06,
3223
+ "loss": 0.0535,
3224
+ "step": 1072
3225
+ },
3226
+ {
3227
+ "epoch": 2.5,
3228
+ "learning_rate": 1.325519764196519e-06,
3229
+ "loss": 0.0496,
3230
+ "step": 1074
3231
+ },
3232
+ {
3233
+ "epoch": 2.51,
3234
+ "learning_rate": 1.301296247173638e-06,
3235
+ "loss": 0.0536,
3236
+ "step": 1076
3237
+ },
3238
+ {
3239
+ "epoch": 2.51,
3240
+ "learning_rate": 1.2772807032465895e-06,
3241
+ "loss": 0.0546,
3242
+ "step": 1078
3243
+ },
3244
+ {
3245
+ "epoch": 2.51,
3246
+ "learning_rate": 1.2534737065913839e-06,
3247
+ "loss": 0.062,
3248
+ "step": 1080
3249
+ },
3250
+ {
3251
+ "epoch": 2.52,
3252
+ "learning_rate": 1.229875826397976e-06,
3253
+ "loss": 0.0482,
3254
+ "step": 1082
3255
+ },
3256
+ {
3257
+ "epoch": 2.52,
3258
+ "learning_rate": 1.2064876268566572e-06,
3259
+ "loss": 0.0526,
3260
+ "step": 1084
3261
+ },
3262
+ {
3263
+ "epoch": 2.53,
3264
+ "learning_rate": 1.1833096671445644e-06,
3265
+ "loss": 0.0513,
3266
+ "step": 1086
3267
+ },
3268
+ {
3269
+ "epoch": 2.53,
3270
+ "learning_rate": 1.1603425014123126e-06,
3271
+ "loss": 0.06,
3272
+ "step": 1088
3273
+ },
3274
+ {
3275
+ "epoch": 2.54,
3276
+ "learning_rate": 1.1375866787707435e-06,
3277
+ "loss": 0.0553,
3278
+ "step": 1090
3279
+ },
3280
+ {
3281
+ "epoch": 2.54,
3282
+ "learning_rate": 1.1150427432778078e-06,
3283
+ "loss": 0.0504,
3284
+ "step": 1092
3285
+ },
3286
+ {
3287
+ "epoch": 2.55,
3288
+ "learning_rate": 1.0927112339255374e-06,
3289
+ "loss": 0.0512,
3290
+ "step": 1094
3291
+ },
3292
+ {
3293
+ "epoch": 2.55,
3294
+ "learning_rate": 1.0705926846271787e-06,
3295
+ "loss": 0.05,
3296
+ "step": 1096
3297
+ },
3298
+ {
3299
+ "epoch": 2.56,
3300
+ "learning_rate": 1.0486876242044153e-06,
3301
+ "loss": 0.0577,
3302
+ "step": 1098
3303
+ },
3304
+ {
3305
+ "epoch": 2.56,
3306
+ "learning_rate": 1.0269965763747292e-06,
3307
+ "loss": 0.0533,
3308
+ "step": 1100
3309
+ },
3310
+ {
3311
+ "epoch": 2.57,
3312
+ "learning_rate": 1.0055200597388793e-06,
3313
+ "loss": 0.0556,
3314
+ "step": 1102
3315
+ },
3316
+ {
3317
+ "epoch": 2.57,
3318
+ "learning_rate": 9.84258587768504e-07,
3319
+ "loss": 0.0501,
3320
+ "step": 1104
3321
+ },
3322
+ {
3323
+ "epoch": 2.58,
3324
+ "learning_rate": 9.632126687938392e-07,
3325
+ "loss": 0.0472,
3326
+ "step": 1106
3327
+ },
3328
+ {
3329
+ "epoch": 2.58,
3330
+ "learning_rate": 9.423828059915685e-07,
3331
+ "loss": 0.0637,
3332
+ "step": 1108
3333
+ },
3334
+ {
3335
+ "epoch": 2.58,
3336
+ "learning_rate": 9.217694973728009e-07,
3337
+ "loss": 0.0508,
3338
+ "step": 1110
3339
+ },
3340
+ {
3341
+ "epoch": 2.59,
3342
+ "learning_rate": 9.013732357711469e-07,
3343
+ "loss": 0.0614,
3344
+ "step": 1112
3345
+ },
3346
+ {
3347
+ "epoch": 2.59,
3348
+ "learning_rate": 8.811945088309493e-07,
3349
+ "loss": 0.0534,
3350
+ "step": 1114
3351
+ },
3352
+ {
3353
+ "epoch": 2.6,
3354
+ "learning_rate": 8.612337989956199e-07,
3355
+ "loss": 0.0569,
3356
+ "step": 1116
3357
+ },
3358
+ {
3359
+ "epoch": 2.6,
3360
+ "learning_rate": 8.414915834961035e-07,
3361
+ "loss": 0.053,
3362
+ "step": 1118
3363
+ },
3364
+ {
3365
+ "epoch": 2.61,
3366
+ "learning_rate": 8.219683343394691e-07,
