RayDu0010 commited on
Commit
32a17a4
·
verified ·
1 Parent(s): eb7b107

Upload folder using huggingface_hub

Browse files
132_128_e3_3e-5/adapter_config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "ibm-granite/granite-3.3-8b-base",
5
+ "bias": "none",
6
+ "corda_config": null,
7
+ "eva_config": null,
8
+ "exclude_modules": null,
9
+ "fan_in_fan_out": false,
10
+ "inference_mode": true,
11
+ "init_lora_weights": true,
12
+ "layer_replication": null,
13
+ "layers_pattern": null,
14
+ "layers_to_transform": null,
15
+ "loftq_config": {},
16
+ "lora_alpha": 256,
17
+ "lora_bias": false,
18
+ "lora_dropout": 0.05,
19
+ "megatron_config": null,
20
+ "megatron_core": "megatron.core",
21
+ "modules_to_save": null,
22
+ "peft_type": "LORA",
23
+ "r": 128,
24
+ "rank_pattern": {},
25
+ "revision": null,
26
+ "target_modules": [
27
+ "o_proj",
28
+ "gate_proj",
29
+ "down_proj",
30
+ "k_proj",
31
+ "q_proj",
32
+ "v_proj",
33
+ "up_proj"
34
+ ],
35
+ "task_type": "CAUSAL_LM",
36
+ "trainable_token_indices": null,
37
+ "use_dora": false,
38
+ "use_rslora": false
39
+ }
132_128_e3_3e-5/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30fd49e81df26b1e6a11b788c07d30f05139dae7a77b955781046c7fd57550b4
3
+ size 791751704
132_128_e3_3e-5/latest ADDED
@@ -0,0 +1 @@
 
 
1
+ global_step2094
132_128_e3_3e-5/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
132_128_e3_3e-5/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80c26ff3e2db01bc288bdd685f222fb976ba73ce8792a53ea587c792fc7aaed4
3
+ size 16389
132_128_e3_3e-5/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79929ff63a7d73a15239ddd6869193270042e58e4c92ebc8170713f5d5dac123
3
+ size 16389
132_128_e3_3e-5/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be5e2536ee0b4c0f434da2049e38e7eae9f232685d1ade50b55a0d6ac3a24c2b
3
+ size 16389
132_128_e3_3e-5/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9682d4fc0af3cdab26e1c06715705bedc990e716999f3cad86f3aeed41af786
3
+ size 16389
132_128_e3_3e-5/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1b3f7337adc072ad351c06d46aa0ce4bbed520bc2a01a734930533126ce31b8
3
+ size 16389
132_128_e3_3e-5/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97b98ef5c542e1c6543eccd5c9f3f3990bcf6b48ab06ceed0c0a95eea1236778
3
+ size 16389
132_128_e3_3e-5/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ac6fce76303fec3a7599cba08202e751fb41cf347132a78ec283b1eca0fdb2b
3
+ size 16389
132_128_e3_3e-5/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d1c30a3f9f3644a226699eda27fcaa5d7e920adeb6aa639e9ad00c4578649ed
3
+ size 16389
132_128_e3_3e-5/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2e7acce1660f836ca34b02346b9caba1ed12b9b5763d0e885b1d3bb2646c001b
3
+ size 1401
132_128_e3_3e-5/special_tokens_map.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|endoftext|>",
4
+ "<fim_prefix>",
5
+ "<fim_middle>",
6
+ "<fim_suffix>",
7
+ "<fim_pad>",
8
+ "<filename>",
9
+ "<gh_stars>",
10
+ "<issue_start>",
11
+ "<issue_comment>",
12
+ "<issue_closed>",
13
+ "<jupyter_start>",
14
+ "<jupyter_text>",
15
+ "<jupyter_code>",
16
+ "<jupyter_output>",
17
+ "<empty_output>",
18
+ "<commit_before>",
19
+ "<commit_msg>",
20
+ "<commit_after>",
21
+ "<reponame>"
22
+ ],
23
+ "bos_token": {
24
+ "content": "<|endoftext|>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "eos_token": {
31
+ "content": "<|endoftext|>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ },
37
+ "pad_token": "<reponame>",
38
+ "unk_token": {
39
+ "content": "<|endoftext|>",
40
+ "lstrip": false,
41
+ "normalized": false,
42
+ "rstrip": false,
43
+ "single_word": false
44
+ }
45
+ }
132_128_e3_3e-5/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
132_128_e3_3e-5/tokenizer_config.json ADDED
@@ -0,0 +1,188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "0": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": false,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ },
12
+ "1": {
13
+ "content": "<fim_prefix>",
14
+ "lstrip": false,
15
+ "normalized": false,
16
+ "rstrip": false,
17
+ "single_word": false,
18
+ "special": true
19
+ },
20
+ "2": {
21
+ "content": "<fim_middle>",
22
+ "lstrip": false,
23
+ "normalized": false,
24
+ "rstrip": false,
25
+ "single_word": false,
26
+ "special": true
27
+ },
28
+ "3": {
29
+ "content": "<fim_suffix>",
30
+ "lstrip": false,
31
+ "normalized": false,
32
+ "rstrip": false,
33
+ "single_word": false,
34
+ "special": true
35
+ },
36
+ "4": {
37
+ "content": "<fim_pad>",
38
+ "lstrip": false,
39
+ "normalized": false,
40
+ "rstrip": false,
41
+ "single_word": false,
42
+ "special": true
43
+ },
44
+ "5": {
45
+ "content": "<filename>",
46
+ "lstrip": false,
47
+ "normalized": false,
48
+ "rstrip": false,
49
+ "single_word": false,
50
+ "special": true
51
+ },
52
+ "6": {
53
+ "content": "<gh_stars>",
54
+ "lstrip": false,
55
+ "normalized": false,
56
+ "rstrip": false,
57
+ "single_word": false,
58
+ "special": true
59
+ },
60
+ "7": {
61
+ "content": "<issue_start>",
62
+ "lstrip": false,
63
+ "normalized": false,
64
+ "rstrip": false,
65
+ "single_word": false,
66
+ "special": true
67
+ },
68
+ "8": {
69
+ "content": "<issue_comment>",
70
+ "lstrip": false,
71
+ "normalized": false,
72
+ "rstrip": false,
73
+ "single_word": false,
74
+ "special": true
75
+ },
76
+ "9": {
77
+ "content": "<issue_closed>",
78
+ "lstrip": false,
79
+ "normalized": false,
80
+ "rstrip": false,
81
+ "single_word": false,
82
+ "special": true
83
+ },
84
+ "10": {
85
+ "content": "<jupyter_start>",
86
+ "lstrip": false,
87
+ "normalized": false,
88
+ "rstrip": false,
89
+ "single_word": false,
90
+ "special": true
91
+ },
92
+ "11": {
93
+ "content": "<jupyter_text>",
94
+ "lstrip": false,
95
+ "normalized": false,
96
+ "rstrip": false,
97
+ "single_word": false,
98
+ "special": true
99
+ },
100
+ "12": {
101
+ "content": "<jupyter_code>",
102
+ "lstrip": false,
103
+ "normalized": false,
104
+ "rstrip": false,
105
+ "single_word": false,
106
+ "special": true
107
+ },
108
+ "13": {
109
+ "content": "<jupyter_output>",
110
+ "lstrip": false,
111
+ "normalized": false,
112
+ "rstrip": false,
113
+ "single_word": false,
114
+ "special": true
115
+ },
116
+ "14": {
117
+ "content": "<empty_output>",
118
+ "lstrip": false,
119
+ "normalized": false,
120
+ "rstrip": false,
121
+ "single_word": false,
122
+ "special": true
123
+ },
124
+ "15": {
125
+ "content": "<commit_before>",
126
+ "lstrip": false,
127
+ "normalized": false,
128
+ "rstrip": false,
129
+ "single_word": false,
130
+ "special": true
131
+ },
132
+ "16": {
133
+ "content": "<commit_msg>",
134
+ "lstrip": false,
135
+ "normalized": false,
136
+ "rstrip": false,
137
+ "single_word": false,
138
+ "special": true
139
+ },
140
+ "17": {
141
+ "content": "<commit_after>",
142
+ "lstrip": false,
143
+ "normalized": false,
144
+ "rstrip": false,
145
+ "single_word": false,
146
+ "special": true
147
+ },
148
+ "18": {
149
+ "content": "<reponame>",
150
+ "lstrip": false,
151
+ "normalized": false,
152
+ "rstrip": false,
153
+ "single_word": false,
154
+ "special": true
155
+ }
156
+ },
157
+ "additional_special_tokens": [
158
+ "<|endoftext|>",
159
+ "<fim_prefix>",
160
+ "<fim_middle>",
161
+ "<fim_suffix>",
162
+ "<fim_pad>",
163
+ "<filename>",
164
+ "<gh_stars>",
165
+ "<issue_start>",
166
+ "<issue_comment>",
167
+ "<issue_closed>",
168
+ "<jupyter_start>",
169
+ "<jupyter_text>",
170
+ "<jupyter_code>",
171
+ "<jupyter_output>",
172
+ "<empty_output>",
173
+ "<commit_before>",
174
+ "<commit_msg>",
175
+ "<commit_after>",
176
+ "<reponame>"
177
+ ],
178
+ "bos_token": "<|endoftext|>",
179
+ "clean_up_tokenization_spaces": true,
180
+ "eos_token": "<|endoftext|>",
181
+ "extra_special_tokens": {},
182
+ "model_max_length": 8192,
183
+ "pad_token": "<reponame>",
184
+ "padding_side": "left",
185
+ "tokenizer_class": "GPT2Tokenizer",
186
+ "unk_token": "<|endoftext|>",
187
+ "vocab_size": 49152
188
+ }
132_128_e3_3e-5/trainer_state.json ADDED
@@ -0,0 +1,2960 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 3.0,
6
+ "eval_steps": 500,
7
+ "global_step": 2094,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.0071633237822349575,
14
+ "grad_norm": 1.03425133228302,
15
+ "learning_rate": 1.142857142857143e-06,
16
+ "loss": 1.4229,
17
+ "step": 5
18
+ },
19
+ {
20
+ "epoch": 0.014326647564469915,
21
+ "grad_norm": 0.7657381892204285,
22
+ "learning_rate": 2.5714285714285716e-06,
23
+ "loss": 1.3634,
24
+ "step": 10
25
+ },
26
+ {
27
+ "epoch": 0.02148997134670487,
28
+ "grad_norm": 0.7649133801460266,
29
+ "learning_rate": 4e-06,
30
+ "loss": 1.3897,
31
+ "step": 15
32
+ },
33
+ {
34
+ "epoch": 0.02865329512893983,
35
+ "grad_norm": 0.528989315032959,
36
+ "learning_rate": 5.428571428571429e-06,
37
+ "loss": 1.2828,
38
+ "step": 20
39
+ },
40
+ {
41
+ "epoch": 0.03581661891117478,
42
+ "grad_norm": 0.5923161506652832,
43
+ "learning_rate": 6.857142857142857e-06,
44
+ "loss": 1.3395,
45
+ "step": 25
46
+ },
47
+ {
48
+ "epoch": 0.04297994269340974,
49
+ "grad_norm": 0.5814239382743835,
50
+ "learning_rate": 8.285714285714287e-06,
51
+ "loss": 1.3348,
52
+ "step": 30
53
+ },
54
+ {
55
+ "epoch": 0.050143266475644696,
56
+ "grad_norm": 0.551114022731781,
57
+ "learning_rate": 9.714285714285715e-06,
58
+ "loss": 1.3368,
59
+ "step": 35
60
+ },
61
+ {
62
+ "epoch": 0.05730659025787966,
63
+ "grad_norm": 0.5624629259109497,
64
+ "learning_rate": 1.1142857142857143e-05,
65
+ "loss": 1.3196,
66
+ "step": 40
67
+ },
68
+ {
69
+ "epoch": 0.06446991404011461,
70
+ "grad_norm": 0.5845000743865967,
71
+ "learning_rate": 1.2571428571428572e-05,
72
+ "loss": 1.2878,
73
+ "step": 45
74
+ },
75
+ {
76
+ "epoch": 0.07163323782234957,
77
+ "grad_norm": 0.5532830953598022,
78
+ "learning_rate": 1.4e-05,
79
+ "loss": 1.2995,
80
+ "step": 50
81
+ },
82
+ {
83
+ "epoch": 0.07879656160458452,
84
+ "grad_norm": 0.5344447493553162,
85
+ "learning_rate": 1.5428571428571428e-05,
86
+ "loss": 1.2846,
87
+ "step": 55
88
+ },
89
+ {
90
+ "epoch": 0.08595988538681948,
91
+ "grad_norm": 0.5201062560081482,
92
+ "learning_rate": 1.6857142857142858e-05,
93
+ "loss": 1.2614,
94
+ "step": 60
95
+ },
96
+ {
97
+ "epoch": 0.09312320916905444,
98
+ "grad_norm": 0.5084472894668579,
99
+ "learning_rate": 1.8285714285714288e-05,
100
+ "loss": 1.2504,
101
+ "step": 65
102
+ },
103
+ {
104
+ "epoch": 0.10028653295128939,
105
+ "grad_norm": 0.5360846519470215,
106
+ "learning_rate": 1.9714285714285714e-05,
107
+ "loss": 1.2625,
108
+ "step": 70
109
+ },
110
+ {
111
+ "epoch": 0.10744985673352435,
112
+ "grad_norm": 0.5937910676002502,
113
+ "learning_rate": 2.1142857142857144e-05,
114
+ "loss": 1.2096,
115
+ "step": 75
116
+ },
117
+ {
118
+ "epoch": 0.11461318051575932,
119
+ "grad_norm": 0.5604082345962524,
120
+ "learning_rate": 2.257142857142857e-05,
121
+ "loss": 1.2617,
122
+ "step": 80
123
+ },
124
+ {
125
+ "epoch": 0.12177650429799428,
126
+ "grad_norm": 0.5827813744544983,
127
+ "learning_rate": 2.4e-05,
128
+ "loss": 1.2143,
129
+ "step": 85
130
+ },
131
+ {
132
+ "epoch": 0.12893982808022922,
133
+ "grad_norm": 0.6254940032958984,
134
+ "learning_rate": 2.5428571428571427e-05,
135
+ "loss": 1.2306,
136
+ "step": 90
137
+ },
138
+ {
139
+ "epoch": 0.1361031518624642,
140
+ "grad_norm": 0.614905595779419,
141
+ "learning_rate": 2.685714285714286e-05,
142
+ "loss": 1.2366,
143
+ "step": 95
144
+ },
145
+ {
146
+ "epoch": 0.14326647564469913,
147
+ "grad_norm": 0.6369166374206543,
148
+ "learning_rate": 2.8285714285714287e-05,
149
+ "loss": 1.2248,
150
+ "step": 100
151
+ },
152
+ {
153
+ "epoch": 0.1504297994269341,
154
+ "grad_norm": 0.6786860823631287,
155
+ "learning_rate": 2.9714285714285717e-05,
156
+ "loss": 1.1431,
157
+ "step": 105
158
+ },
159
+ {
160
+ "epoch": 0.15759312320916904,
161
+ "grad_norm": 0.5922759175300598,
162
+ "learning_rate": 2.9999700628825923e-05,
163
+ "loss": 1.1399,
164
+ "step": 110
165
+ },
166
+ {
167
+ "epoch": 0.16475644699140402,
168
+ "grad_norm": 0.6088710427284241,
169
+ "learning_rate": 2.9998484453911492e-05,
170
+ "loss": 1.1571,
171
+ "step": 115
172
+ },
173
+ {
174
+ "epoch": 0.17191977077363896,
175
+ "grad_norm": 0.6352574229240417,
176
+ "learning_rate": 2.9996332840351433e-05,
177
+ "loss": 1.1742,
178
+ "step": 120
179
+ },
180
+ {
181
+ "epoch": 0.17908309455587393,
182
+ "grad_norm": 0.5774803161621094,
183
+ "learning_rate": 2.9993245922339485e-05,
184
+ "loss": 1.107,
185
+ "step": 125
186
+ },
187
+ {
188
+ "epoch": 0.18624641833810887,
189
+ "grad_norm": 0.733860194683075,
190
+ "learning_rate": 2.9989223892403244e-05,
191
+ "loss": 1.1378,
192
+ "step": 130
193
+ },
194
+ {
195
+ "epoch": 0.19340974212034384,
196
+ "grad_norm": 0.8307927250862122,
197
+ "learning_rate": 2.9984267001392217e-05,
198
+ "loss": 1.1422,
199
+ "step": 135
200
+ },
201
+ {
202
+ "epoch": 0.20057306590257878,
203
+ "grad_norm": 0.740251898765564,
204
+ "learning_rate": 2.9978375558462126e-05,
205
+ "loss": 1.0947,
206
+ "step": 140
207
+ },
208
+ {
209
+ "epoch": 0.20773638968481375,
210
+ "grad_norm": 0.6395658254623413,
211
+ "learning_rate": 2.9971549931055665e-05,
212
+ "loss": 1.1072,
213
+ "step": 145
214
+ },
215
+ {
216
+ "epoch": 0.2148997134670487,
217
+ "grad_norm": 0.7063137888908386,
218
+ "learning_rate": 2.9963790544879543e-05,
219
+ "loss": 1.0851,
220
+ "step": 150
221
+ },
222
+ {
223
+ "epoch": 0.22206303724928367,
224
+ "grad_norm": 0.6892749071121216,
225
+ "learning_rate": 2.9955097883877975e-05,
226
+ "loss": 1.0828,
227
+ "step": 155
228
+ },
229
+ {
230
+ "epoch": 0.22922636103151864,
231
+ "grad_norm": 0.7910078763961792,
232
+ "learning_rate": 2.994547249020248e-05,
233
+ "loss": 1.0479,
234
+ "step": 160
235
+ },
236
+ {
237
+ "epoch": 0.23638968481375358,
238
+ "grad_norm": 0.7675610184669495,
239
+ "learning_rate": 2.9934914964178053e-05,
240
+ "loss": 1.0644,
241
+ "step": 165
242
+ },
243
+ {
244
+ "epoch": 0.24355300859598855,
245
+ "grad_norm": 0.7545039653778076,
246
+ "learning_rate": 2.992342596426575e-05,
247
+ "loss": 1.0815,
248
+ "step": 170
249
+ },
250
+ {
251
+ "epoch": 0.2507163323782235,
252
+ "grad_norm": 0.8199554681777954,
253
+ "learning_rate": 2.99110062070216e-05,
254
+ "loss": 0.9924,
255
+ "step": 175
256
+ },
257
+ {
258
+ "epoch": 0.25787965616045844,
259
+ "grad_norm": 0.7176316380500793,
260
+ "learning_rate": 2.989765646705193e-05,
261
+ "loss": 1.0116,
262
+ "step": 180
263
+ },
264
+ {
265
+ "epoch": 0.26504297994269344,
266
+ "grad_norm": 0.8339550495147705,
267
+ "learning_rate": 2.988337757696504e-05,
268
+ "loss": 1.0325,
269
+ "step": 185
270
+ },
271
+ {
272
+ "epoch": 0.2722063037249284,
273
+ "grad_norm": 0.8716332316398621,
274
+ "learning_rate": 2.986817042731928e-05,
275
+ "loss": 0.9799,
276
+ "step": 190
277
+ },
278
+ {
279
+ "epoch": 0.2793696275071633,
280
+ "grad_norm": 0.7907215356826782,
281
+ "learning_rate": 2.9852035966567515e-05,
282
+ "loss": 0.9742,
283
+ "step": 195
284
+ },
285
+ {
286
+ "epoch": 0.28653295128939826,
287
+ "grad_norm": 1.0076730251312256,
288
+ "learning_rate": 2.9834975200997954e-05,
289
+ "loss": 1.0236,
290
+ "step": 200
291
+ },
292
+ {
293
+ "epoch": 0.29369627507163326,
294
+ "grad_norm": 0.8593704700469971,
295
+ "learning_rate": 2.9816989194671403e-05,
296
+ "loss": 0.9882,
297
+ "step": 205
298
+ },
299
+ {
300
+ "epoch": 0.3008595988538682,
301
+ "grad_norm": 0.8797978162765503,
302
+ "learning_rate": 2.9798079069354893e-05,
303
+ "loss": 0.951,
304
+ "step": 210
305
+ },
306
+ {
307
+ "epoch": 0.30802292263610315,
308
+ "grad_norm": 0.8882305026054382,
309
+ "learning_rate": 2.9778246004451725e-05,
310
+ "loss": 0.9432,
311
+ "step": 215
312
+ },
313
+ {
314
+ "epoch": 0.3151862464183381,
315
+ "grad_norm": 0.9005177021026611,
316
+ "learning_rate": 2.9757491236927895e-05,
317
+ "loss": 0.9163,
318
+ "step": 220
319
+ },
320
+ {
321
+ "epoch": 0.3223495702005731,
322
+ "grad_norm": 0.8990908861160278,
323
+ "learning_rate": 2.973581606123497e-05,
324
+ "loss": 0.9105,
325
+ "step": 225
326
+ },
327
+ {
328
+ "epoch": 0.32951289398280803,
329
+ "grad_norm": 0.9843989014625549,
330
+ "learning_rate": 2.9713221829229337e-05,
331
+ "loss": 0.9352,
332
+ "step": 230
333
+ },
334
+ {
335
+ "epoch": 0.336676217765043,
336
+ "grad_norm": 1.0960330963134766,
337
+ "learning_rate": 2.9689709950087894e-05,
338
+ "loss": 0.8826,
339
+ "step": 235
340
+ },
341
+ {
342
+ "epoch": 0.3438395415472779,
343
+ "grad_norm": 0.9707636833190918,
344
+ "learning_rate": 2.966528189022016e-05,
345
+ "loss": 0.8781,
346
+ "step": 240
347
+ },
348
+ {
349
+ "epoch": 0.3510028653295129,
350
+ "grad_norm": 1.0142745971679688,
351
+ "learning_rate": 2.9639939173176808e-05,
352
+ "loss": 0.8868,
353
+ "step": 245
354
+ },
355
+ {
356
+ "epoch": 0.35816618911174786,
357
+ "grad_norm": 0.8420681953430176,
358
+ "learning_rate": 2.9613683379554675e-05,
359
+ "loss": 0.927,
360
+ "step": 250
361
+ },
362
+ {
363
+ "epoch": 0.3653295128939828,
364
+ "grad_norm": 0.993446409702301,
365
+ "learning_rate": 2.958651614689814e-05,
366
+ "loss": 0.839,
367
+ "step": 255
368
+ },
369
+ {
370
+ "epoch": 0.37249283667621774,
371
+ "grad_norm": 0.9465077519416809,
372
+ "learning_rate": 2.955843916959703e-05,
373
+ "loss": 0.8951,
374
+ "step": 260
375
+ },
376
+ {
377
+ "epoch": 0.37965616045845274,
378
+ "grad_norm": 0.9627001881599426,
379
+ "learning_rate": 2.9529454198780906e-05,
380
+ "loss": 0.8727,
381
+ "step": 265
382
+ },
383
+ {
384
+ "epoch": 0.3868194842406877,
385
+ "grad_norm": 1.0283046960830688,
386
+ "learning_rate": 2.949956304220987e-05,
387
+ "loss": 0.8285,
388
+ "step": 270
389
+ },
390
+ {
391
+ "epoch": 0.3939828080229226,
392
+ "grad_norm": 0.9945497512817383,
393
+ "learning_rate": 2.946876756416183e-05,
394
+ "loss": 0.816,
395
+ "step": 275
396
+ },
397
+ {
398
+ "epoch": 0.40114613180515757,
399
+ "grad_norm": 1.0398837327957153,
400
+ "learning_rate": 2.943706968531618e-05,
401
+ "loss": 0.8491,
402
+ "step": 280
403
+ },
404
+ {
405
+ "epoch": 0.40830945558739257,
406
+ "grad_norm": 1.0406376123428345,
407
+ "learning_rate": 2.940447138263407e-05,
408
+ "loss": 0.8263,
409
+ "step": 285
410
+ },
411
+ {
412
+ "epoch": 0.4154727793696275,
413
+ "grad_norm": 1.020673394203186,
414
+ "learning_rate": 2.9370974689235052e-05,
415
+ "loss": 0.8349,
416
+ "step": 290
417
+ },
418
+ {
419
+ "epoch": 0.42263610315186245,
420
+ "grad_norm": 0.9811904430389404,
421
+ "learning_rate": 2.9336581694270306e-05,
422
+ "loss": 0.7652,
423
+ "step": 295
424
+ },
425
+ {
426
+ "epoch": 0.4297994269340974,
427
+ "grad_norm": 1.0167213678359985,
428
+ "learning_rate": 2.930129454279234e-05,
429
+ "loss": 0.8358,
430
+ "step": 300
431
+ },
432
+ {
433
+ "epoch": 0.4369627507163324,
434
+ "grad_norm": 1.084373950958252,
435
+ "learning_rate": 2.9265115435621194e-05,
436
+ "loss": 0.8329,
437
+ "step": 305
438
+ },
439
+ {
440
+ "epoch": 0.44412607449856734,
441
+ "grad_norm": 0.9555304646492004,
442
+ "learning_rate": 2.922804662920718e-05,
443
+ "loss": 0.759,
444
+ "step": 310
445
+ },
446
+ {
447
+ "epoch": 0.4512893982808023,
448
+ "grad_norm": 0.971975564956665,
449
+ "learning_rate": 2.919009043549017e-05,
450
+ "loss": 0.7929,
451
+ "step": 315
452
+ },
453
+ {
454
+ "epoch": 0.4584527220630373,
455
+ "grad_norm": 1.3004662990570068,
456
+ "learning_rate": 2.9151249221755374e-05,
457
+ "loss": 0.7566,
458
+ "step": 320
459
+ },
460
+ {
461
+ "epoch": 0.4656160458452722,
462
+ "grad_norm": 0.9859579205513,
463
+ "learning_rate": 2.91115254104857e-05,
464
+ "loss": 0.7736,
465
+ "step": 325
466
+ },
467
+ {
468
+ "epoch": 0.47277936962750716,
469
+ "grad_norm": 1.1479684114456177,
470
+ "learning_rate": 2.9070921479210693e-05,
471
+ "loss": 0.8137,
472
+ "step": 330
473
+ },
474
+ {
475
+ "epoch": 0.4799426934097421,
476
+ "grad_norm": 1.0291324853897095,
477
+ "learning_rate": 2.9029439960351973e-05,
478
+ "loss": 0.746,
479
+ "step": 335
480
+ },
481
+ {
482
+ "epoch": 0.4871060171919771,
483
+ "grad_norm": 1.0725616216659546,
484
+ "learning_rate": 2.8987083441065335e-05,
485
+ "loss": 0.7111,
486
+ "step": 340
487
+ },
488
+ {
489
+ "epoch": 0.49426934097421205,
490
+ "grad_norm": 1.3750799894332886,
491
+ "learning_rate": 2.894385456307936e-05,
492
+ "loss": 0.7383,
493
+ "step": 345
494
+ },
495
+ {
496
+ "epoch": 0.501432664756447,
497
+ "grad_norm": 1.1375631093978882,
498
+ "learning_rate": 2.8899756022530653e-05,
499
+ "loss": 0.7223,
500
+ "step": 350
501
+ },
502
+ {
503
+ "epoch": 0.5085959885386819,
504
+ "grad_norm": 1.0370551347732544,
505
+ "learning_rate": 2.8854790569795712e-05,
506
+ "loss": 0.7349,
507
+ "step": 355
508
+ },
509
+ {
510
+ "epoch": 0.5157593123209169,
511
+ "grad_norm": 1.1560977697372437,
512
+ "learning_rate": 2.880896100931936e-05,
513
+ "loss": 0.712,
514
+ "step": 360
515
+ },
516
+ {
517
+ "epoch": 0.5229226361031518,
518
+ "grad_norm": 1.1957367658615112,
519
+ "learning_rate": 2.8762270199439873e-05,
520
+ "loss": 0.7336,
521
+ "step": 365
522
+ },
523
+ {
524
+ "epoch": 0.5300859598853869,
525
+ "grad_norm": 1.184022068977356,
526
+ "learning_rate": 2.8714721052210656e-05,
527
+ "loss": 0.6946,
528
+ "step": 370
529
+ },
530
+ {
531
+ "epoch": 0.5372492836676218,
532
+ "grad_norm": 1.082257628440857,
533
+ "learning_rate": 2.8666316533218673e-05,
534
+ "loss": 0.7018,
535
+ "step": 375
536
+ },
537
+ {
538
+ "epoch": 0.5444126074498568,
539
+ "grad_norm": 1.091844916343689,
540
+ "learning_rate": 2.861705966139946e-05,
541
+ "loss": 0.707,
542
+ "step": 380
543
+ },
544
+ {
545
+ "epoch": 0.5515759312320917,
546
+ "grad_norm": 1.3097453117370605,
547
+ "learning_rate": 2.8566953508848844e-05,
548
+ "loss": 0.6996,
549
+ "step": 385
550
+ },
551
+ {
552
+ "epoch": 0.5587392550143266,
553
+ "grad_norm": 1.3055963516235352,
554
+ "learning_rate": 2.8516001200631333e-05,
555
+ "loss": 0.6836,
556
+ "step": 390
557
+ },
558
+ {
559
+ "epoch": 0.5659025787965616,
560
+ "grad_norm": 1.4905264377593994,
561
+ "learning_rate": 2.8464205914585212e-05,
562
+ "loss": 0.6652,
563
+ "step": 395
564
+ },
565
+ {
566
+ "epoch": 0.5730659025787965,
567
+ "grad_norm": 1.25338876247406,
568
+ "learning_rate": 2.8411570881124355e-05,
569
+ "loss": 0.676,
570
+ "step": 400
571
+ },
572
+ {
573
+ "epoch": 0.5802292263610315,
574
+ "grad_norm": 1.1054009199142456,
575
+ "learning_rate": 2.8358099383036745e-05,
576
+ "loss": 0.6654,
577
+ "step": 405
578
+ },
579
+ {
580
+ "epoch": 0.5873925501432665,
581
+ "grad_norm": 1.0774147510528564,
582
+ "learning_rate": 2.8303794755279708e-05,
583
+ "loss": 0.6704,
584
+ "step": 410
585
+ },
586
+ {
587
+ "epoch": 0.5945558739255015,
588
+ "grad_norm": 1.2063066959381104,
589
+ "learning_rate": 2.8248660384771947e-05,
590
+ "loss": 0.6797,
591
+ "step": 415
592
+ },
593
+ {
594
+ "epoch": 0.6017191977077364,
595
+ "grad_norm": 1.1049535274505615,
596
+ "learning_rate": 2.8192699710182282e-05,
597
+ "loss": 0.6099,
598
+ "step": 420
599
+ },
600
+ {
601
+ "epoch": 0.6088825214899714,
602
+ "grad_norm": 1.1662622690200806,
603
+ "learning_rate": 2.8135916221715197e-05,
604
+ "loss": 0.6416,
605
+ "step": 425
606
+ },
607
+ {
608
+ "epoch": 0.6160458452722063,
609
+ "grad_norm": 1.2611438035964966,
610
+ "learning_rate": 2.8078313460893148e-05,
611
+ "loss": 0.6443,
612
+ "step": 430
613
+ },
614
+ {
615
+ "epoch": 0.6232091690544412,
616
+ "grad_norm": 1.267148494720459,
617
+ "learning_rate": 2.801989502033569e-05,
618
+ "loss": 0.633,
619
+ "step": 435
620
+ },
621
+ {
622
+ "epoch": 0.6303724928366762,
623
+ "grad_norm": 1.2322561740875244,
624
+ "learning_rate": 2.796066454353541e-05,
625
+ "loss": 0.6019,
626
+ "step": 440
627
+ },
628
+ {
629
+ "epoch": 0.6375358166189111,
630
+ "grad_norm": 1.3672127723693848,
631
+ "learning_rate": 2.790062572463069e-05,
632
+ "loss": 0.6158,
633
+ "step": 445
634
+ },
635
+ {
636
+ "epoch": 0.6446991404011462,
637
+ "grad_norm": 1.2686625719070435,
638
+ "learning_rate": 2.783978230817529e-05,
639
+ "loss": 0.5929,
640
+ "step": 450
641
+ },
642
+ {
643
+ "epoch": 0.6518624641833811,
644
+ "grad_norm": 1.251196265220642,
645
+ "learning_rate": 2.7778138088904824e-05,
646
+ "loss": 0.614,
647
+ "step": 455
648
+ },
649
+ {
650
+ "epoch": 0.6590257879656161,
651
+ "grad_norm": 1.180862545967102,
652
+ "learning_rate": 2.771569691150007e-05,
653
+ "loss": 0.613,
654
+ "step": 460
655
+ },
656
+ {
657
+ "epoch": 0.666189111747851,
658
+ "grad_norm": 1.320216178894043,
659