3367
+ "loss": 0.0554,
3368
+ "step": 1120
3369
+ },
3370
+ {
3371
+ "epoch": 2.61,
3372
+ "learning_rate": 8.0266451829763e-07,
3373
+ "loss": 0.0538,
3374
+ "step": 1122
3375
+ },
3376
+ {
3377
+ "epoch": 2.62,
3378
+ "learning_rate": 7.835805968961762e-07,
3379
+ "loss": 0.0529,
3380
+ "step": 1124
3381
+ },
3382
+ {
3383
+ "epoch": 2.62,
3384
+ "learning_rate": 7.647170264033422e-07,
3385
+ "loss": 0.0535,
3386
+ "step": 1126
3387
+ },
3388
+ {
3389
+ "epoch": 2.63,
3390
+ "learning_rate": 7.460742578191016e-07,
3391
+ "loss": 0.0457,
3392
+ "step": 1128
3393
+ },
3394
+ {
3395
+ "epoch": 2.63,
3396
+ "learning_rate": 7.276527368643793e-07,
3397
+ "loss": 0.0531,
3398
+ "step": 1130
3399
+ },
3400
+ {
3401
+ "epoch": 2.64,
3402
+ "learning_rate": 7.094529039704013e-07,
3403
+ "loss": 0.052,
3404
+ "step": 1132
3405
+ },
3406
+ {
3407
+ "epoch": 2.64,
3408
+ "learning_rate": 6.914751942681585e-07,
3409
+ "loss": 0.0527,
3410
+ "step": 1134
3411
+ },
3412
+ {
3413
+ "epoch": 2.64,
3414
+ "learning_rate": 6.737200375780073e-07,
3415
+ "loss": 0.0555,
3416
+ "step": 1136
3417
+ },
3418
+ {
3419
+ "epoch": 2.65,
3420
+ "learning_rate": 6.561878583993897e-07,
3421
+ "loss": 0.0502,
3422
+ "step": 1138
3423
+ },
3424
+ {
3425
+ "epoch": 2.65,
3426
+ "learning_rate": 6.388790759006902e-07,
3427
+ "loss": 0.0502,
3428
+ "step": 1140
3429
+ },
3430
+ {
3431
+ "epoch": 2.66,
3432
+ "learning_rate": 6.217941039092068e-07,
3433
+ "loss": 0.0602,
3434
+ "step": 1142
3435
+ },
3436
+ {
3437
+ "epoch": 2.66,
3438
+ "learning_rate": 6.049333509012611e-07,
3439
+ "loss": 0.0564,
3440
+ "step": 1144
3441
+ },
3442
+ {
3443
+ "epoch": 2.67,
3444
+ "learning_rate": 5.882972199924353e-07,
3445
+ "loss": 0.0578,
3446
+ "step": 1146
3447
+ },
3448
+ {
3449
+ "epoch": 2.67,
3450
+ "learning_rate": 5.718861089279249e-07,
3451
+ "loss": 0.0553,
3452
+ "step": 1148
3453
+ },
3454
+ {
3455
+ "epoch": 2.68,
3456
+ "learning_rate": 5.557004100730357e-07,
3457
+ "loss": 0.0531,
3458
+ "step": 1150
3459
+ },
3460
+ {
3461
+ "epoch": 2.68,
3462
+ "learning_rate": 5.39740510403809e-07,
3463
+ "loss": 0.0499,
3464
+ "step": 1152
3465
+ },
3466
+ {
3467
+ "epoch": 2.69,
3468
+ "learning_rate": 5.240067914977554e-07,
3469
+ "loss": 0.054,
3470
+ "step": 1154
3471
+ },
3472
+ {
3473
+ "epoch": 2.69,
3474
+ "learning_rate": 5.084996295247402e-07,
3475
+ "loss": 0.0512,
3476
+ "step": 1156
3477
+ },
3478
+ {
3479
+ "epoch": 2.7,
3480
+ "learning_rate": 4.932193952379915e-07,
3481
+ "loss": 0.0537,
3482
+ "step": 1158
3483
+ },
3484
+ {
3485
+ "epoch": 2.7,
3486
+ "learning_rate": 4.781664539652319e-07,
3487
+ "loss": 0.0541,
3488
+ "step": 1160
3489
+ },
3490
+ {
3491
+ "epoch": 2.71,
3492
+ "learning_rate": 4.633411655999431e-07,
3493
+ "loss": 0.0515,
3494
+ "step": 1162
3495
+ },
3496
+ {
3497
+ "epoch": 2.71,
3498
+ "learning_rate": 4.487438845927683e-07,
3499
+ "loss": 0.054,
3500
+ "step": 1164
3501
+ },
3502
+ {
3503
+ "epoch": 2.71,
3504
+ "learning_rate": 4.34374959943028e-07,
3505
+ "loss": 0.0516,
3506
+ "step": 1166
3507
+ },
3508
+ {
3509
+ "epoch": 2.72,
3510
+ "learning_rate": 4.202347351903857e-07,
3511
+ "loss": 0.0551,
3512
+ "step": 1168
3513
+ },
3514
+ {
3515
+ "epoch": 2.72,
3516
+ "learning_rate": 4.063235484066275e-07,
3517
+ "loss": 0.0498,
3518
+ "step": 1170
3519
+ },
3520
+ {
3521
+ "epoch": 2.73,
3522
+ "learning_rate": 3.9264173218758083e-07,
3523
+ "loss": 0.0523,
3524
+ "step": 1172
3525
+ },
3526
+ {
3527
+ "epoch": 2.73,
3528
+ "learning_rate": 3.791896136451656e-07,
3529
+ "loss": 0.0535,
3530
+ "step": 1174
3531
+ },
3532
+ {
3533
+ "epoch": 2.74,
3534
+ "learning_rate": 3.6596751439957003e-07,
3535
+ "loss": 0.0515,
3536
+ "step": 1176
3537
+ },
3538
+ {
3539
+ "epoch": 2.74,
3540
+ "learning_rate": 3.5297575057156255e-07,
3541
+ "loss": 0.0642,
3542
+ "step": 1178
3543
+ },
3544
+ {
3545
+ "epoch": 2.75,
3546
+ "learning_rate": 3.4021463277493337e-07,
3547
+ "loss": 0.0576,
3548
+ "step": 1180
3549
+ },
3550
+ {
3551
+ "epoch": 2.75,
3552
+ "learning_rate": 3.2768446610906834e-07,
3553
+ "loss": 0.0512,
3554
+ "step": 1182
3555
+ },
3556
+ {
3557
+ "epoch": 2.76,
3558
+ "learning_rate": 3.153855501516545e-07,
3559
+ "loss": 0.0497,
3560
+ "step": 1184
3561
+ },
3562
+ {
3563
+ "epoch": 2.76,
3564
+ "learning_rate": 3.0331817895151827e-07,
3565
+ "loss": 0.0493,
3566
+ "step": 1186
3567
+ },
3568
+ {
3569
+ "epoch": 2.77,
3570
+ "learning_rate": 2.9148264102159316e-07,
3571
+ "loss": 0.0526,
3572
+ "step": 1188
3573
+ },
3574
+ {
3575
+ "epoch": 2.77,
3576
+ "learning_rate": 2.7987921933202655e-07,
3577
+ "loss": 0.0616,
3578
+ "step": 1190
3579
+ },
3580
+ {
3581
+ "epoch": 2.78,
3582
+ "learning_rate": 2.685081913034082e-07,
3583
+ "loss": 0.0558,
3584
+ "step": 1192
3585
+ },
3586
+ {
3587
+ "epoch": 2.78,
3588
+ "learning_rate": 2.573698288001403e-07,
3589
+ "loss": 0.0537,
3590
+ "step": 1194
3591
+ },
3592
+ {
3593
+ "epoch": 2.78,
3594
+ "learning_rate": 2.46464398123939e-07,
3595
+ "loss": 0.0549,
3596
+ "step": 1196
3597
+ },
3598
+ {
3599
+ "epoch": 2.79,
3600
+ "learning_rate": 2.3579216000746418e-07,
3601
+ "loss": 0.0523,
3602
+ "step": 1198
3603
+ },
3604
+ {
3605
+ "epoch": 2.79,
3606
+ "learning_rate": 2.2535336960809118e-07,
3607
+ "loss": 0.0558,
3608
+ "step": 1200
3609
+ },
3610
+ {
3611
+ "epoch": 2.8,
3612
+ "learning_rate": 2.1514827650180425e-07,
3613
+ "loss": 0.0488,
3614
+ "step": 1202
3615
+ },
3616
+ {
3617
+ "epoch": 2.8,
3618
+ "learning_rate": 2.051771246772305e-07,
3619
+ "loss": 0.052,
3620
+ "step": 1204
3621
+ },
3622
+ {
3623
+ "epoch": 2.81,
3624
+ "learning_rate": 1.954401525298144e-07,
3625
+ "loss": 0.0518,
3626
+ "step": 1206
3627
+ },
3628
+ {
3629
+ "epoch": 2.81,
3630
+ "learning_rate": 1.859375928561058e-07,
3631
+ "loss": 0.0477,
3632
+ "step": 1208
3633
+ },
3634
+ {
3635
+ "epoch": 2.82,
3636
+ "learning_rate": 1.7666967284820202e-07,
3637
+ "loss": 0.055,
3638
+ "step": 1210
3639
+ },
3640
+ {
3641
+ "epoch": 2.82,
3642
+ "learning_rate": 1.6763661408831677e-07,
3643
+ "loss": 0.0534,
3644
+ "step": 1212
3645
+ },
3646
+ {
3647
+ "epoch": 2.83,
3648
+ "learning_rate": 1.5883863254347653e-07,
3649
+ "loss": 0.0522,
3650
+ "step": 1214
3651
+ },
3652
+ {
3653
+ "epoch": 2.83,
3654