+ "learning_rate": 2.7652462670347207e-05,
660
+ "loss": 0.6483,
661
+ "step": 465
662
+ },
663
+ {
664
+ "epoch": 0.673352435530086,
665
+ "grad_norm": 1.2085721492767334,
666
+ "learning_rate": 2.7588439309294902e-05,
667
+ "loss": 0.6393,
668
+ "step": 470
669
+ },
670
+ {
671
+ "epoch": 0.6805157593123209,
672
+ "grad_norm": 1.4307420253753662,
673
+ "learning_rate": 2.7523630821408334e-05,
674
+ "loss": 0.6161,
675
+ "step": 475
676
+ },
677
+ {
678
+ "epoch": 0.6876790830945558,
679
+ "grad_norm": 1.1866905689239502,
680
+ "learning_rate": 2.7458041248720175e-05,
681
+ "loss": 0.5839,
682
+ "step": 480
683
+ },
684
+ {
685
+ "epoch": 0.6948424068767909,
686
+ "grad_norm": 1.1969176530838013,
687
+ "learning_rate": 2.739167468197848e-05,
688
+ "loss": 0.5837,
689
+ "step": 485
690
+ },
691
+ {
692
+ "epoch": 0.7020057306590258,
693
+ "grad_norm": 1.13490629196167,
694
+ "learning_rate": 2.732453526039154e-05,
695
+ "loss": 0.6116,
696
+ "step": 490
697
+ },
698
+ {
699
+ "epoch": 0.7091690544412608,
700
+ "grad_norm": 1.167480230331421,
701
+ "learning_rate": 2.7256627171369747e-05,
702
+ "loss": 0.6088,
703
+ "step": 495
704
+ },
705
+ {
706
+ "epoch": 0.7163323782234957,
707
+ "grad_norm": 1.2040870189666748,
708
+ "learning_rate": 2.718795465026441e-05,
709
+ "loss": 0.5179,
710
+ "step": 500
711
+ },
712
+ {
713
+ "epoch": 0.7234957020057307,
714
+ "grad_norm": 1.2380949258804321,
715
+ "learning_rate": 2.7118521980103618e-05,
716
+ "loss": 0.5853,
717
+ "step": 505
718
+ },
719
+ {
720
+ "epoch": 0.7306590257879656,
721
+ "grad_norm": 1.196877121925354,
722
+ "learning_rate": 2.7048333491325093e-05,
723
+ "loss": 0.6021,
724
+ "step": 510
725
+ },
726
+ {
727
+ "epoch": 0.7378223495702005,
728
+ "grad_norm": 1.1875160932540894,
729
+ "learning_rate": 2.697739356150613e-05,
730
+ "loss": 0.5584,
731
+ "step": 515
732
+ },
733
+ {
734
+ "epoch": 0.7449856733524355,
735
+ "grad_norm": 1.169421672821045,
736
+ "learning_rate": 2.690570661509054e-05,
737
+ "loss": 0.5533,
738
+ "step": 520
739
+ },
740
+ {
741
+ "epoch": 0.7521489971346705,
742
+ "grad_norm": 1.275260329246521,
743
+ "learning_rate": 2.6833277123112742e-05,
744
+ "loss": 0.4973,
745
+ "step": 525
746
+ },
747
+ {
748
+ "epoch": 0.7593123209169055,
749
+ "grad_norm": 1.330531358718872,
750
+ "learning_rate": 2.676010960291887e-05,
751
+ "loss": 0.5145,
752
+ "step": 530
753
+ },
754
+ {
755
+ "epoch": 0.7664756446991404,
756
+ "grad_norm": 1.18343186378479,
757
+ "learning_rate": 2.6686208617885057e-05,
758
+ "loss": 0.5291,
759
+ "step": 535
760
+ },
761
+ {
762
+ "epoch": 0.7736389684813754,
763
+ "grad_norm": 1.236130952835083,
764
+ "learning_rate": 2.661157877713282e-05,
765
+ "loss": 0.5307,
766
+ "step": 540
767
+ },
768
+ {
769
+ "epoch": 0.7808022922636103,
770
+ "grad_norm": 1.26006019115448,
771
+ "learning_rate": 2.653622473524158e-05,
772
+ "loss": 0.5282,
773
+ "step": 545
774
+ },
775
+ {
776
+ "epoch": 0.7879656160458453,
777
+ "grad_norm": 1.2142411470413208,
778
+ "learning_rate": 2.6460151191958378e-05,
779
+ "loss": 0.4956,
780
+ "step": 550
781
+ },
782
+ {
783
+ "epoch": 0.7951289398280802,
784
+ "grad_norm": 1.2132776975631714,
785
+ "learning_rate": 2.6383362891904746e-05,
786
+ "loss": 0.5193,
787
+ "step": 555
788
+ },
789
+ {
790
+ "epoch": 0.8022922636103151,
791
+ "grad_norm": 1.2030400037765503,
792
+ "learning_rate": 2.6305864624280795e-05,
793
+ "loss": 0.5304,
794
+ "step": 560
795
+ },
796
+ {
797
+ "epoch": 0.8094555873925502,
798
+ "grad_norm": 1.3670672178268433,
799
+ "learning_rate": 2.6227661222566516e-05,
800
+ "loss": 0.4965,
801
+ "step": 565
802
+ },
803
+ {
804
+ "epoch": 0.8166189111747851,
805
+ "grad_norm": 1.164743185043335,
806
+ "learning_rate": 2.6148757564220323e-05,
807
+ "loss": 0.5332,
808
+ "step": 570
809
+ },
810
+ {
811
+ "epoch": 0.8237822349570201,
812
+ "grad_norm": 1.0929417610168457,
813
+ "learning_rate": 2.6069158570374846e-05,
814
+ "loss": 0.459,
815
+ "step": 575
816
+ },
817
+ {
818
+ "epoch": 0.830945558739255,
819
+ "grad_norm": 1.2855541706085205,
820
+ "learning_rate": 2.5988869205530013e-05,
821
+ "loss": 0.4856,
822
+ "step": 580
823
+ },
824
+ {
825
+ "epoch": 0.83810888252149,
826
+ "grad_norm": 1.2150049209594727,
827
+ "learning_rate": 2.5907894477243424e-05,
828
+ "loss": 0.492,
829
+ "step": 585
830
+ },
831
+ {
832
+ "epoch": 0.8452722063037249,
833
+ "grad_norm": 1.759722352027893,
834
+ "learning_rate": 2.5826239435818015e-05,
835
+ "loss": 0.4816,
836
+ "step": 590
837
+ },
838
+ {
839
+ "epoch": 0.8524355300859598,
840
+ "grad_norm": 1.2279518842697144,
841
+ "learning_rate": 2.5743909173987093e-05,
842
+ "loss": 0.47,
843
+ "step": 595
844
+ },
845
+ {
846
+ "epoch": 0.8595988538681948,
847
+ "grad_norm": 1.5168720483779907,
848
+ "learning_rate": 2.5660908826596707e-05,
849
+ "loss": 0.4666,
850
+ "step": 600
851
+ },
852
+ {
853
+ "epoch": 0.8667621776504298,
854
+ "grad_norm": 1.203741431236267,
855
+ "learning_rate": 2.5577243570285397e-05,
856
+ "loss": 0.4405,
857
+ "step": 605
858
+ },
859
+ {
860
+ "epoch": 0.8739255014326648,
861
+ "grad_norm": 1.3632906675338745,
862
+ "learning_rate": 2.5492918623161306e-05,
863
+ "loss": 0.4942,
864
+ "step": 610
865
+ },
866
+ {
867
+ "epoch": 0.8810888252148997,
868
+ "grad_norm": 1.2164058685302734,
869
+ "learning_rate": 2.5407939244476773e-05,
870
+ "loss": 0.4978,
871
+ "step": 615
872
+ },
873
+ {
874
+ "epoch": 0.8882521489971347,
875
+ "grad_norm": 1.2076077461242676,
876
+ "learning_rate": 2.532231073430029e-05,
877
+ "loss": 0.459,
878
+ "step": 620
879
+ },
880
+ {
881
+ "epoch": 0.8954154727793696,
882
+ "grad_norm": 1.232103943824768,
883
+ "learning_rate": 2.5236038433185952e-05,
884
+ "loss": 0.4548,
885
+ "step": 625
886
+ },
887
+ {
888
+ "epoch": 0.9025787965616046,
889
+ "grad_norm": 1.200344443321228,
890
+ "learning_rate": 2.5149127721840377e-05,
891
+ "loss": 0.4329,
892
+ "step": 630
893
+ },
894
+ {
895
+ "epoch": 0.9097421203438395,
896
+ "grad_norm": 1.2939461469650269,
897
+ "learning_rate": 2.5061584020787104e-05,
898
+ "loss": 0.4274,
899
+ "step": 635
900
+ },
901
+ {
902
+ "epoch": 0.9169054441260746,
903
+ "grad_norm": 1.426303505897522,
904
+ "learning_rate": 2.497341279002854e-05,
905
+ "loss": 0.3945,
906
+ "step": 640
907
+ },
908
+ {
909
+ "epoch": 0.9240687679083095,
910
+ "grad_norm": 1.1224136352539062,
911
+ "learning_rate": 2.4884619528705407e-05,
912
+ "loss": 0.4724,
913
+ "step": 645
914
+ },
915
+ {
916
+ "epoch": 0.9312320916905444,
917
+ "grad_norm": 1.2926020622253418,
918
+ "learning_rate": 2.4795209774753777e-05,
919
+ "loss": 0.4112,
920
+ "step": 650
921
+ },
922
+ {
923
+ "epoch": 0.9383954154727794,
924
+ "grad_norm": 1.3288722038269043,
925
+ "learning_rate": 2.4705189104559683e-05,
926
+ "loss": 0.4761,
927
+ "step": 655
928
+ },
929
+ {
930
+ "epoch": 0.9455587392550143,
931
+ "grad_norm": 1.2000527381896973,
932
+ "learning_rate": 2.4614563132611314e-05,
933
+ "loss": 0.4713,
934
+ "step": 660
935
+ },
936
+ {
937
+ "epoch": 0.9527220630372493,
938
+ "grad_norm": 1.2532076835632324,
939
+ "learning_rate": 2.4523337511148843e-05,
940
+ "loss": 0.4483,
941
+ "step": 665
942
+ },
943
+ {
944
+ "epoch": 0.9598853868194842,
945
+ "grad_norm": 1.4833226203918457,
946
+ "learning_rate": 2.443151792981192e-05,
947
+ "loss": 0.4365,
948
+ "step": 670
949
+ },
950
+ {
951
+ "epoch": 0.9670487106017192,
952
+ "grad_norm": 1.3556255102157593,
953
+ "learning_rate": 2.4339110115284802e-05,
954
+ "loss": 0.4078,
955
+ "step": 675
956
+ },
957
+ {
958
+ "epoch": 0.9742120343839542,
959
+ "grad_norm": 1.2245787382125854,
960
+ "learning_rate": 2.4246119830939193e-05,
961
+ "loss": 0.4029,
962
+ "step": 680
963
+ },
964
+ {
965
+ "epoch": 0.9813753581661891,
966
+ "grad_norm": 1.3031940460205078,
967
+ "learning_rate": 2.4152552876474774e-05,
968
+ "loss": 0.4134,
969
+ "step": 685
970
+ },
971
+ {
972
+ "epoch": 0.9885386819484241,
973
+ "grad_norm": 1.1140525341033936,
974
+ "learning_rate": 2.4058415087557516e-05,
975
+ "loss": 0.3884,
976
+ "step": 690
977
+ },
978
+ {
979
+ "epoch": 0.995702005730659,
980
+ "grad_norm": 1.2650887966156006,
981
+ "learning_rate": 2.3963712335455674e-05,
982
+ "loss": 0.4043,
983
+ "step": 695
984
+ },
985
+ {
986
+ "epoch": 1.002865329512894,
987
+ "grad_norm": 1.3683085441589355,
988
+ "learning_rate": 2.3868450526673638e-05,
989
+ "loss": 0.3403,
990
+ "step": 700
991
+ },
992
+ {
993
+ "epoch": 1.010028653295129,
994
+ "grad_norm": 1.114241123199463,
995
+ "learning_rate": 2.377263560258352e-05,
996
+ "loss": 0.3354,
997
+ "step": 705
998
+ },
999
+ {
1000
+ "epoch": 1.0171919770773639,
1001
+ "grad_norm": 1.4985027313232422,
1002
+ "learning_rate": 2.3676273539054626e-05,
1003
+ "loss": 0.3407,
1004
+ "step": 710
1005
+ },
1006
+ {
1007
+ "epoch": 1.024355300859599,
1008
+ "grad_norm": 1.2135462760925293,
1009
+ "learning_rate": 2.3579370346080723e-05,
1010
+ "loss": 0.3196,
1011
+ "step": 715
1012
+ },
1013
+ {
1014
+ "epoch": 1.0315186246418337,
1015
+ "grad_norm": 1.3170359134674072,
1016
+ "learning_rate": 2.348193206740524e-05,
1017
+ "loss": 0.3378,
1018
+ "step": 720
1019
+ },
1020
+ {
1021
+ "epoch": 1.0386819484240688,
1022
+ "grad_norm": 1.2661170959472656,
1023
+ "learning_rate": 2.3383964780144263e-05,
1024
+ "loss": 0.3424,
1025
+ "step": 725
1026
+ },
1027
+ {
1028
+ "epoch": 1.0458452722063036,
1029
+ "grad_norm": 1.148691177368164,
1030
+ "learning_rate": 2.3285474594407588e-05,
1031
+ "loss": 0.3328,
1032
+ "step": 730
1033
+ },
1034
+ {
1035
+ "epoch": 1.0530085959885387,
1036
+ "grad_norm": 1.279922604560852,
1037
+ "learning_rate": 2.318646765291757e-05,
1038
+ "loss": 0.294,
1039
+ "step": 735
1040
+ },
1041
+ {
1042
+ "epoch": 1.0601719197707737,
1043
+ "grad_norm": 1.2511733770370483,
1044
+ "learning_rate": 2.3086950130626058e-05,
1045
+ "loss": 0.332,
1046
+ "step": 740
1047
+ },
1048
+ {
1049
+ "epoch": 1.0673352435530086,
1050
+ "grad_norm": 1.2338162660598755,
1051
+ "learning_rate": 2.2986928234329247e-05,
1052
+ "loss": 0.3335,
1053
+ "step": 745
1054
+ },
1055
+ {
1056
+ "epoch": 1.0744985673352436,
1057
+ "grad_norm": 1.5571306943893433,
1058
+ "learning_rate": 2.2886408202280582e-05,
1059
+ "loss": 0.2784,
1060
+ "step": 750
1061
+ },
1062
+ {
1063
+ "epoch": 1.0816618911174785,
1064
+ "grad_norm": 1.1600881814956665,
1065
+ "learning_rate": 2.278539630380166e-05,
1066
+ "loss": 0.3029,
1067
+ "step": 755
1068
+ },
1069
+ {
1070
+ "epoch": 1.0888252148997135,
1071
+ "grad_norm": 1.1799629926681519,
1072
+ "learning_rate": 2.2683898838891246e-05,
1073
+ "loss": 0.3072,
1074
+ "step": 760
1075
+ },
1076
+ {
1077
+ "epoch": 1.0959885386819483,
1078
+ "grad_norm": 1.3337632417678833,
1079
+ "learning_rate": 2.258192213783234e-05,
1080
+ "loss": 0.2964,
1081
+ "step": 765
1082
+ },
1083
+ {
1084
+ "epoch": 1.1031518624641834,
1085
+ "grad_norm": 1.2311087846755981,
1086
+ "learning_rate": 2.247947256079736e-05,
1087
+ "loss": 0.3242,
1088
+ "step": 770
1089
+ },
1090
+ {
1091
+ "epoch": 1.1103151862464182,
1092
+ "grad_norm": 1.1536506414413452,
1093
+ "learning_rate": 2.2376556497451457e-05,
1094
+ "loss": 0.32,
1095
+ "step": 775
1096
+ },
1097
+ {
1098
+ "epoch": 1.1174785100286533,
1099
+ "grad_norm": 1.196093201637268,
1100
+ "learning_rate": 2.2273180366554034e-05,
1101
+ "loss": 0.3068,
1102
+ "step": 780
1103
+ },
1104
+ {
1105
+ "epoch": 1.1246418338108883,
1106
+ "grad_norm": 1.181283950805664,
1107
+ "learning_rate": 2.2169350615558367e-05,
1108
+ "loss": 0.3114,
1109
+ "step": 785
1110
+ },
1111
+ {
1112
+ "epoch": 1.1318051575931232,
1113
+ "grad_norm": 1.2184926271438599,
1114
+ "learning_rate": 2.2065073720209522e-05,
1115
+ "loss": 0.3052,
1116
+ "step": 790
1117
+ },
1118
+ {
1119
+ "epoch": 1.1389684813753582,
1120
+ "grad_norm": 1.2585797309875488,
1121
+ "learning_rate": 2.1960356184140453e-05,
1122
+ "loss": 0.3023,
1123
+ "step": 795
1124
+ },
1125
+ {
1126
+ "epoch": 1.146131805157593,
1127
+ "grad_norm": 1.1815505027770996,
1128
+ "learning_rate": 2.185520453846638e-05,
1129
+ "loss": 0.2883,
1130
+ "step": 800
1131
+ },
1132
+ {
1133
+ "epoch": 1.153295128939828,
1134
+ "grad_norm": 1.185673713684082,
1135
+ "learning_rate": 2.1749625341377454e-05,
1136
+ "loss": 0.3158,
1137
+ "step": 805
1138
+ },
1139
+ {
1140
+ "epoch": 1.1604584527220632,
1141
+ "grad_norm": 1.4234334230422974,
1142
+ "learning_rate": 2.1643625177729727e-05,
1143
+ "loss": 0.3267,
1144
+ "step": 810
1145
+ },
1146
+ {
1147
+ "epoch": 1.167621776504298,
1148
+ "grad_norm": 1.0580003261566162,
1149
+ "learning_rate": 2.1537210658634465e-05,
1150
+ "loss": 0.2991,
1151
+ "step": 815
1152
+ },
1153
+ {
1154
+ "epoch": 1.174785100286533,
1155
+ "grad_norm": 1.446027398109436,
1156
+ "learning_rate": 2.1430388421045812e-05,
1157
+ "loss": 0.3197,
1158
+ "step": 820
1159
+ },
1160
+ {
1161
+ "epoch": 1.1819484240687679,
1162
+ "grad_norm": 1.3555097579956055,
1163
+ "learning_rate": 2.132316512734687e-05,
1164
+ "loss": 0.2752,
1165
+ "step": 825
1166
+ },
1167
+ {
1168
+ "epoch": 1.189111747851003,
1169
+ "grad_norm": 1.3425337076187134,
1170
+ "learning_rate": 2.121554746493414e-05,
1171
+ "loss": 0.2683,
1172
+ "step": 830
1173
+ },
1174
+ {
1175
+ "epoch": 1.1962750716332378,
1176
+ "grad_norm": 1.2778831720352173,
1177
+ "learning_rate": 2.110754214580048e-05,
1178
+ "loss": 0.2944,
1179
+ "step": 835
1180
+ },
1181
+ {
1182
+ "epoch": 1.2034383954154728,
1183
+ "grad_norm": 1.1608521938323975,
1184
+ "learning_rate": 2.0999155906116444e-05,
1185
+ "loss": 0.2494,
1186
+ "step": 840
1187
+ },
1188
+ {
1189
+ "epoch": 1.2106017191977076,
1190
+ "grad_norm": 1.2077664136886597,
1191
+ "learning_rate": 2.0890395505810182e-05,
1192
+ "loss": 0.276,
1193
+ "step": 845
1194
+ },
1195
+ {
1196
+ "epoch": 1.2177650429799427,
1197
+ "grad_norm": 1.1834725141525269,
1198
+ "learning_rate": 2.078126772814583e-05,
1199
+ "loss": 0.2772,
1200
+ "step": 850
1201
+ },
1202
+ {
1203
+ "epoch": 1.2249283667621778,
1204
+ "grad_norm": 1.2521677017211914,
1205
+ "learning_rate": 2.067177937930042e-05,
1206
+ "loss": 0.2776,
1207
+ "step": 855
1208
+ },
1209
+ {
1210
+ "epoch": 1.2320916905444126,
1211
+ "grad_norm": 1.2379565238952637,
1212
+ "learning_rate": 2.0561937287939413e-05,
1213
+ "loss": 0.2786,
1214
+ "step": 860
1215
+ },
1216
+ {
1217
+ "epoch": 1.2392550143266476,
1218
+ "grad_norm": 1.1778956651687622,
1219
+ "learning_rate": 2.0451748304790797e-05,
1220
+ "loss": 0.2736,
1221
+ "step": 865
1222
+ },
1223
+ {
1224
+ "epoch": 1.2464183381088825,
1225
+ "grad_norm": 1.1758215427398682,
1226
+ "learning_rate": 2.0341219302217808e-05,
1227
+ "loss": 0.2464,
1228
+ "step": 870
1229
+ },
1230
+ {
1231
+ "epoch": 1.2535816618911175,
1232
+ "grad_norm": 1.3063814640045166,
1233
+ "learning_rate": 2.023035717379031e-05,
1234
+ "loss": 0.2542,
1235
+ "step": 875
1236
+ },
1237
+ {
1238
+ "epoch": 1.2607449856733524,
1239
+ "grad_norm": 1.2965805530548096,
1240
+ "learning_rate": 2.011916883385486e-05,
1241
+ "loss": 0.2788,
1242
+ "step": 880
1243
+ },
1244
+ {
1245
+ "epoch": 1.2679083094555874,
1246
+ "grad_norm": 1.3286978006362915,
1247
+ "learning_rate": 2.0007661217103447e-05,
1248
+ "loss": 0.3239,
1249
+ "step": 885
1250
+ },
1251
+ {
1252
+ "epoch": 1.2750716332378222,
1253
+ "grad_norm": 1.2331782579421997,
1254
+ "learning_rate": 1.9895841278141007e-05,
1255
+ "loss": 0.2668,
1256
+ "step": 890
1257
+ },
1258
+ {
1259
+ "epoch": 1.2822349570200573,
1260
+ "grad_norm": 1.2736456394195557,
1261
+ "learning_rate": 1.978371599105167e-05,
1262
+ "loss": 0.2561,
1263
+ "step": 895
1264
+ },
1265
+ {
1266
+ "epoch": 1.2893982808022924,
1267
+ "grad_norm": 1.2978795766830444,
1268
+ "learning_rate": 1.967129234896376e-05,
1269
+ "loss": 0.248,
1270
+ "step": 900
1271
+ },
1272
+ {
1273
+ "epoch": 1.2965616045845272,
1274
+ "grad_norm": 1.1861902475357056,
1275
+ "learning_rate": 1.9558577363613704e-05,
1276
+ "loss": 0.2619,
1277
+ "step": 905
1278
+ },
1279
+ {
1280
+ "epoch": 1.3037249283667622,
1281
+ "grad_norm": 1.3414446115493774,
1282
+ "learning_rate": 1.9445578064908645e-05,
1283
+ "loss": 0.2735,
1284
+ "step": 910
1285
+ },
1286
+ {
1287
+ "epoch": 1.310888252148997,
1288
+ "grad_norm": 1.2330338954925537,
1289
+ "learning_rate": 1.9332301500488052e-05,
1290
+ "loss": 0.2627,
1291
+ "step": 915
1292
+ },
1293
+ {
1294
+ "epoch": 1.3180515759312321,
1295
+ "grad_norm": 1.3122456073760986,
1296
+ "learning_rate": 1.921875473528414e-05,
1297
+ "loss": 0.2568,
1298
+ "step": 920
1299
+ },
1300
+ {
1301
+ "epoch": 1.3252148997134672,
1302
+ "grad_norm": 1.29453706741333,
1303
+ "learning_rate": 1.9104944851081247e-05,
1304
+ "loss": 0.2522,
1305
+ "step": 925
1306
+ },
1307
+ {
1308
+ "epoch": 1.332378223495702,
1309
+ "grad_norm": 1.0213770866394043,
1310
+ "learning_rate": 1.8990878946074136e-05,
1311
+ "loss": 0.2488,
1312
+ "step": 930
1313
+ },
1314
+ {
1315
+ "epoch": 1.3395415472779368,
1316
+ "grad_norm": 1.1831605434417725,
1317
+ "learning_rate": 1.887656413442532e-05,
1318
+ "loss": 0.2551,
1319
+ "step": 935
1320
+ },
1321
+ {
1322
+ "epoch": 1.346704871060172,
1323
+ "grad_norm": 1.23065185546875,
1324
+ "learning_rate": 1.8762007545821323e-05,
1325
+ "loss": 0.2395,
1326
+ "step": 940
1327
+ },
1328
+ {
1329
+ "epoch": 1.353868194842407,
1330
+ "grad_norm": 1.2198963165283203,
1331
+ "learning_rate": 1.8647216325028037e-05,
1332
+ "loss": 0.2589,
1333
+ "step": 945
1334
+ },
1335
+ {
1336
+ "epoch": 1.3610315186246418,
1337
+ "grad_norm": 1.225571632385254,
1338
+ "learning_rate": 1.8532197631445095e-05,
1339
+ "loss": 0.2463,
1340
+ "step": 950
1341
+ },
1342
+ {
1343
+ "epoch": 1.3681948424068768,
1344
+ "grad_norm": 1.2289516925811768,
1345
+ "learning_rate": 1.8416958638659348e-05,
1346
+ "loss": 0.2541,
1347
+ "step": 955
1348
+ },
1349
+ {
1350
+ "epoch": 1.3753581661891117,
1351
+ "grad_norm": 1.115930199623108,
1352
+ "learning_rate": 1.8301506533997465e-05,
1353
+ "loss": 0.2393,
1354
+ "step": 960
1355
+ },
1356
+ {
1357
+ "epoch": 1.3825214899713467,
1358
+ "grad_norm": 1.249009132385254,
1359
+ "learning_rate": 1.8185848518077676e-05,
1360
+ "loss": 0.2249,
1361
+ "step": 965
1362
+ },
1363
+ {
1364
+ "epoch": 1.3896848137535818,
1365
+ "grad_norm": 1.1407043933868408,
1366
+ "learning_rate": 1.806999180436065e-05,
1367
+ "loss": 0.2644,
1368
+ "step": 970
1369
+ },
1370
+ {
1371
+ "epoch": 1.3968481375358166,
1372
+ "grad_norm": 1.2310693264007568,
1373
+ "learning_rate": 1.7953943618699616e-05,
1374
+ "loss": 0.2648,
1375
+ "step": 975
1376
+ },
1377
+ {
1378
+ "epoch": 1.4040114613180517,
1379
+ "grad_norm": 1.0873064994812012,
1380
+ "learning_rate": 1.7837711198889703e-05,
1381
+ "loss": 0.2133,
1382
+ "step": 980
1383
+ },
1384
+ {
1385
+ "epoch": 1.4111747851002865,
1386
+ "grad_norm": 1.1521344184875488,
1387
+ "learning_rate": 1.7721301794216498e-05,
1388
+ "loss": 0.2282,
1389
+ "step": 985
1390
+ },
1391
+ {
1392
+ "epoch": 1.4183381088825215,
1393
+ "grad_norm": 1.293166995048523,
1394
+ "learning_rate": 1.760472266500396e-05,
1395
+ "loss": 0.1975,
1396
+ "step": 990
1397
+ },
1398
+ {
1399
+ "epoch": 1.4255014326647564,
1400
+ "grad_norm": 0.9224392771720886,
1401
+ "learning_rate": 1.748798108216155e-05,
1402
+ "loss": 0.2272,
1403
+ "step": 995
1404
+ },
1405
+ {
1406
+ "epoch": 1.4326647564469914,
1407
+ "grad_norm": 1.08697509765625,
1408
+ "learning_rate": 1.7371084326730813e-05,
1409
+ "loss": 0.207,
1410
+ "step": 1000
1411
+ },
1412
+ {
1413
+ "epoch": 1.4398280802292263,
1414
+ "grad_norm": 1.134006142616272,
1415
+ "learning_rate": 1.7254039689431206e-05,
1416
+ "loss": 0.2067,
1417
+ "step": 1005
1418
+ },
1419
+ {
1420
+ "epoch": 1.4469914040114613,
1421
+ "grad_norm": 1.0207308530807495,
1422
+ "learning_rate": 1.7136854470205424e-05,
1423
+ "loss": 0.2038,
1424
+ "step": 1010
1425
+ },
1426
+ {
1427
+ "epoch": 1.4541547277936964,
1428
+ "grad_norm": 1.2723283767700195,
1429
+ "learning_rate": 1.701953597776409e-05,
1430
+ "loss": 0.2165,
1431
+ "step": 1015
1432
+ },
1433
+ {
1434
+ "epoch": 1.4613180515759312,
1435
+ "grad_norm": 1.2360570430755615,
1436
+ "learning_rate": 1.6902091529129946e-05,
1437
+ "loss": 0.196,
1438
+ "step": 1020
1439
+ },
1440
+ {
1441
+ "epoch": 1.4684813753581663,
1442
+ "grad_norm": 1.1582695245742798,
1443
+ "learning_rate": 1.678452844918146e-05,
1444
+ "loss": 0.217,
1445
+ "step": 1025
1446
+ },
1447
+ {
1448
+ "epoch": 1.475644699140401,
1449
+ "grad_norm": 1.1291875839233398,
1450
+ "learning_rate": 1.666685407019601e-05,
1451
+ "loss": 0.1952,
1452
+ "step": 1030
1453
+ },
1454
+ {
1455
+ "epoch": 1.4828080229226361,
1456
+ "grad_norm": 1.1681057214736938,
1457
+ "learning_rate": 1.6549075731392586e-05,
1458
+ "loss": 0.2201,
1459
+ "step": 1035
1460
+ },
1461
+ {
1462
+ "epoch": 1.4899713467048712,
1463
+ "grad_norm": 1.1959859132766724,
1464
+ "learning_rate": 1.6431200778474002e-05,
1465
+ "loss": 0.2147,
1466
+ "step": 1040
1467
+ },
1468
+ {
1469
+ "epoch": 1.497134670487106,
1470
+ "grad_norm": 1.270689606666565,
1471
+ "learning_rate": 1.631323656316881e-05,
1472
+ "loss": 0.2238,
1473
+ "step": 1045
1474
+ },
1475
+ {
1476
+ "epoch": 1.5042979942693409,
1477
+ "grad_norm": 1.2683687210083008,
1478
+ "learning_rate": 1.6195190442772764e-05,
1479
+ "loss": 0.2156,
1480
+ "step": 1050
1481
+ },
1482
+ {
1483
+ "epoch": 1.511461318051576,
1484
+ "grad_norm": 1.0826033353805542,
1485
+ "learning_rate": 1.6077069779689915e-05,
1486
+ "loss": 0.1823,
1487
+ "step": 1055
1488
+ },
1489
+ {
1490
+ "epoch": 1.518624641833811,
1491
+ "grad_norm": 1.2781873941421509,
1492
+ "learning_rate": 1.5958881940973487e-05,
1493
+ "loss": 0.2059,
1494
+ "step": 1060
1495
+ },
1496
+ {
1497
+ "epoch": 1.5257879656160458,
1498
+ "grad_norm": 1.1984552145004272,
1499
+ "learning_rate": 1.5840634297866354e-05,
1500
+ "loss": 0.2281,
1501
+ "step": 1065
1502
+ },
1503
+ {
1504
+ "epoch": 1.5329512893982808,
1505
+ "grad_norm": 1.1112569570541382,
1506
+ "learning_rate": 1.5722334225341345e-05,
1507
+ "loss": 0.2119,
1508
+ "step": 1070
1509
+ },
1510
+ {
1511
+ "epoch": 1.5401146131805157,
1512
+ "grad_norm": 1.36013925075531,
1513
+ "learning_rate": 1.5603989101641228e-05,
1514
+ "loss": 0.1905,
1515
+ "step": 1075
1516
+ },
1517
+ {
1518
+ "epoch": 1.5472779369627507,
1519
+ "grad_norm": 1.2900772094726562,
1520
+ "learning_rate": 1.548560630781858e-05,
1521
+ "loss": 0.1952,
1522
+ "step": 1080
1523
+ },
1524
+ {
1525
+ "epoch": 1.5544412607449858,
1526
+ "grad_norm": 1.240225911140442,
1527
+ "learning_rate": 1.5367193227275408e-05,
1528
+ "loss": 0.21,
1529
+ "step": 1085
1530
+ },
1531
+ {
1532
+ "epoch": 1.5616045845272206,
1533
+ "grad_norm": 1.2521461248397827,
1534
+ "learning_rate": 1.5248757245302672e-05,
1535
+ "loss": 0.2088,
1536
+ "step": 1090
1537
+ },
1538
+ {
1539
+ "epoch": 1.5687679083094554,
1540
+ "grad_norm": 1.1497176885604858,
1541
+ "learning_rate": 1.5130305748619686e-05,
1542
+ "loss": 0.2078,
1543
+ "step": 1095
1544
+ },
1545
+ {
1546
+ "epoch": 1.5759312320916905,
1547
+ "grad_norm": 1.4251394271850586,
1548
+ "learning_rate": 1.501184612491337e-05,
1549
+ "loss": 0.189,
1550
+ "step": 1100
1551
+ },
1552
+ {
1553
+ "epoch": 1.5830945558739256,
1554
+ "grad_norm": 1.315163016319275,
1555
+ "learning_rate": 1.4893385762377536e-05,
1556
+ "loss": 0.2019,
1557
+ "step": 1105
1558
+ },
1559
+ {
1560
+ "epoch": 1.5902578796561606,
1561
+ "grad_norm": 1.158630609512329,
1562
+ "learning_rate": 1.4774932049252073e-05,
1563
+ "loss": 0.1926,
1564
+ "step": 1110
1565
+ },
1566
+ {
1567
+ "epoch": 1.5974212034383954,
1568
+ "grad_norm": 1.2413978576660156,
1569
+ "learning_rate": 1.4656492373362148e-05,
1570
+ "loss": 0.179,
1571
+ "step": 1115
1572
+ },
1573
+ {
1574
+ "epoch": 1.6045845272206303,
1575
+ "grad_norm": 1.2058839797973633,
1576
+ "learning_rate": 1.4538074121657448e-05,
1577
+ "loss": 0.1949,
1578
+ "step": 1120
1579
+ },
1580
+ {
1581
+ "epoch": 1.6117478510028653,
1582
+ "grad_norm": 1.6929833889007568,
1583
+ "learning_rate": 1.4419684679751455e-05,
1584
+ "loss": 0.1949,
1585
+ "step": 1125
1586
+ },
1587
+ {
1588
+ "epoch": 1.6189111747851004,
1589
+ "grad_norm": 1.1389960050582886,
1590
+ "learning_rate": 1.4301331431460814e-05,
1591
+ "loss": 0.2017,
1592
+ "step": 1130
1593
+ },
1594
+ {
1595
+ "epoch": 1.6260744985673352,
1596
+ "grad_norm": 1.1719030141830444,
1597
+ "learning_rate": 1.4183021758344812e-05,