+ "learning_rate": 1.5027593856036137e-07,
3655
+ "loss": 0.0518,
3656
+ "step": 1216
3657
+ },
3658
+ {
3659
+ "epoch": 2.84,
3660
+ "learning_rate": 1.4194873686027566e-07,
3661
+ "loss": 0.0484,
3662
+ "step": 1218
3663
+ },
3664
+ {
3665
+ "epoch": 2.84,
3666
+ "learning_rate": 1.3385722653425304e-07,
3667
+ "loss": 0.0571,
3668
+ "step": 1220
3669
+ },
3670
+ {
3671
+ "epoch": 2.85,
3672
+ "learning_rate": 1.2600160103829584e-07,
3673
+ "loss": 0.048,
3674
+ "step": 1222
3675
+ },
3676
+ {
3677
+ "epoch": 2.85,
3678
+ "learning_rate": 1.1838204818874877e-07,
3679
+ "loss": 0.0521,
3680
+ "step": 1224
3681
+ },
3682
+ {
3683
+ "epoch": 2.85,
3684
+ "learning_rate": 1.1099875015781359e-07,
3685
+ "loss": 0.0546,
3686
+ "step": 1226
3687
+ },
3688
+ {
3689
+ "epoch": 2.86,
3690
+ "learning_rate": 1.0385188346918485e-07,
3691
+ "loss": 0.0492,
3692
+ "step": 1228
3693
+ },
3694
+ {
3695
+ "epoch": 2.86,
3696
+ "learning_rate": 9.694161899383992e-08,
3697
+ "loss": 0.0555,
3698
+ "step": 1230
3699
+ },
3700
+ {
3701
+ "epoch": 2.87,
3702
+ "learning_rate": 9.026812194594448e-08,
3703
+ "loss": 0.0494,
3704
+ "step": 1232
3705
+ },
3706
+ {
3707
+ "epoch": 2.87,
3708
+ "learning_rate": 8.383155187890901e-08,
3709
+ "loss": 0.0527,
3710
+ "step": 1234
3711
+ },
3712
+ {
3713
+ "epoch": 2.88,
3714
+ "learning_rate": 7.763206268156964e-08,
3715
+ "loss": 0.0527,
3716
+ "step": 1236
3717
+ },
3718
+ {
3719
+ "epoch": 2.88,
3720
+ "learning_rate": 7.166980257451106e-08,
3721
+ "loss": 0.0618,
3722
+ "step": 1238
3723
+ },
3724
+ {
3725
+ "epoch": 2.89,
3726
+ "learning_rate": 6.594491410652493e-08,
3727
+ "loss": 0.0506,
3728
+ "step": 1240
3729
+ },
3730
+ {
3731
+ "epoch": 2.89,
3732
+ "learning_rate": 6.045753415119593e-08,
3733
+ "loss": 0.0577,
3734
+ "step": 1242
3735
+ },
3736
+ {
3737
+ "epoch": 2.9,
3738
+ "learning_rate": 5.520779390363551e-08,
3739
+ "loss": 0.0495,
3740
+ "step": 1244
3741
+ },
3742
+ {
3743
+ "epoch": 2.9,
3744
+ "learning_rate": 5.019581887733993e-08,
3745
+ "loss": 0.0563,
3746
+ "step": 1246
3747
+ },
3748
+ {
3749
+ "epoch": 2.91,
3750
+ "learning_rate": 4.542172890119267e-08,
3751
+ "loss": 0.0556,
3752
+ "step": 1248
3753
+ },
3754
+ {
3755
+ "epoch": 2.91,
3756
+ "learning_rate": 4.0885638116601176e-08,
3757
+ "loss": 0.0518,
3758
+ "step": 1250
3759
+ },
3760
+ {
3761
+ "epoch": 2.92,
3762
+ "learning_rate": 3.6587654974761246e-08,
3763
+ "loss": 0.0517,
3764
+ "step": 1252
3765
+ },
3766
+ {
3767
+ "epoch": 2.92,
3768
+ "learning_rate": 3.252788223407244e-08,
3769
+ "loss": 0.0531,
3770
+ "step": 1254
3771
+ },
3772
+ {
3773
+ "epoch": 2.92,
3774
+ "learning_rate": 2.870641695767451e-08,
3775
+ "loss": 0.0525,
3776
+ "step": 1256
3777
+ },
3778
+ {
3779
+ "epoch": 2.93,
3780
+ "learning_rate": 2.5123350511129242e-08,
3781
+ "loss": 0.0609,
3782
+ "step": 1258
3783
+ },
3784
+ {
3785
+ "epoch": 2.93,
3786
+ "learning_rate": 2.177876856023997e-08,
3787
+ "loss": 0.0578,
3788
+ "step": 1260
3789
+ },
3790
+ {
3791
+ "epoch": 2.94,
3792
+ "learning_rate": 1.8672751068995464e-08,
3793
+ "loss": 0.0544,
3794
+ "step": 1262
3795
+ },
3796
+ {
3797
+ "epoch": 2.94,
3798
+ "learning_rate": 1.5805372297662546e-08,
3799
+ "loss": 0.0557,
3800
+ "step": 1264
3801
+ },
3802
+ {
3803
+ "epoch": 2.95,
3804
+ "learning_rate": 1.3176700801014186e-08,
3805
+ "loss": 0.0509,
3806
+ "step": 1266
3807
+ },
3808
+ {
3809
+ "epoch": 2.95,
3810
+ "learning_rate": 1.0786799426683037e-08,
3811
+ "loss": 0.0564,
3812
+ "step": 1268
3813
+ },
3814
+ {
3815
+ "epoch": 2.96,
3816
+ "learning_rate": 8.635725313663745e-09,
3817
+ "loss": 0.055,
3818
+ "step": 1270
3819
+ },
3820
+ {
3821
+ "epoch": 2.96,
3822
+ "learning_rate": 6.723529890946268e-09,
3823
+ "loss": 0.0494,
3824
+ "step": 1272
3825
+ },
3826
+ {
3827
+ "epoch": 2.97,
3828
+ "learning_rate": 5.05025887628352e-09,
3829
+ "loss": 0.0501,
3830
+ "step": 1274
3831
+ },
3832
+ {
3833
+ "epoch": 2.97,
3834
+ "learning_rate": 3.615952275104473e-09,
3835
+ "loss": 0.0505,
3836
+ "step": 1276
3837
+ },
3838
+ {
3839
+ "epoch": 2.98,
3840
+ "learning_rate": 2.420644379549364e-09,
3841
+ "loss": 0.0515,
3842
+ "step": 1278
3843
+ },
3844
+ {
3845
+ "epoch": 2.98,
3846
+ "learning_rate": 1.4643637676559074e-09,
3847
+ "loss": 0.2096,
3848
+ "step": 1280
3849
+ },
3850
+ {
3851
+ "epoch": 2.98,
3852
+ "learning_rate": 7.471333026742856e-10,
3853
+ "loss": 0.0539,
3854
+ "step": 1282
3855
+ },
3856
+ {
3857
+ "epoch": 2.99,
3858
+ "learning_rate": 2.689701325209182e-10,
3859
+ "loss": 0.0536,
3860
+ "step": 1284
3861
+ },
3862
+ {
3863
+ "epoch": 2.99,
3864
+ "learning_rate": 2.988568936768132e-11,
3865
+ "loss": 0.0512,
3866
+ "step": 1286
3867
+ }
3868
+ ],
3869
+ "max_steps": 1287,
3870
+ "num_train_epochs": 3,
3871
+ "total_flos": 999035450556416.0,
3872
+ "trial_name": null,
3873
+ "trial_params": null
3874
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1aa0c429a2a24dffa8e1b675d54e650f542f904e47dfea597b9ba0b965212f79
3
+ size 5115
zero_to_fp32.py ADDED
@@ -0,0 +1,584 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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 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
+ # record shared parameters so that they can be recovered based on partners
124
+ # this is because such parameters holding reference only are not saved by optimizer
125
+ shared_params = []
126
+ for param in state_dict["module"]:
127
+ if param not in [*param_names, *buffer_names]:
128
+ for share_param in state_dict["module"]:
129
+ if (state_dict["module"][share_param].data_ptr() == state_dict["module"][param].data_ptr()
130
+ and share_param != param):
131
+ shared_params.append([param, share_param])
132
+ break
133
+
134
+ ds_version = state_dict.get(DS_VERSION, None)
135
+
136
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
137
+
138
+ z_model_state = zero_model_state(buffers=buffers,
139
+ param_shapes=param_shapes,
140
+ shared_params=shared_params,
141
+ ds_version=ds_version,
142
+ frozen_param_shapes=frozen_param_shapes,
143
+ frozen_param_fragments=frozen_param_fragments)
144
+ zero_model_states.append(z_model_state)
145
+
146
+ return zero_model_states
147
+
148
+
149
+ def parse_optim_states(files, ds_checkpoint_dir):
150
+
151
+ total_files = len(files)
152