1598
+ "loss": 0.2029,
1599
+ "step": 1135
1600
+ },
1601
+ {
1602
+ "epoch": 1.63323782234957,
1603
+ "grad_norm": 1.1961101293563843,
1604
+ "learning_rate": 1.4064763039245019e-05,
1605
+ "loss": 0.1924,
1606
+ "step": 1140
1607
+ },
1608
+ {
1609
+ "epoch": 1.640401146131805,
1610
+ "grad_norm": 1.313583254814148,
1611
+ "learning_rate": 1.3946562649825038e-05,
1612
+ "loss": 0.1979,
1613
+ "step": 1145
1614
+ },
1615
+ {
1616
+ "epoch": 1.6475644699140402,
1617
+ "grad_norm": 1.1494886875152588,
1618
+ "learning_rate": 1.382842796211053e-05,
1619
+ "loss": 0.178,
1620
+ "step": 1150
1621
+ },
1622
+ {
1623
+ "epoch": 1.6547277936962752,
1624
+ "grad_norm": 1.1895718574523926,
1625
+ "learning_rate": 1.3710366344029408e-05,
1626
+ "loss": 0.1581,
1627
+ "step": 1155
1628
+ },
1629
+ {
1630
+ "epoch": 1.66189111747851,
1631
+ "grad_norm": 1.0595122575759888,
1632
+ "learning_rate": 1.3592385158952312e-05,
1633
+ "loss": 0.1748,
1634
+ "step": 1160
1635
+ },
1636
+ {
1637
+ "epoch": 1.6690544412607449,
1638
+ "grad_norm": 0.964249312877655,
1639
+ "learning_rate": 1.3474491765233368e-05,
1640
+ "loss": 0.15,
1641
+ "step": 1165
1642
+ },
1643
+ {
1644
+ "epoch": 1.67621776504298,
1645
+ "grad_norm": 1.2320538759231567,
1646
+ "learning_rate": 1.335669351575126e-05,
1647
+ "loss": 0.1907,
1648
+ "step": 1170
1649
+ },
1650
+ {
1651
+ "epoch": 1.683381088825215,
1652
+ "grad_norm": 1.211921215057373,
1653
+ "learning_rate": 1.3238997757450622e-05,
1654
+ "loss": 0.1893,
1655
+ "step": 1175
1656
+ },
1657
+ {
1658
+ "epoch": 1.6905444126074498,
1659
+ "grad_norm": 1.0907567739486694,
1660
+ "learning_rate": 1.3121411830883838e-05,
1661
+ "loss": 0.1657,
1662
+ "step": 1180
1663
+ },
1664
+ {
1665
+ "epoch": 1.6977077363896849,
1666
+ "grad_norm": 1.188611626625061,
1667
+ "learning_rate": 1.30039430697532e-05,
1668
+ "loss": 0.1693,
1669
+ "step": 1185
1670
+ },
1671
+ {
1672
+ "epoch": 1.7048710601719197,
1673
+ "grad_norm": 0.9809935092926025,
1674
+ "learning_rate": 1.2886598800453527e-05,
1675
+ "loss": 0.185,
1676
+ "step": 1190
1677
+ },
1678
+ {
1679
+ "epoch": 1.7120343839541547,
1680
+ "grad_norm": 1.2036057710647583,
1681
+ "learning_rate": 1.2769386341615222e-05,
1682
+ "loss": 0.1645,
1683
+ "step": 1195
1684
+ },
1685
+ {
1686
+ "epoch": 1.7191977077363898,
1687
+ "grad_norm": 1.0425801277160645,
1688
+ "learning_rate": 1.2652313003647832e-05,
1689
+ "loss": 0.1736,
1690
+ "step": 1200
1691
+ },
1692
+ {
1693
+ "epoch": 1.7263610315186246,
1694
+ "grad_norm": 1.159801721572876,
1695
+ "learning_rate": 1.253538608828407e-05,
1696
+ "loss": 0.174,
1697
+ "step": 1205
1698
+ },
1699
+ {
1700
+ "epoch": 1.7335243553008595,
1701
+ "grad_norm": 1.0157557725906372,
1702
+ "learning_rate": 1.241861288812444e-05,
1703
+ "loss": 0.1702,
1704
+ "step": 1210
1705
+ },
1706
+ {
1707
+ "epoch": 1.7406876790830945,
1708
+ "grad_norm": 1.0609725713729858,
1709
+ "learning_rate": 1.230200068618241e-05,
1710
+ "loss": 0.1398,
1711
+ "step": 1215
1712
+ },
1713
+ {
1714
+ "epoch": 1.7478510028653296,
1715
+ "grad_norm": 1.1517068147659302,
1716
+ "learning_rate": 1.218555675543016e-05,
1717
+ "loss": 0.1564,
1718
+ "step": 1220
1719
+ },
1720
+ {
1721
+ "epoch": 1.7550143266475646,
1722
+ "grad_norm": 1.2830114364624023,
1723
+ "learning_rate": 1.2069288358344997e-05,
1724
+ "loss": 0.1663,
1725
+ "step": 1225
1726
+ },
1727
+ {
1728
+ "epoch": 1.7621776504297995,
1729
+ "grad_norm": 1.0851353406906128,
1730
+ "learning_rate": 1.1953202746456377e-05,
1731
+ "loss": 0.1645,
1732
+ "step": 1230
1733
+ },
1734
+ {
1735
+ "epoch": 1.7693409742120343,
1736
+ "grad_norm": 1.1211943626403809,
1737
+ "learning_rate": 1.1837307159893656e-05,
1738
+ "loss": 0.1522,
1739
+ "step": 1235
1740
+ },
1741
+ {
1742
+ "epoch": 1.7765042979942693,
1743
+ "grad_norm": 3.025891065597534,
1744
+ "learning_rate": 1.1721608826934517e-05,
1745
+ "loss": 0.1665,
1746
+ "step": 1240
1747
+ },
1748
+ {
1749
+ "epoch": 1.7836676217765044,
1750
+ "grad_norm": 1.019081950187683,
1751
+ "learning_rate": 1.1606114963554171e-05,
1752
+ "loss": 0.1544,
1753
+ "step": 1245
1754
+ },
1755
+ {
1756
+ "epoch": 1.7908309455587392,
1757
+ "grad_norm": 1.0036916732788086,
1758
+ "learning_rate": 1.1490832772975275e-05,
1759
+ "loss": 0.1422,
1760
+ "step": 1250
1761
+ },
1762
+ {
1763
+ "epoch": 1.797994269340974,
1764
+ "grad_norm": 1.0035783052444458,
1765
+ "learning_rate": 1.13757694452187e-05,
1766
+ "loss": 0.1462,
1767
+ "step": 1255
1768
+ },
1769
+ {
1770
+ "epoch": 1.8051575931232091,
1771
+ "grad_norm": 1.0962002277374268,
1772
+ "learning_rate": 1.126093215665508e-05,
1773
+ "loss": 0.1527,
1774
+ "step": 1260
1775
+ },
1776
+ {
1777
+ "epoch": 1.8123209169054442,
1778
+ "grad_norm": 1.1270493268966675,
1779
+ "learning_rate": 1.1146328069557252e-05,
1780
+ "loss": 0.1476,
1781
+ "step": 1265
1782
+ },
1783
+ {
1784
+ "epoch": 1.8194842406876792,
1785
+ "grad_norm": 1.1545363664627075,
1786
+ "learning_rate": 1.1031964331653526e-05,
1787
+ "loss": 0.1412,
1788
+ "step": 1270
1789
+ },
1790
+ {
1791
+ "epoch": 1.826647564469914,
1792
+ "grad_norm": 0.9560638070106506,
1793
+ "learning_rate": 1.0917848075681915e-05,
1794
+ "loss": 0.1326,
1795
+ "step": 1275
1796
+ },
1797
+ {
1798
+ "epoch": 1.8338108882521489,
1799
+ "grad_norm": 1.1150048971176147,
1800
+ "learning_rate": 1.0803986418945247e-05,
1801
+ "loss": 0.1491,
1802
+ "step": 1280
1803
+ },
1804
+ {
1805
+ "epoch": 1.840974212034384,
1806
+ "grad_norm": 1.172191858291626,
1807
+ "learning_rate": 1.0690386462867294e-05,
1808
+ "loss": 0.1473,
1809
+ "step": 1285
1810
+ },
1811
+ {
1812
+ "epoch": 1.848137535816619,
1813
+ "grad_norm": 1.1756559610366821,
1814
+ "learning_rate": 1.057705529254984e-05,
1815
+ "loss": 0.1508,
1816
+ "step": 1290
1817
+ },
1818
+ {
1819
+ "epoch": 1.8553008595988538,
1820
+ "grad_norm": 0.9707951545715332,
1821
+ "learning_rate": 1.046399997633081e-05,
1822
+ "loss": 0.1545,
1823
+ "step": 1295
1824
+ },
1825
+ {
1826
+ "epoch": 1.8624641833810889,
1827
+ "grad_norm": 1.12907075881958,
1828
+ "learning_rate": 1.0351227565343422e-05,
1829
+ "loss": 0.1416,
1830
+ "step": 1300
1831
+ },
1832
+ {
1833
+ "epoch": 1.8696275071633237,
1834
+ "grad_norm": 1.0165517330169678,
1835
+ "learning_rate": 1.0238745093076407e-05,
1836
+ "loss": 0.1443,
1837
+ "step": 1305
1838
+ },
1839
+ {
1840
+ "epoch": 1.8767908309455588,
1841
+ "grad_norm": 1.2274993658065796,
1842
+ "learning_rate": 1.0126559574935343e-05,
1843
+ "loss": 0.1428,
1844
+ "step": 1310
1845
+ },
1846
+ {
1847
+ "epoch": 1.8839541547277938,
1848
+ "grad_norm": 1.0325816869735718,
1849
+ "learning_rate": 1.0014678007805108e-05,
1850
+ "loss": 0.147,
1851
+ "step": 1315
1852
+ },
1853
+ {
1854
+ "epoch": 1.8911174785100286,
1855
+ "grad_norm": 0.9291435480117798,
1856
+ "learning_rate": 9.903107369613505e-06,
1857
+ "loss": 0.1348,
1858
+ "step": 1320
1859
+ },
1860
+ {
1861
+ "epoch": 1.8982808022922635,
1862
+ "grad_norm": 1.184050440788269,
1863
+ "learning_rate": 9.791854618896049e-06,
1864
+ "loss": 0.143,
1865
+ "step": 1325
1866
+ },
1867
+ {
1868
+ "epoch": 1.9054441260744985,
1869
+ "grad_norm": 1.1466131210327148,
1870
+ "learning_rate": 9.680926694361966e-06,
1871
+ "loss": 0.1418,
1872
+ "step": 1330
1873
+ },
1874
+ {
1875
+ "epoch": 1.9126074498567336,
1876
+ "grad_norm": 1.0839155912399292,
1877
+ "learning_rate": 9.570330514461431e-06,
1878
+ "loss": 0.1437,
1879
+ "step": 1335
1880
+ },
1881
+ {
1882
+ "epoch": 1.9197707736389686,
1883
+ "grad_norm": 1.0298680067062378,
1884
+ "learning_rate": 9.46007297695409e-06,
1885
+ "loss": 0.1483,
1886
+ "step": 1340
1887
+ },
1888
+ {
1889
+ "epoch": 1.9269340974212035,
1890
+ "grad_norm": 1.0673662424087524,
1891
+ "learning_rate": 9.350160958478834e-06,
1892
+ "loss": 0.1224,
1893
+ "step": 1345
1894
+ },
1895
+ {
1896
+ "epoch": 1.9340974212034383,
1897
+ "grad_norm": 1.1538810729980469,
1898
+ "learning_rate": 9.240601314124925e-06,
1899
+ "loss": 0.1663,
1900
+ "step": 1350
1901
+ },
1902
+ {
1903
+ "epoch": 1.9412607449856734,
1904
+ "grad_norm": 1.090591311454773,
1905
+ "learning_rate": 9.131400877004446e-06,
1906
+ "loss": 0.1222,
1907
+ "step": 1355
1908
+ },
1909
+ {
1910
+ "epoch": 1.9484240687679084,
1911
+ "grad_norm": 1.3217843770980835,
1912
+ "learning_rate": 9.022566457826125e-06,
1913
+ "loss": 0.1305,
1914
+ "step": 1360
1915
+ },
1916
+ {
1917
+ "epoch": 1.9555873925501432,
1918
+ "grad_norm": 0.9129636883735657,
1919
+ "learning_rate": 8.914104844470563e-06,
1920
+ "loss": 0.134,
1921
+ "step": 1365
1922
+ },
1923
+ {
1924
+ "epoch": 1.962750716332378,
1925
+ "grad_norm": 0.8943307399749756,
1926
+ "learning_rate": 8.806022801566877e-06,
1927
+ "loss": 0.1289,
1928
+ "step": 1370
1929
+ },
1930
+ {
1931
+ "epoch": 1.9699140401146131,
1932
+ "grad_norm": 0.8952675461769104,
1933
+ "learning_rate": 8.698327070070812e-06,
1934
+ "loss": 0.1259,
1935
+ "step": 1375
1936
+ },
1937
+ {
1938
+ "epoch": 1.9770773638968482,
1939
+ "grad_norm": 0.9694656133651733,
1940
+ "learning_rate": 8.591024366844291e-06,
1941
+ "loss": 0.1288,
1942
+ "step": 1380
1943
+ },
1944
+ {
1945
+ "epoch": 1.9842406876790832,
1946
+ "grad_norm": 0.9191892147064209,
1947
+ "learning_rate": 8.484121384236505e-06,
1948
+ "loss": 0.1476,
1949
+ "step": 1385
1950
+ },
1951
+ {
1952
+ "epoch": 1.991404011461318,
1953
+ "grad_norm": 1.4172223806381226,
1954
+ "learning_rate": 8.377624789666536e-06,
1955
+ "loss": 0.1235,
1956
+ "step": 1390
1957
+ },
1958
+ {
1959
+ "epoch": 1.998567335243553,
1960
+ "grad_norm": 1.0805190801620483,
1961
+ "learning_rate": 8.271541225207483e-06,
1962
+ "loss": 0.1223,
1963
+ "step": 1395
1964
+ },
1965
+ {
1966
+ "epoch": 2.005730659025788,
1967
+ "grad_norm": 0.8016635775566101,
1968
+ "learning_rate": 8.165877307172222e-06,
1969
+ "loss": 0.1018,
1970
+ "step": 1400
1971
+ },
1972
+ {
1973
+ "epoch": 2.012893982808023,
1974
+ "grad_norm": 0.9081824421882629,
1975
+ "learning_rate": 8.06063962570076e-06,
1976
+ "loss": 0.1036,
1977
+ "step": 1405
1978
+ },
1979
+ {
1980
+ "epoch": 2.020057306590258,
1981
+ "grad_norm": 0.95320063829422,
1982
+ "learning_rate": 7.955834744349192e-06,
1983
+ "loss": 0.0883,
1984
+ "step": 1410
1985
+ },
1986
+ {
1987
+ "epoch": 2.0272206303724927,
1988
+ "grad_norm": 0.9411661028862,
1989
+ "learning_rate": 7.851469199680382e-06,
1990
+ "loss": 0.1031,
1991
+ "step": 1415
1992
+ },
1993
+ {
1994
+ "epoch": 2.0343839541547277,
1995
+ "grad_norm": 0.9394431710243225,
1996
+ "learning_rate": 7.747549500856237e-06,
1997
+ "loss": 0.1024,
1998
+ "step": 1420
1999
+ },
2000
+ {
2001
+ "epoch": 2.041547277936963,
2002
+ "grad_norm": 0.9306779503822327,
2003
+ "learning_rate": 7.644082129231762e-06,
2004
+ "loss": 0.0961,
2005
+ "step": 1425
2006
+ },
2007
+ {
2008
+ "epoch": 2.048710601719198,
2009
+ "grad_norm": 0.8523784279823303,
2010
+ "learning_rate": 7.541073537950826e-06,
2011
+ "loss": 0.0997,
2012
+ "step": 1430
2013
+ },
2014
+ {
2015
+ "epoch": 2.055873925501433,
2016
+ "grad_norm": 0.8980006575584412,
2017
+ "learning_rate": 7.438530151543682e-06,
2018
+ "loss": 0.1091,
2019
+ "step": 1435
2020
+ },
2021
+ {
2022
+ "epoch": 2.0630372492836675,
2023
+ "grad_norm": 0.8851286172866821,
2024
+ "learning_rate": 7.336458365526278e-06,
2025
+ "loss": 0.1004,
2026
+ "step": 1440
2027
+ },
2028
+ {
2029
+ "epoch": 2.0702005730659025,
2030
+ "grad_norm": 0.9961656332015991,
2031
+ "learning_rate": 7.234864546001364e-06,
2032
+ "loss": 0.0979,
2033
+ "step": 1445
2034
+ },
2035
+ {
2036
+ "epoch": 2.0773638968481376,
2037
+ "grad_norm": 0.8253286480903625,
2038
+ "learning_rate": 7.133755029261478e-06,
2039
+ "loss": 0.1062,
2040
+ "step": 1450
2041
+ },
2042
+ {
2043
+ "epoch": 2.0845272206303727,
2044
+ "grad_norm": 0.799706220626831,
2045
+ "learning_rate": 7.03313612139372e-06,
2046
+ "loss": 0.1007,
2047
+ "step": 1455
2048
+ },
2049
+ {
2050
+ "epoch": 2.0916905444126073,
2051
+ "grad_norm": 0.9287534952163696,
2052
+ "learning_rate": 6.933014097886464e-06,
2053
+ "loss": 0.0939,
2054
+ "step": 1460
2055
+ },
2056
+ {
2057
+ "epoch": 2.0988538681948423,
2058
+ "grad_norm": 1.0432243347167969,
2059
+ "learning_rate": 6.833395203237979e-06,
2060