+ state_dicts = []
153
+ for f in files:
154
+ state_dicts.append(torch.load(f, map_location=device))
155
+
156
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
157
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
158
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
159
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
160
+
161
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
162
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
163
+ # use the max of the partition_count to get the dp world_size.
164
+
165
+ if type(world_size) is list:
166
+ world_size = max(world_size)
167
+
168
+ if world_size != total_files:
169
+ raise ValueError(
170
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
171
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
172
+ )
173
+
174
+ # the groups are named differently in each stage
175
+ if zero_stage == 2:
176
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
177
+ elif zero_stage == 3:
178
+ fp32_groups_key = FP32_FLAT_GROUPS
179
+ else:
180
+ raise ValueError(f"unknown zero stage {zero_stage}")
181
+
182
+ if zero_stage == 2:
183
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
184
+ elif zero_stage == 3:
185
+ # if there is more than one param group, there will be multiple flattened tensors - one
186
+ # flattened tensor per group - for simplicity merge them into a single tensor
187
+ #
188
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
189
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
190
+
191
+ fp32_flat_groups = [
192
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
193
+ ]
194
+
195
+ return zero_stage, world_size, fp32_flat_groups
196
+
197
+
198
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
199
+ """
200
+ Returns fp32 state_dict reconstructed from ds checkpoint
201
+
202
+ Args:
203
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
204
+
205
+ """
206
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
207
+
208
+ optim_files = get_optim_files(ds_checkpoint_dir)
209
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
210
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
211
+
212
+ model_files = get_model_state_files(ds_checkpoint_dir)
213
+
214
+ zero_model_states = parse_model_states(model_files)
215
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
216
+
217
+ if zero_stage == 2:
218
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
219
+ elif zero_stage == 3:
220
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
221
+
222
+
223
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
224
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
225
+ return
226
+
227
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
228
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
229
+
230
+ if debug:
231
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
232
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
233
+
234
+ wanted_params = len(frozen_param_shapes)
235
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
236
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
237
+ print(f'Frozen params: Have {avail_numel} numels to process.')