+ "loss": 0.0971,
2061
+ "step": 1465
2062
+ },
2063
+ {
2064
+ "epoch": 2.1060171919770774,
2065
+ "grad_norm": 1.0664498805999756,
2066
+ "learning_rate": 6.734285650566938e-06,
2067
+ "loss": 0.1053,
2068
+ "step": 1470
2069
+ },
2070
+ {
2071
+ "epoch": 2.1131805157593124,
2072
+ "grad_norm": 0.8691348433494568,
2073
+ "learning_rate": 6.6356916212249294e-06,
2074
+ "loss": 0.0965,
2075
+ "step": 1475
2076
+ },
2077
+ {
2078
+ "epoch": 2.1203438395415475,
2079
+ "grad_norm": 1.0136277675628662,
2080
+ "learning_rate": 6.53761926441093e-06,
2081
+ "loss": 0.1072,
2082
+ "step": 1480
2083
+ },
2084
+ {
2085
+ "epoch": 2.127507163323782,
2086
+ "grad_norm": 0.9431965947151184,
2087
+ "learning_rate": 6.440074696787791e-06,
2088
+ "loss": 0.1038,
2089
+ "step": 1485
2090
+ },
2091
+ {
2092
+ "epoch": 2.134670487106017,
2093
+ "grad_norm": 0.9201871156692505,
2094
+ "learning_rate": 6.343064002100737e-06,
2095
+ "loss": 0.0909,
2096
+ "step": 1490
2097
+ },
2098
+ {
2099
+ "epoch": 2.141833810888252,
2100
+ "grad_norm": 1.684442400932312,
2101
+ "learning_rate": 6.246593230797942e-06,
2102
+ "loss": 0.1072,
2103
+ "step": 1495
2104
+ },
2105
+ {
2106
+ "epoch": 2.1489971346704873,
2107
+ "grad_norm": 0.8413323760032654,
2108
+ "learning_rate": 6.150668399653161e-06,
2109
+ "loss": 0.0929,
2110
+ "step": 1500
2111
+ },
2112
+ {
2113
+ "epoch": 2.156160458452722,
2114
+ "grad_norm": 0.9635283350944519,
2115
+ "learning_rate": 6.055295491390472e-06,
2116
+ "loss": 0.0977,
2117
+ "step": 1505
2118
+ },
2119
+ {
2120
+ "epoch": 2.163323782234957,
2121
+ "grad_norm": 1.1035348176956177,
2122
+ "learning_rate": 5.960480454311155e-06,
2123
+ "loss": 0.1024,
2124
+ "step": 1510
2125
+ },
2126
+ {
2127
+ "epoch": 2.170487106017192,
2128
+ "grad_norm": 0.9366750717163086,
2129
+ "learning_rate": 5.8662292019226715e-06,
2130
+ "loss": 0.0962,
2131
+ "step": 1515
2132
+ },
2133
+ {
2134
+ "epoch": 2.177650429799427,
2135
+ "grad_norm": 0.933452844619751,
2136
+ "learning_rate": 5.772547612569884e-06,
2137
+ "loss": 0.0959,
2138
+ "step": 1520
2139
+ },
2140
+ {
2141
+ "epoch": 2.184813753581662,
2142
+ "grad_norm": 0.8540327548980713,
2143
+ "learning_rate": 5.679441529068392e-06,
2144
+ "loss": 0.0905,
2145
+ "step": 1525
2146
+ },
2147
+ {
2148
+ "epoch": 2.1919770773638967,
2149
+ "grad_norm": 0.8865665793418884,
2150
+ "learning_rate": 5.5869167583401556e-06,
2151
+ "loss": 0.0933,
2152
+ "step": 1530
2153
+ },
2154
+ {
2155
+ "epoch": 2.1991404011461317,
2156
+ "grad_norm": 0.878319501876831,
2157
+ "learning_rate": 5.494979071051297e-06,
2158
+ "loss": 0.094,
2159
+ "step": 1535
2160
+ },
2161
+ {
2162
+ "epoch": 2.206303724928367,
2163
+ "grad_norm": 0.7740932703018188,
2164
+ "learning_rate": 5.4036342012522135e-06,
2165
+ "loss": 0.0911,
2166
+ "step": 1540
2167
+ },
2168
+ {
2169
+ "epoch": 2.213467048710602,
2170
+ "grad_norm": 0.9036451578140259,
2171
+ "learning_rate": 5.312887846019935e-06,
2172
+ "loss": 0.0943,
2173
+ "step": 1545
2174
+ },
2175
+ {
2176
+ "epoch": 2.2206303724928365,
2177
+ "grad_norm": 1.1190263032913208,
2178
+ "learning_rate": 5.22274566510281e-06,
2179
+ "loss": 0.0948,
2180
+ "step": 1550
2181
+ },
2182
+ {
2183
+ "epoch": 2.2277936962750715,
2184
+ "grad_norm": 0.9147911667823792,
2185
+ "learning_rate": 5.1332132805675165e-06,
2186
+ "loss": 0.088,
2187
+ "step": 1555
2188
+ },
2189
+ {
2190
+ "epoch": 2.2349570200573066,
2191
+ "grad_norm": 1.006125569343567,
2192
+ "learning_rate": 5.044296276448407e-06,
2193
+ "loss": 0.0753,
2194
+ "step": 1560
2195
+ },
2196
+ {
2197
+ "epoch": 2.2421203438395416,
2198
+ "grad_norm": 0.7414736747741699,
2199
+ "learning_rate": 4.956000198399266e-06,
2200
+ "loss": 0.1022,
2201
+ "step": 1565
2202
+ },
2203
+ {
2204
+ "epoch": 2.2492836676217767,
2205
+ "grad_norm": 0.9095657467842102,
2206
+ "learning_rate": 4.868330553347393e-06,
2207
+ "loss": 0.1069,
2208
+ "step": 1570
2209
+ },
2210
+ {
2211
+ "epoch": 2.2564469914040113,
2212
+ "grad_norm": 0.9873449802398682,
2213
+ "learning_rate": 4.7812928091501865e-06,
2214
+ "loss": 0.081,
2215
+ "step": 1575
2216
+ },
2217
+ {
2218
+ "epoch": 2.2636103151862463,
2219
+ "grad_norm": 1.2000373601913452,
2220
+ "learning_rate": 4.694892394254083e-06,
2221
+ "loss": 0.0802,
2222
+ "step": 1580
2223
+ },
2224
+ {
2225
+ "epoch": 2.2707736389684814,
2226
+ "grad_norm": 0.764998197555542,
2227
+ "learning_rate": 4.609134697356009e-06,
2228
+ "loss": 0.0856,
2229
+ "step": 1585
2230
+ },
2231
+ {
2232
+ "epoch": 2.2779369627507164,
2233
+ "grad_norm": 0.8754514455795288,
2234
+ "learning_rate": 4.5240250670672915e-06,
2235
+ "loss": 0.0892,
2236
+ "step": 1590
2237
+ },
2238
+ {
2239
+ "epoch": 2.285100286532951,
2240
+ "grad_norm": 0.7414029240608215,
2241
+ "learning_rate": 4.439568811580071e-06,
2242
+ "loss": 0.0901,
2243
+ "step": 1595
2244
+ },
2245
+ {
2246
+ "epoch": 2.292263610315186,
2247
+ "grad_norm": 0.9194661974906921,
2248
+ "learning_rate": 4.355771198336235e-06,
2249
+ "loss": 0.1029,
2250
+ "step": 1600
2251
+ },
2252
+ {
2253
+ "epoch": 2.299426934097421,
2254
+ "grad_norm": 0.8973520398139954,
2255
+ "learning_rate": 4.27263745369889e-06,
2256
+ "loss": 0.0855,
2257
+ "step": 1605
2258
+ },
2259
+ {
2260
+ "epoch": 2.306590257879656,
2261
+ "grad_norm": 0.8022074699401855,
2262
+ "learning_rate": 4.190172762626401e-06,
2263
+ "loss": 0.0897,
2264
+ "step": 1610
2265
+ },
2266
+ {
2267
+ "epoch": 2.3137535816618913,
2268
+ "grad_norm": 0.9117087125778198,
2269
+ "learning_rate": 4.108382268349026e-06,
2270
+ "loss": 0.0748,
2271
+ "step": 1615
2272
+ },
2273
+ {
2274
+ "epoch": 2.3209169054441263,
2275
+ "grad_norm": 0.8679396510124207,
2276
+ "learning_rate": 4.027271072048113e-06,
2277
+ "loss": 0.087,
2278
+ "step": 1620
2279
+ },
2280
+ {
2281
+ "epoch": 2.328080229226361,
2282
+ "grad_norm": 0.7531956434249878,
2283
+ "learning_rate": 3.946844232537958e-06,
2284
+ "loss": 0.082,
2285
+ "step": 1625
2286
+ },
2287
+ {
2288
+ "epoch": 2.335243553008596,
2289
+ "grad_norm": 0.7955215573310852,
2290
+ "learning_rate": 3.867106765950291e-06,
2291
+ "loss": 0.09,
2292
+ "step": 1630
2293
+ },
2294
+ {
2295
+ "epoch": 2.342406876790831,
2296
+ "grad_norm": 0.8443688750267029,
2297
+ "learning_rate": 3.7880636454214408e-06,
2298
+ "loss": 0.0958,
2299
+ "step": 1635
2300
+ },
2301
+ {
2302
+ "epoch": 2.349570200573066,
2303
+ "grad_norm": 0.7366068363189697,
2304
+ "learning_rate": 3.709719800782133e-06,
2305
+ "loss": 0.08,
2306
+ "step": 1640
2307
+ },
2308
+ {
2309
+ "epoch": 2.3567335243553007,
2310
+ "grad_norm": 1.0085781812667847,
2311
+ "learning_rate": 3.632080118250051e-06,
2312
+ "loss": 0.0953,
2313
+ "step": 1645
2314
+ },
2315
+ {
2316
+ "epoch": 2.3638968481375358,
2317
+ "grad_norm": 0.7856600284576416,
2318
+ "learning_rate": 3.5551494401250727e-06,
2319
+ "loss": 0.0866,
2320
+ "step": 1650
2321
+ },
2322
+ {
2323
+ "epoch": 2.371060171919771,
2324
+ "grad_norm": 0.7357582449913025,
2325
+ "learning_rate": 3.4789325644872695e-06,
2326
+ "loss": 0.0888,
2327
+ "step": 1655
2328
+ },
2329
+ {
2330
+ "epoch": 2.378223495702006,
2331
+ "grad_norm": 0.675508439540863,
2332
+ "learning_rate": 3.4034342448976533e-06,
2333
+ "loss": 0.083,
2334
+ "step": 1660
2335
+ },
2336
+ {
2337
+ "epoch": 2.385386819484241,
2338
+ "grad_norm": 0.7333245873451233,
2339
+ "learning_rate": 3.3286591901016956e-06,
2340
+ "loss": 0.0825,
2341
+ "step": 1665
2342
+ },
2343
+ {
2344
+ "epoch": 2.3925501432664755,
2345
+ "grad_norm": 0.7654566764831543,
2346
+ "learning_rate": 3.2546120637356676e-06,
2347
+ "loss": 0.0892,
2348
+ "step": 1670
2349
+ },
2350
+ {
2351
+ "epoch": 2.3997134670487106,
2352
+ "grad_norm": 0.7408206462860107,
2353
+ "learning_rate": 3.181297484035748e-06,
2354
+ "loss": 0.0843,
2355
+ "step": 1675
2356
+ },
2357
+ {
2358
+ "epoch": 2.4068767908309456,
2359
+ "grad_norm": 0.7130251526832581,
2360
+ "learning_rate": 3.108720023550005e-06,
2361
+ "loss": 0.0809,
2362
+ "step": 1680
2363
+ },
2364
+ {
2365
+ "epoch": 2.4140401146131807,
2366
+ "grad_norm": 0.7174748182296753,
2367
+ "learning_rate": 3.0368842088532146e-06,
2368
+ "loss": 0.0743,
2369
+ "step": 1685
2370
+ },
2371
+ {
2372
+ "epoch": 2.4212034383954153,
2373
+ "grad_norm": 0.8282214999198914,
2374
+ "learning_rate": 2.965794520264528e-06,
2375
+ "loss": 0.0814,
2376
+ "step": 1690
2377
+ },
2378
+ {
2379
+ "epoch": 2.4283667621776504,
2380
+ "grad_norm": 0.7426462769508362,
2381
+ "learning_rate": 2.8954553915680526e-06,
2382
+ "loss": 0.078,
2383
+ "step": 1695
2384
+ },
2385
+ {
2386
+ "epoch": 2.4355300859598854,
2387
+ "grad_norm": 0.7813041806221008,
2388
+ "learning_rate": 2.8258712097363215e-06,
2389
+ "loss": 0.0828,
2390
+ "step": 1700
2391
+ },
2392
+ {
2393
+ "epoch": 2.4426934097421205,
2394
+ "grad_norm": 0.657216489315033,
2395
+ "learning_rate": 2.757046314656676e-06,
2396
+ "loss": 0.0858,
2397
+ "step": 1705
2398
+ },
2399
+ {
2400
+ "epoch": 2.4498567335243555,
2401
+ "grad_norm": 0.8834379315376282,
2402
+ "learning_rate": 2.6889849988605914e-06,
2403
+ "loss": 0.0796,
2404
+ "step": 1710
2405
+ },
2406
+ {
2407
+ "epoch": 2.45702005730659,
2408
+ "grad_norm": 0.8047406077384949,
2409
+ "learning_rate": 2.6216915072559577e-06,
2410
+ "loss": 0.0773,
2411
+ "step": 1715
2412
+ },
2413
+ {
2414
+ "epoch": 2.464183381088825,
2415
+ "grad_norm": 0.6732385158538818,
2416
+ "learning_rate": 2.5551700368623455e-06,
2417
+ "loss": 0.0818,
2418
+ "step": 1720
2419
+ },
2420
+ {
2421
+ "epoch": 2.4713467048710602,
2422
+ "grad_norm": 0.6131224036216736,
2423
+ "learning_rate": 2.4894247365492155e-06,
2424
+ "loss": 0.0856,
2425
+ "step": 1725
2426
+ },
2427
+ {
2428
+ "epoch": 2.4785100286532953,
2429
+ "grad_norm": 0.6395134925842285,
2430
+ "learning_rate": 2.424459706777176e-06,
2431
+ "loss": 0.0714,
2432
+ "step": 1730
2433
+ },
2434
+ {
2435
+ "epoch": 2.48567335243553,
2436
+ "grad_norm": 0.7728132605552673,
2437
+ "learning_rate": 2.360278999342236e-06,
2438
+ "loss": 0.0788,
2439
+ "step": 1735
2440
+ },
2441
+ {
2442
+ "epoch": 2.492836676217765,
2443
+ "grad_norm": 0.6614881753921509,
2444
+ "learning_rate": 2.296886617123101e-06,
2445
+ "loss": 0.0776,
2446
+ "step": 1740
2447
+ },
2448
+ {
2449
+ "epoch": 2.5,
2450
+ "grad_norm": 1.1581789255142212,
2451
+ "learning_rate": 2.2342865138315187e-06,
2452
+ "loss": 0.0754,
2453
+ "step": 1745
2454
+ },
2455
+ {
2456
+ "epoch": 2.507163323782235,
2457
+ "grad_norm": 0.6569383144378662,
2458
+ "learning_rate": 2.1724825937656884e-06,
2459
+ "loss": 0.0758,
2460
+ "step": 1750
2461
+ },
2462
+ {
2463
+ "epoch": 2.51432664756447,
2464
+ "grad_norm": 0.7573304176330566,
2465
+ "learning_rate": 2.1114787115667476e-06,
2466
+ "loss": 0.0797,
2467
+ "step": 1755
2468
+ },
2469
+ {
2470
+ "epoch": 2.5214899713467047,
2471
+ "grad_norm": 0.888360857963562,
2472
+ "learning_rate": 2.0512786719783887e-06,
2473
+ "loss": 0.0818,
2474
+ "step": 1760
2475
+ },
2476
+ {
2477
+ "epoch": 2.5286532951289398,
2478
+ "grad_norm": 0.9539448618888855,
2479
+ "learning_rate": 1.991886229609522e-06,
2480
+ "loss": 0.0854,
2481
+ "step": 1765
2482
+ },
2483
+ {
2484
+ "epoch": 2.535816618911175,
2485
+ "grad_norm": 0.7275035977363586,
2486
+ "learning_rate": 1.9333050887001337e-06,
2487
+ "loss": 0.0736,
2488
+ "step": 1770
2489
+ },
2490
+ {
2491
+ "epoch": 2.54297994269341,
2492
+ "grad_norm": 0.750778079032898,
2493
+ "learning_rate": 1.875538902890256e-06,
2494
+ "loss": 0.0724,
2495
+ "step": 1775
2496
+ },
2497
+ {
2498
+ "epoch": 2.5501432664756445,
2499
+ "grad_norm": 0.7562493681907654,
2500
+ "learning_rate": 1.8185912749920708e-06,
2501
+ "loss": 0.0754,
2502
+ "step": 1780
2503
+ },
2504
+ {
2505
+ "epoch": 2.5573065902578795,
2506
+ "grad_norm": 0.7872411012649536,
2507
+ "learning_rate": 1.7624657567652347e-06,
2508
+ "loss": 0.0781,
2509
+ "step": 1785
2510
+ },
2511
+ {
2512
+ "epoch": 2.5644699140401146,
2513
+ "grad_norm": 0.9371019005775452,
2514
+ "learning_rate": 1.7071658486953373e-06,
2515
+ "loss": 0.0848,
2516
+ "step": 1790
2517
+ },
2518
+ {
2519
+ "epoch": 2.5716332378223496,
2520
+ "grad_norm": 0.7029364109039307,
2521
+ "learning_rate": 1.6526949997755947e-06,
2522
+ "loss": 0.0754,
2523
+ "step": 1795
2524
+ },
2525
+ {
2526
+ "epoch": 2.5787965616045847,
2527
+ "grad_norm": 0.7276107668876648,
2528
+ "learning_rate": 1.5990566072917312e-06,
2529
+ "loss": 0.0733,
2530
+ "step": 1800
2531
+ },
2532
+ {
2533
+ "epoch": 2.5859598853868198,
2534
+ "grad_norm": 0.7803687453269958,
2535
+ "learning_rate": 1.5462540166100992e-06,
2536
+ "loss": 0.0833,
2537
+ "step": 1805
2538
+ },
2539
+ {
2540
+ "epoch": 2.5931232091690544,
2541
+ "grad_norm": 0.7932295799255371,
2542
+ "learning_rate": 1.494290520969029e-06,
2543
+ "loss": 0.0794,
2544
+ "step": 1810
2545
+ },
2546
+ {
2547
+ "epoch": 2.6002865329512894,
2548
+ "grad_norm": 0.8229401111602783,
2549
+ "learning_rate": 1.443169361273438e-06,
2550
+ "loss": 0.0711,
2551
+ "step": 1815
2552
+ },
2553
+ {
2554
+ "epoch": 2.6074498567335245,
2555
+ "grad_norm": 0.6098470091819763,
2556
+ "learning_rate": 1.392893725892689e-06,
2557
+ "loss": 0.0691,
2558
+ "step": 1820
2559
+ },
2560
+ {
2561
+ "epoch": 2.614613180515759,
2562
+ "grad_norm": 0.7762227654457092,
2563
+ "learning_rate": 1.3434667504617526e-06,
2564
+ "loss": 0.0902,
2565
+ "step": 1825
2566
+ },
2567
+ {
2568
+ "epoch": 2.621776504297994,
2569
+ "grad_norm": 0.7130775451660156,
2570
+ "learning_rate": 1.2948915176856208e-06,
2571
+ "loss": 0.0729,
2572
+ "step": 1830
2573
+ },
2574
+ {
2575
+ "epoch": 2.628939828080229,
2576
+ "grad_norm": 0.7970885634422302,
2577
+ "learning_rate": 1.2471710571470579e-06,
2578
+ "loss": 0.0785,
2579
+ "step": 1835
2580
+ },
2581
+ {
2582
+ "epoch": 2.6361031518624642,
2583
+ "grad_norm": 0.8026163578033447,
2584
+ "learning_rate": 1.2003083451176366e-06,
2585
+ "loss": 0.0796,
2586
+ "step": 1840
2587
+ },
2588
+ {
2589
+ "epoch": 2.6432664756446993,
2590
+ "grad_norm": 0.7174281477928162,
2591
+ "learning_rate": 1.1543063043721258e-06,
2592
+ "loss": 0.0718,
2593
+ "step": 1845
2594
+ },
2595
+ {
2596
+ "epoch": 2.6504297994269344,
2597
+ "grad_norm": 1.2479575872421265,
2598
+ "learning_rate": 1.1091678040061865e-06,
2599
+ "loss": 0.0888,
2600
+ "step": 1850
2601
+ },
2602
+ {
2603
+ "epoch": 2.657593123209169,
2604
+ "grad_norm": 0.5801442265510559,
2605
+ "learning_rate": 1.064895659257435e-06,
2606
+ "loss": 0.0657,
2607
+ "step": 1855
2608
+ },
2609
+ {
2610
+ "epoch": 2.664756446991404,
2611
+ "grad_norm": 0.7601532340049744,
2612
+ "learning_rate": 1.021492631329865e-06,
2613
+ "loss": 0.069,
2614
+ "step": 1860
2615
+ },
2616
+ {
2617
+ "epoch": 2.671919770773639,
2618
+ "grad_norm": 0.7267557382583618,
2619
+ "learning_rate": 9.789614272216207e-07,
2620
+ "loss": 0.0766,
2621
+ "step": 1865
2622
+ },
2623
+ {
2624
+ "epoch": 2.6790830945558737,
2625
+ "grad_norm": 0.6664993166923523,
2626
+ "learning_rate": 9.373046995561829e-07,
2627
+ "loss": 0.0816,
2628
+ "step": 1870
2629
+ },
2630
+ {
2631
+ "epoch": 2.6862464183381087,
2632
+ "grad_norm": 0.6393699645996094,
2633
+ "learning_rate": 8.965250464169133e-07,
2634
+ "loss": 0.0763,
2635
+ "step": 1875
2636
+ },
2637
+ {
2638
+ "epoch": 2.693409742120344,
2639
+ "grad_norm": 0.6990323066711426,
2640
+ "learning_rate": 8.566250111850143e-07,
2641
+ "loss": 0.0805,
2642
+ "step": 1880
2643
+ },
2644
+ {
2645
+ "epoch": 2.700573065902579,
2646
+ "grad_norm": 0.6114766001701355,
2647
+ "learning_rate": 8.176070823809173e-07,
2648
+ "loss": 0.0756,
2649
+ "step": 1885
2650
+ },
2651
+ {
2652
+ "epoch": 2.707736389684814,
2653
+ "grad_norm": 0.7426799535751343,
2654
+ "learning_rate": 7.794736935090585e-07,
2655
+ "loss": 0.0799,
2656
+ "step": 1890
2657
+ },
2658
+ {
2659
+ "epoch": 2.714899713467049,
2660
+ "grad_norm": 0.8868831992149353,
2661
+ "learning_rate": 7.422272229061106e-07,
2662
+ "loss": 0.0788,
2663
+ "step": 1895
2664
+ },
2665
+ {
2666
+ "epoch": 2.7220630372492836,
2667
+ "grad_norm": 0.7065321207046509,
2668
+ "learning_rate": 7.058699935926527e-07,
2669
+ "loss": 0.082,
2670
+ "step": 1900
2671
+ },
2672
+ {
2673
+ "epoch": 2.7292263610315186,
2674
+ "grad_norm": 0.7753829956054688,
2675
+ "learning_rate": 6.704042731282784e-07,
2676
+ "loss": 0.0832,
2677
+ "step": 1905
2678
+ },
2679
+ {
2680
+ "epoch": 2.7363896848137537,
2681
+ "grad_norm": 0.6844934225082397,
2682
+ "learning_rate": 6.358322734701799e-07,
2683
+ "loss": 0.0694,
2684
+ "step": 1910
2685
+ },
2686
+ {
2687
+ "epoch": 2.7435530085959883,
2688
+ "grad_norm": 0.6393133997917175,
2689
+ "learning_rate": 6.021561508351808e-07,
2690
+ "loss": 0.0785,
2691
+ "step": 1915
2692
+ },
2693
+ {
2694
+ "epoch": 2.7507163323782233,
2695
+ "grad_norm": 0.7742157578468323,
2696
+ "learning_rate": 5.693780055652615e-07,
2697
+ "loss": 0.0754,
2698
+ "step": 1920
2699
+ },
2700
+ {
2701
+ "epoch": 2.7578796561604584,
2702
+ "grad_norm": 0.6114529371261597,
2703
+ "learning_rate": 5.374998819965654e-07,
2704
+ "loss": 0.0594,
2705
+ "step": 1925
2706
+ },
2707
+ {
2708
+ "epoch": 2.7650429799426934,
2709
+ "grad_norm": 0.6473120450973511,
2710
+ "learning_rate": 5.065237683318852e-07,
2711
+ "loss": 0.0801,
2712
+ "step": 1930
2713
+ },
2714
+ {
2715
+ "epoch": 2.7722063037249285,
2716
+ "grad_norm": 0.6858311295509338,
2717
+ "learning_rate": 4.7645159651667435e-07,
2718
+ "loss": 0.0761,
2719
+ "step": 1935
2720
+ },
2721
+ {
2722
+ "epoch": 2.7793696275071635,
2723
+ "grad_norm": 0.6852824091911316,
2724
+ "learning_rate": 4.472852421185425e-07,
2725
+ "loss": 0.0741,
2726
+ "step": 1940
2727
+ },
2728
+ {
2729
+ "epoch": 2.786532951289398,
2730
+ "grad_norm": 0.7606457471847534,
2731
+ "learning_rate": 4.1902652421029033e-07,
2732
+ "loss": 0.0823,
2733
+ "step": 1945
2734
+ },
2735
+ {
2736
+ "epoch": 2.793696275071633,
2737
+ "grad_norm": 0.7371909022331238,
2738
+ "learning_rate": 3.9167720525644044e-07,
2739
+ "loss": 0.0856,
2740
+ "step": 1950
2741
+ },
2742
+ {
2743
+ "epoch": 2.8008595988538683,
2744
+ "grad_norm": 0.6579184532165527,
2745
+ "learning_rate": 3.652389910033305e-07,
2746
+ "loss": 0.0627,
2747
+ "step": 1955
2748
+ },
2749
+ {
2750
+ "epoch": 2.8080229226361033,
2751
+ "grad_norm": 0.7221111059188843,
2752
+ "learning_rate": 3.397135303727161e-07,
2753
+ "loss": 0.0779,
2754
+ "step": 1960
2755
+ },
2756
+ {
2757
+ "epoch": 2.815186246418338,
2758
+ "grad_norm": 0.7001911997795105,
2759
+ "learning_rate": 3.1510241535893215e-07,
2760
+ "loss": 0.0877,
2761
+ "step": 1965
2762
+ },
2763
+ {
2764
+ "epoch": 2.822349570200573,
2765
+ "grad_norm": 0.77513587474823,
2766
+ "learning_rate": 2.9140718092960985e-07,
2767
+ "loss": 0.0744,
2768
+ "step": 1970
2769
+ },
2770
+ {
2771
+ "epoch": 2.829512893982808,
2772
+ "grad_norm": 0.7423258423805237,
2773
+ "learning_rate": 2.6862930492993043e-07,
2774
+ "loss": 0.0759,
2775
+ "step": 1975
2776
+ },
2777
+ {
2778
+ "epoch": 2.836676217765043,
2779
+ "grad_norm": 0.5900893211364746,
2780
+ "learning_rate": 2.4677020799046044e-07,
2781
+ "loss": 0.0687,
2782
+ "step": 1980
2783
+ },
2784
+ {
2785
+ "epoch": 2.843839541547278,
2786
+ "grad_norm": 0.6805256009101868,
2787
+ "learning_rate": 2.2583125343854927e-07,
2788
+ "loss": 0.0715,
2789
+ "step": 1985
2790
+ },
2791
+ {
2792
+ "epoch": 2.8510028653295127,
2793
+ "grad_norm": 0.7028383612632751,
2794
+ "learning_rate": 2.058137472132954e-07,
2795
+ "loss": 0.0667,
2796
+ "step": 1990
2797
+ },
2798
+ {
2799
+ "epoch": 2.858166189111748,
2800
+ "grad_norm": 0.6828755140304565,
2801
+ "learning_rate": 1.8671893778410166e-07,
2802
+ "loss": 0.0616,
2803
+ "step": 1995
2804
+ },
2805
+ {
2806
+ "epoch": 2.865329512893983,
2807
+ "grad_norm": 0.6552565097808838,
2808
+ "learning_rate": 1.685480160728059e-07,
2809
+ "loss": 0.0713,
2810
+ "step": 2000
2811
+ },
2812
+ {
2813
+ "epoch": 2.872492836676218,
2814
+ "grad_norm": 0.7110970616340637,
2815
+ "learning_rate": 1.5130211537940364e-07,
2816
+ "loss": 0.0772,
2817
+ "step": 2005
2818
+ },
2819
+ {
2820
+ "epoch": 2.8796561604584525,
2821
+ "grad_norm": 0.5577581524848938,
2822
+ "learning_rate": 1.3498231131137296e-07,
2823
+ "loss": 0.0675,
2824
+ "step": 2010
2825
+ },
2826
+ {
2827
+ "epoch": 2.8868194842406876,
2828
+ "grad_norm": 0.741245687007904,
2829
+ "learning_rate": 1.1958962171657984e-07,
2830
+ "loss": 0.0781,
2831
+ "step": 2015
2832
+ },
2833
+ {
2834
+ "epoch": 2.8939828080229226,
2835
+ "grad_norm": 0.6453272700309753,
2836
+ "learning_rate": 1.0512500661980063e-07,
2837
+ "loss": 0.0731,
2838
+ "step": 2020
2839
+ },
2840
+ {
2841
+ "epoch": 2.9011461318051577,
2842
+ "grad_norm": 0.9927032589912415,
2843
+ "learning_rate": 9.158936816284824e-08,
2844
+ "loss": 0.0744,
2845
+ "step": 2025
2846
+ },
2847
+ {
2848
+ "epoch": 2.9083094555873927,
2849
+ "grad_norm": 0.6868733763694763,
2850
+ "learning_rate": 7.898355054830719e-08,
2851
+ "loss": 0.0698,
2852
+ "step": 2030
2853
+ },
2854
+ {
2855
+ "epoch": 2.9154727793696273,
2856
+ "grad_norm": 0.6451588869094849,
2857
+ "learning_rate": 6.730833998687569e-08,
2858
+ "loss": 0.0712,
2859
+ "step": 2035
2860
+ },
2861
+ {
2862
+ "epoch": 2.9226361031518624,
2863
+ "grad_norm": 0.5942579507827759,
2864
+ "learning_rate": 5.656446464833653e-08,
2865
+ "loss": 0.0675,
2866
+ "step": 2040
2867
+ },
2868
+ {
2869
+ "epoch": 2.9297994269340975,
2870
+ "grad_norm": 0.7502886056900024,
2871
+ "learning_rate": 4.675259461613679e-08,
2872
+ "loss": 0.0702,
2873
+ "step": 2045
2874
+ },
2875
+ {
2876
+ "epoch": 2.9369627507163325,
2877
+ "grad_norm": 0.6487383246421814,
2878
+ "learning_rate": 3.787334184560287e-08,
2879
+ "loss": 0.0768,
2880
+ "step": 2050
2881
+ },
2882
+ {
2883
+ "epoch": 2.944126074498567,
2884
+ "grad_norm": 0.6319767236709595,
2885
+ "learning_rate": 2.992726012576607e-08,
2886
+ "loss": 0.0687,
2887
+ "step": 2055
2888
+ },
2889
+ {
2890
+ "epoch": 2.951289398280802,
2891
+ "grad_norm": 0.6914830207824707,
2892
+ "learning_rate": 2.2914845044826883e-08,
2893
+ "loss": 0.0728,
2894
+ "step": 2060
2895
+ },
2896
+ {
2897
+ "epoch": 2.958452722063037,
2898
+ "grad_norm": 0.8084492087364197,
2899
+ "learning_rate": 1.683653395924467e-08,
2900
+ "loss": 0.0716,
2901
+ "step": 2065
2902
+ },
2903
+ {
2904
+ "epoch": 2.9656160458452723,
2905
+ "grad_norm": 0.7632797360420227,
2906
+ "learning_rate": 1.1692705966461214e-08,
2907
+ "loss": 0.0724,
2908
+ "step": 2070
2909
+ },
2910
+ {
2911
+ "epoch": 2.9727793696275073,
2912
+ "grad_norm": 0.7611112594604492,
2913
+ "learning_rate": 7.48368188125792e-09,
2914
+ "loss": 0.0762,
2915
+ "step": 2075
2916
+ },
2917
+ {
2918
+ "epoch": 2.9799426934097424,
2919
+ "grad_norm": 0.789293646812439,
2920
+ "learning_rate": 4.209724215743504e-09,
2921
+ "loss": 0.0762,
2922
+ "step": 2080
2923
+ },
2924
+ {
2925
+ "epoch": 2.987106017191977,
2926
+ "grad_norm": 0.7104436159133911,
2927
+ "learning_rate": 1.8710371629837575e-09,
2928
+ "loss": 0.0682,
2929
+ "step": 2085
2930
+ },
2931
+ {
2932
+ "epoch": 2.994269340974212,
2933
+ "grad_norm": 0.6485856771469116,
2934
+ "learning_rate": 4.677665842650613e-10,
2935
+ "loss": 0.0712,
2936
+ "step": 2090
2937
+ }
2938
+ ],
2939
+ "logging_steps": 5,
2940
+ "max_steps": 2094,
2941
+ "num_input_tokens_seen": 0,
2942
+ "num_train_epochs": 3,
2943
+ "save_steps": 2000,
2944
+ "stateful_callbacks": {
2945
+ "TrainerControl": {
2946
+ "args": {
2947
+ "should_epoch_stop": false,
2948
+ "should_evaluate": false,
2949
+ "should_log": false,
2950
+ "should_save": true,
2951
+ "should_training_stop": true
2952
+ },
2953
+ "attributes": {}
2954
+ }
2955
+ },
2956
+ "total_flos": 2.834346006366126e+18,
2957
+ "train_batch_size": 2,
2958
+ "trial_name": null,
2959
+ "trial_params": null
2960
+ }
132_128_e3_3e-5/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6e69f666717339f2d7f2d89115c1f0ded9f2163adb49892605a0e4131352705
3
+ size 8209
132_128_e3_3e-5/vocab.json ADDED
The diff for this file is too large to render. See raw diff
 
132_128_e3_3e-5/zero_to_fp32.py ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+
3
+ # Copyright (c) Microsoft Corporation.
4
+ # SPDX-License-Identifier: Apache-2.0
5
+
6
+ # DeepSpeed Team
7
+
8
+ # This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
9
+ # copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
10
+ # the future. Once extracted, the weights don't require DeepSpeed and can be used in any
11
+ # application.