238
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
239
+
240
+ total_params = 0
241
+ total_numel = 0
242
+ for name, shape in frozen_param_shapes.items():
243
+ total_params += 1
244
+ unpartitioned_numel = shape.numel()
245
+ total_numel += unpartitioned_numel
246
+
247
+ state_dict[name] = frozen_param_fragments[name]
248
+
249
+ if debug:
250
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
251
+
252
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
253
+
254
+
255
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
256
+ param_shapes = zero_model_states[0].param_shapes
257
+
258
+ # Reconstruction protocol:
259
+ #
260
+ # XXX: document this
261
+
262
+ if debug:
263
+ for i in range(world_size):
264
+ for j in range(len(fp32_flat_groups[0])):
265
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
266
+
267
+ # XXX: memory usage doubles here (zero2)
268
+ num_param_groups = len(fp32_flat_groups[0])
269
+ merged_single_partition_of_fp32_groups = []
270
+ for i in range(num_param_groups):
271
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
272
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
273
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
274
+ avail_numel = sum(
275
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
276
+
277
+ if debug:
278
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
279
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
280
+ # not asserting if there is a mismatch due to possible padding
281
+ print(f"Have {avail_numel} numels to process.")
282
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
283
+
284
+ # params
285
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
286
+ # out-of-core computing solution
287
+ total_numel = 0
288
+ total_params = 0
289
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
290
+ offset = 0
291
+ avail_numel = full_single_fp32_vector.numel()
292
+ for name, shape in shapes.items():
293
+
294
+ unpartitioned_numel = shape.numel()
295
+ total_numel += unpartitioned_numel
296
+ total_params += 1
297
+
298
+ if debug:
299
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
300
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
301
+ offset += unpartitioned_numel
302
+
303
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
304
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
305
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
306
+ # live optimizer object, so we are checking that the numbers are within the right range
307
+ align_to = 2 * world_size
308
+
309
+ def zero2_align(x):
310
+ return align_to * math.ceil(x / align_to)
311
+
312
+ if debug:
313
+ print(f"original offset={offset}, avail_numel={avail_numel}")
314
+
315
+ offset = zero2_align(offset)
316
+ avail_numel = zero2_align(avail_numel)
317
+
318
+ if debug:
319
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
320
+
321
+ # Sanity check
322
+ if offset != avail_numel:
323
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
324
+
325
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
326
+
327
+
328
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
329
+ state_dict = OrderedDict()
330
+
331
+ # buffers
332
+ buffers = zero_model_states[0].buffers
333
+ state_dict.update(buffers)
334
+ if debug:
335
+ print(f"added {len(buffers)} buffers")
336
+
337
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
338
+
339
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
340
+
341
+ # recover shared parameters
342
+ for pair in zero_model_states[0].shared_params:
343
+ state_dict[pair[0]] = state_dict[pair[1]]
344
+
345
+ return state_dict
346
+
347
+
348
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
349
+ remainder = unpartitioned_numel % world_size
350
+ padding_numel = (world_size - remainder) if remainder else 0
351
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
352
+ return partitioned_numel, padding_numel
353
+
354
+
355
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
356
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
357
+ return
358
+
359
+ if debug:
360
+ for i in range(world_size):
361
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
362
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
363
+
364
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
365
+ wanted_params = len(frozen_param_shapes)
366
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
367
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
368
+ print(f'Frozen params: Have {avail_numel} numels to process.')
369
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
370
+
371
+ total_params = 0
372
+ total_numel = 0
373
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
374
+ total_params += 1
375
+ unpartitioned_numel = shape.numel()
376
+ total_numel += unpartitioned_numel
377
+
378
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
379
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
380
+
381
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
382
+
383
+ if debug:
384
+ print(
385
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
386
+ )
387
+
388
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
389
+
390
+
391
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
392
+ param_shapes = zero_model_states[0].param_shapes
393
+ avail_numel = fp32_flat_groups[0].numel() * world_size
394
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
395
+ # param, re-consolidating each param, while dealing with padding if any
396
+
397
+ # merge list of dicts, preserving order
398
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
399
+
400
+ if debug:
401
+ for i in range(world_size):
402
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
403
+
404
+ wanted_params = len(param_shapes)
405
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
406
+ # not asserting if there is a mismatch due to possible padding
407
+ avail_numel = fp32_flat_groups[0].numel() * world_size
408
+ print(f"Trainable params: Have {avail_numel} numels to process.")