12
+ #
13
+ # example: python zero_to_fp32.py . pytorch_model.bin
14
+
15
+ import argparse
16
+ import torch
17
+ import glob
18
+ import math
19
+ import os
20
+ import re
21
+ from collections import OrderedDict
22
+ from dataclasses import dataclass
23
+
24
+ # while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
25
+ # DeepSpeed data structures it has to be available in the current python environment.
26
+ from deepspeed.utils import logger
27
+ from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
28
+ FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
29
+ FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
30
+
31
+
32
+ @dataclass
33
+ class zero_model_state:
34
+ buffers: dict()
35
+ param_shapes: dict()
36
+ shared_params: list
37
+ ds_version: int
38
+ frozen_param_shapes: dict()
39
+ frozen_param_fragments: dict()
40
+
41
+
42
+ debug = 0
43
+
44
+ # load to cpu
45
+ device = torch.device('cpu')
46
+
47
+
48
+ def atoi(text):
49
+ return int(text) if text.isdigit() else text
50
+
51
+
52
+ def natural_keys(text):
53
+ '''
54
+ alist.sort(key=natural_keys) sorts in human order
55
+ http://nedbatchelder.com/blog/200712/human_sorting.html
56
+ (See Toothy's implementation in the comments)
57
+ '''
58
+ return [atoi(c) for c in re.split(r'(\d+)', text)]
59
+
60
+
61
+ def get_model_state_file(checkpoint_dir, zero_stage):
62
+ if not os.path.isdir(checkpoint_dir):
63
+ raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
64
+
65
+ # there should be only one file
66
+ if zero_stage <= 2:
67
+ file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
68
+ elif zero_stage == 3:
69
+ file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
70
+
71
+ if not os.path.exists(file):
72
+ raise FileNotFoundError(f"can't find model states file at '{file}'")
73
+
74
+ return file
75
+
76
+
77
+ def get_checkpoint_files(checkpoint_dir, glob_pattern):
78
+ # XXX: need to test that this simple glob rule works for multi-node setup too
79
+ ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
80
+
81
+ if len(ckpt_files) == 0:
82
+ raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
83
+
84
+ return ckpt_files
85
+
86
+
87
+ def get_optim_files(checkpoint_dir):
88
+ return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
89
+
90
+
91
+ def get_model_state_files(checkpoint_dir):
92
+ return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
93
+
94
+
95
+ def parse_model_states(files):
96
+ zero_model_states = []
97
+ for file in files:
98
+ state_dict = torch.load(file, map_location=device)
99
+
100
+ if BUFFER_NAMES not in state_dict:
101
+ raise ValueError(f"{file} is not a model state checkpoint")
102
+ buffer_names = state_dict[BUFFER_NAMES]
103
+ if debug:
104
+ print("Found buffers:", buffer_names)
105
+
106
+ # recover just the buffers while restoring them to fp32 if they were saved in fp16
107
+ buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
108
+ param_shapes = state_dict[PARAM_SHAPES]
109
+
110
+ # collect parameters that are included in param_shapes
111
+ param_names = []
112
+ for s in param_shapes:
113
+ for name in s.keys():
114
+ param_names.append(name)
115
+
116
+ # update with frozen parameters
117
+ frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
118
+ if frozen_param_shapes is not None:
119
+ if debug:
120
+ print(f"Found frozen_param_shapes: {frozen_param_shapes}")
121
+ param_names += list(frozen_param_shapes.keys())
122
+
123
+ # handle shared params
124
+ shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
125
+
126
+ ds_version = state_dict.get(DS_VERSION, None)
127
+
128
+ frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
129
+
130
+ z_model_state = zero_model_state(buffers=buffers,
131
+ param_shapes=param_shapes,
132
+ shared_params=shared_params,
133
+ ds_version=ds_version,
134
+ frozen_param_shapes=frozen_param_shapes,
135
+ frozen_param_fragments=frozen_param_fragments)
136
+ zero_model_states.append(z_model_state)
137
+
138
+ return zero_model_states
139
+
140
+
141
+ def parse_optim_states(files, ds_checkpoint_dir):
142
+
143
+ total_files = len(files)
144
+ state_dicts = []
145
+ for f in files:
146
+ state_dict = torch.load(f, map_location=device)
147
+ # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
148
+ # and also handle the case where it was already removed by another helper script
149
+ state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
150
+ state_dicts.append(state_dict)
151
+
152
+ if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
153
+ raise ValueError(f"{files[0]} is not a zero checkpoint")
154
+ zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
155
+ world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
156
+
157
+ # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
158
+ # parameters can be different from data parallelism for non-expert parameters. So we can just
159
+ # use the max of the partition_count to get the dp world_size.
160
+
161
+ if type(world_size) is list:
162
+ world_size = max(world_size)
163
+
164
+ if world_size != total_files:
165
+ raise ValueError(
166
+ f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
167
+ "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
168
+ )
169
+
170
+ # the groups are named differently in each stage
171
+ if zero_stage <= 2:
172
+ fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
173
+ elif zero_stage == 3:
174
+ fp32_groups_key = FP32_FLAT_GROUPS
175
+ else:
176
+ raise ValueError(f"unknown zero stage {zero_stage}")
177
+
178
+ if zero_stage <= 2:
179
+ fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
180
+ elif zero_stage == 3:
181
+ # if there is more than one param group, there will be multiple flattened tensors - one
182
+ # flattened tensor per group - for simplicity merge them into a single tensor
183
+ #
184
+ # XXX: could make the script more memory efficient for when there are multiple groups - it
185
+ # will require matching the sub-lists of param_shapes for each param group flattened tensor
186
+
187
+ fp32_flat_groups = [
188
+ torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
189
+ ]
190
+
191
+ return zero_stage, world_size, fp32_flat_groups
192
+
193
+
194
+ def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
195
+ """
196
+ Returns fp32 state_dict reconstructed from ds checkpoint
197
+
198
+ Args:
199
+ - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
200
+
201
+ """
202
+ print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
203
+
204
+ optim_files = get_optim_files(ds_checkpoint_dir)
205
+ zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
206
+ print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
207
+
208
+ model_files = get_model_state_files(ds_checkpoint_dir)
209
+
210
+ zero_model_states = parse_model_states(model_files)
211
+ print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
212
+
213
+ if zero_stage <= 2:
214
+ return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
215
+ exclude_frozen_parameters)
216
+ elif zero_stage == 3:
217
+ return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
218
+ exclude_frozen_parameters)
219
+
220
+
221
+ def _zero2_merge_frozen_params(state_dict, zero_model_states):
222
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
223
+ return
224
+
225
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
226
+ frozen_param_fragments = zero_model_states[0].frozen_param_fragments
227
+
228
+ if debug:
229
+ num_elem = sum(s.numel() for s in frozen_param_shapes.values())
230
+ print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
231
+
232
+ wanted_params = len(frozen_param_shapes)
233
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
234
+ avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
235
+ print(f'Frozen params: Have {avail_numel} numels to process.')
236
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
237
+
238
+ total_params = 0
239
+ total_numel = 0
240
+ for name, shape in frozen_param_shapes.items():
241
+ total_params += 1
242
+ unpartitioned_numel = shape.numel()
243
+ total_numel += unpartitioned_numel
244
+
245
+ state_dict[name] = frozen_param_fragments[name]
246
+
247
+ if debug:
248
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
249
+
250
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
251
+
252
+
253
+ def _has_callable(obj, fn):
254
+ attr = getattr(obj, fn, None)
255
+ return callable(attr)
256
+
257
+
258
+ def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
259
+ param_shapes = zero_model_states[0].param_shapes
260
+
261
+ # Reconstruction protocol:
262
+ #
263
+ # XXX: document this
264
+
265
+ if debug:
266
+ for i in range(world_size):
267
+ for j in range(len(fp32_flat_groups[0])):
268
+ print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
269
+
270
+ # XXX: memory usage doubles here (zero2)
271
+ num_param_groups = len(fp32_flat_groups[0])
272
+ merged_single_partition_of_fp32_groups = []
273
+ for i in range(num_param_groups):
274
+ merged_partitions = [sd[i] for sd in fp32_flat_groups]
275
+ full_single_fp32_vector = torch.cat(merged_partitions, 0)
276
+ merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
277
+ avail_numel = sum(
278
+ [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
279
+
280
+ if debug:
281
+ wanted_params = sum([len(shapes) for shapes in param_shapes])
282
+ wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
283
+ # not asserting if there is a mismatch due to possible padding
284
+ print(f"Have {avail_numel} numels to process.")
285
+ print(f"Need {wanted_numel} numels in {wanted_params} params.")
286
+
287
+ # params
288
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
289
+ # out-of-core computing solution
290
+ total_numel = 0
291
+ total_params = 0
292
+ for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
293
+ offset = 0
294
+ avail_numel = full_single_fp32_vector.numel()
295
+ for name, shape in shapes.items():
296
+
297
+ unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
298
+ total_numel += unpartitioned_numel
299
+ total_params += 1
300
+
301
+ if debug:
302
+ print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
303
+ state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
304
+ offset += unpartitioned_numel
305
+
306
+ # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
307
+ # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
308
+ # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
309
+ # live optimizer object, so we are checking that the numbers are within the right range
310
+ align_to = 2 * world_size
311
+
312
+ def zero2_align(x):
313
+ return align_to * math.ceil(x / align_to)
314
+
315
+ if debug:
316
+ print(f"original offset={offset}, avail_numel={avail_numel}")
317
+
318
+ offset = zero2_align(offset)
319
+ avail_numel = zero2_align(avail_numel)
320
+
321
+ if debug:
322
+ print(f"aligned offset={offset}, avail_numel={avail_numel}")
323
+
324
+ # Sanity check
325
+ if offset != avail_numel:
326
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
327
+
328
+ print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
329
+
330
+
331
+ def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
332
+ exclude_frozen_parameters):
333
+ state_dict = OrderedDict()
334
+
335
+ # buffers
336
+ buffers = zero_model_states[0].buffers
337
+ state_dict.update(buffers)
338
+ if debug:
339
+ print(f"added {len(buffers)} buffers")
340
+
341
+ if not exclude_frozen_parameters:
342
+ _zero2_merge_frozen_params(state_dict, zero_model_states)
343
+
344
+ _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
345
+
346
+ # recover shared parameters
347
+ for pair in zero_model_states[0].shared_params:
348
+ if pair[1] in state_dict:
349
+ state_dict[pair[0]] = state_dict[pair[1]]
350
+
351
+ return state_dict
352
+
353
+
354
+ def zero3_partitioned_param_info(unpartitioned_numel, world_size):
355
+ remainder = unpartitioned_numel % world_size
356
+ padding_numel = (world_size - remainder) if remainder else 0
357
+ partitioned_numel = math.ceil(unpartitioned_numel / world_size)
358
+ return partitioned_numel, padding_numel
359
+
360
+
361
+ def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
362
+ if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
363
+ return
364
+
365
+ if debug:
366
+ for i in range(world_size):
367
+ num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
368
+ print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
369
+
370
+ frozen_param_shapes = zero_model_states[0].frozen_param_shapes
371
+ wanted_params = len(frozen_param_shapes)
372
+ wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
373
+ avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
374
+ print(f'Frozen params: Have {avail_numel} numels to process.')
375
+ print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
376
+
377
+ total_params = 0
378
+ total_numel = 0
379
+ for name, shape in zero_model_states[0].frozen_param_shapes.items():
380
+ total_params += 1
381
+ unpartitioned_numel = shape.numel()
382
+ total_numel += unpartitioned_numel
383
+
384
+ param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
385
+ state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
386
+
387
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
388
+
389
+ if debug:
390
+ print(
391
+ f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
392
+ )
393
+
394
+ print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
395
+
396
+
397
+ def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
398
+ param_shapes = zero_model_states[0].param_shapes
399
+ avail_numel = fp32_flat_groups[0].numel() * world_size
400
+ # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
401
+ # param, re-consolidating each param, while dealing with padding if any
402
+
403
+ # merge list of dicts, preserving order
404
+ param_shapes = {k: v for d in param_shapes for k, v in d.items()}
405
+
406
+ if debug:
407
+ for i in range(world_size):
408
+ print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
409
+
410
+ wanted_params = len(param_shapes)
411
+ wanted_numel = sum(shape.numel() for shape in param_shapes.values())
412
+ # not asserting if there is a mismatch due to possible padding
413
+ avail_numel = fp32_flat_groups[0].numel() * world_size
414
+ print(f"Trainable params: Have {avail_numel} numels to process.")
415
+ print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
416
+
417
+ # params
418
+ # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
419
+ # out-of-core computing solution
420
+ offset = 0
421
+ total_numel = 0
422
+ total_params = 0
423
+ for name, shape in param_shapes.items():
424
+
425
+ unpartitioned_numel = shape.numel()
426
+ total_numel += unpartitioned_numel
427
+ total_params += 1
428
+
429
+ partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
430
+
431
+ if debug:
432
+ print(
433
+ f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
434
+ )
435
+
436
+ # XXX: memory usage doubles here
437
+ state_dict[name] = torch.cat(
438
+ tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
439
+ 0).narrow(0, 0, unpartitioned_numel).view(shape)
440
+ offset += partitioned_numel
441
+
442
+ offset *= world_size
443
+
444
+ # Sanity check
445
+ if offset != avail_numel:
446
+ raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
447
+
448
+ print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
449
+
450
+
451
+ def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
452
+ exclude_frozen_parameters):
453
+ state_dict = OrderedDict()
454
+
455
+ # buffers
456
+ buffers = zero_model_states[0].buffers
457
+ state_dict.update(buffers)
458
+ if debug:
459
+ print(f"added {len(buffers)} buffers")
460
+
461
+ if not exclude_frozen_parameters:
462
+ _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
463
+
464
+ _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
465
+
466
+ # recover shared parameters
467
+ for pair in zero_model_states[0].shared_params:
468
+ if pair[1] in state_dict:
469
+ state_dict[pair[0]] = state_dict[pair[1]]
470
+
471
+ return state_dict
472
+
473
+
474
+ def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
475
+ """
476
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
477
+ ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
478
+ via a model hub.
479
+
480
+ Args:
481
+ - ``checkpoint_dir``: path to the desired checkpoint folder
482
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
483
+ - ``exclude_frozen_parameters``: exclude frozen parameters
484
+
485
+ Returns:
486
+ - pytorch ``state_dict``
487
+
488
+ Note: this approach may not work if your application doesn't have sufficient free CPU memory and
489
+ you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
490
+ the checkpoint.
491
+
492
+ A typical usage might be ::
493
+
494
+ from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
495
+ # do the training and checkpoint saving
496
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
497
+ model = model.cpu() # move to cpu
498
+ model.load_state_dict(state_dict)
499
+ # submit to model hub or save the model to share with others
500
+
501
+ In this example the ``model`` will no longer be usable in the deepspeed context of the same
502
+ application. i.e. you will need to re-initialize the deepspeed engine, since
503
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
504
+
505
+ If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
506
+
507
+ """
508
+ if tag is None:
509
+ latest_path = os.path.join(checkpoint_dir, 'latest')
510
+ if os.path.isfile(latest_path):
511
+ with open(latest_path, 'r') as fd:
512
+ tag = fd.read().strip()
513
+ else:
514
+ raise ValueError(f"Unable to find 'latest' file at {latest_path}")
515
+
516
+ ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
517
+
518
+ if not os.path.isdir(ds_checkpoint_dir):
519
+ raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
520
+
521
+ return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
522
+
523
+
524
+ def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
525
+ """
526
+ Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
527
+ loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
528
+
529
+ Args:
530
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
531
+ - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
532
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
533
+ - ``exclude_frozen_parameters``: exclude frozen parameters
534
+ """
535
+
536
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
537
+ print(f"Saving fp32 state dict to {output_file}")
538
+ torch.save(state_dict, output_file)
539
+
540
+
541
+ def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
542
+ """
543
+ 1. Put the provided model to cpu
544
+ 2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
545
+ 3. Load it into the provided model
546
+
547
+ Args:
548
+ - ``model``: the model object to update
549
+ - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
550
+ - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
551
+
552
+ Returns:
553
+ - ``model`: modified model
554
+
555
+ Make sure you have plenty of CPU memory available before you call this function. If you don't
556
+ have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
557
+ conveniently placed for you in the checkpoint folder.
558
+
559
+ A typical usage might be ::
560
+
561
+ from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
562
+ model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
563
+ # submit to model hub or save the model to share with others
564
+
565
+ Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
566
+ of the same application. i.e. you will need to re-initialize the deepspeed engine, since
567
+ ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
568
+
569
+ """
570
+ logger.info(f"Extracting fp32 weights")
571
+ state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
572
+
573
+ logger.info(f"Overwriting model with fp32 weights")
574
+ model = model.cpu()
575
+ model.load_state_dict(state_dict, strict=False)
576
+
577
+ return model
578
+
579
+
580
+ if __name__ == "__main__":
581
+
582
+ parser = argparse.ArgumentParser()
583
+ parser.add_argument("checkpoint_dir",
584
+ type=str,
585
+ help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
586
+ parser.add_argument(
587
+ "output_file",
588
+ type=str,
589
+ help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
590
+ parser.add_argument("-t",
591
+ "--tag",
592
+ type=str,
593
+ default=None,
594
+ help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
595
+ parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
596
+ parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
597
+ args = parser.parse_args()
598
+
599
+ debug = args.debug
600
+
601
+ convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
602
+ args.output_file,
603
+ tag=args.tag,
604
+ exclude_frozen_parameters=args.exclude_frozen_parameters)