409
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
410
+
411
+ # params
412
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
413
+ # out-of-core computing solution
414
+ offset = 0
415
+ total_numel = 0
416
+ total_params = 0
417
+ for name, shape in param_shapes.items():
418
+
419
+ unpartitioned_numel = shape.numel()
420
+ total_numel += unpartitioned_numel
421
+ total_params += 1
422
+
423
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
424
+
425
+ if debug:
426
+ print(
427
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
428
+ )
429
+
430
+ # XXX: memory usage doubles here
431
+ state_dict[name] = torch.cat(
432
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
433
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
434
+ offset += partitioned_numel
435
+
436
+ offset *= world_size
437
+
438
+ # Sanity check
439
+ if offset != avail_numel:
440
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
441
+
442
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
443
+
444
+
445
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
446
+ state_dict = OrderedDict()
447
+
448
+ # buffers
449
+ buffers = zero_model_states[0].buffers
450
+ state_dict.update(buffers)
451
+ if debug:
452
+ print(f"added {len(buffers)} buffers")
453
+
454
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
455
+
456
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
457
+
458
+ # recover shared parameters
459
+ for pair in zero_model_states[0].shared_params:
460
+ state_dict[pair[0]] = state_dict[pair[1]]
461
+
462
+ return state_dict
463
+
464
+
465
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
466
+ """
467
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
468
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
469
+ via a model hub.
470
+
471
+ Args:
472
+ - ``checkpoint_dir``: path to the desired checkpoint folder
473
+ - ``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``
474
+
475
+ Returns:
476
+ - pytorch ``state_dict``
477
+
478
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
479
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
480
+ the checkpoint.
481
+
482
+ A typical usage might be ::
483
+
484
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
485
+ # do the training and checkpoint saving
486
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
487
+ model = model.cpu() # move to cpu
488
+ model.load_state_dict(state_dict)
489
+ # submit to model hub or save the model to share with others
490
+
491
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
492
+ application. i.e. you will need to re-initialize the deepspeed engine, since
493
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
494
+
495
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
496
+
497
+ """
498
+ if tag is None:
499
+ latest_path = os.path.join(checkpoint_dir, 'latest')
500
+ if os.path.isfile(latest_path):
501
+ with open(latest_path, 'r') as fd:
502
+ tag = fd.read().strip()
503
+ else:
504
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
505
+
506
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
507
+
508
+ if not os.path.isdir(ds_checkpoint_dir):
509
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
510
+
511
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
512
+
513
+
514
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
515
+ """
516
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
517
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
518
+
519
+ Args:
520
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
521
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
522
+ - ``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``
523
+ """
524
+
525
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
526
+ print(f"Saving fp32 state dict to {output_file}")
527
+ torch.save(state_dict, output_file)
528
+
529
+
530
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
531
+ """
532
+ 1. Put the provided model to cpu
533
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
534
+ 3. Load it into the provided model
535
+
536
+ Args:
537
+ - ``model``: the model object to update
538
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
539
+ - ``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``
540
+
541
+ Returns:
542
+ - ``model`: modified model
543
+
544
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
545
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
546
+ conveniently placed for you in the checkpoint folder.
547
+
548
+ A typical usage might be ::
549
+
550
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
551
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
552
+ # submit to model hub or save the model to share with others
553
+
554
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
555
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
556
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
557
+
558
+ """
559
+ logger.info(f"Extracting fp32 weights")
560
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
561
+
562
+ logger.info(f"Overwriting model with fp32 weights")
563
+ model = model.cpu()
564
+ model.load_state_dict(state_dict, strict=False)
565
+
566
+ return model
567
+
568
+
569
+ if __name__ == "__main__":
570
+
571
+ parser = argparse.ArgumentParser()
572
+ parser.add_argument("checkpoint_dir",
573
+ type=str,
574
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
575
+ parser.add_argument(
576
+ "output_file",
577
+ type=str,
578
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
579
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
580
+ args = parser.parse_args()
581
+
582
+ debug = args.debug
583
+
584
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file)