Hugging Face's logo

IsmaelRR
/
SusurroModel-WhisperTurboV3Spanish

Safetensors
Spanish
whisper
susurro
audio
Model card Files
xet
 Community
IsmaelRR commited on
Commit
b5e21ae
·
verified ·
1 Parent(s): d9058e4

Upload 9 files

Browse files
Files changed (9) hide show
  1. config.json +49 -0
  2. generation_config.json +249 -0
  3. model.safetensors +3 -0
  4. optimizer.pt +3 -0
  5. preprocessor_config.json +14 -0
  6. rng_state.pth +3 -0
  7. scheduler.pt +3 -0
  8. trainer_state.json +3693 -0
  9. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-large-v3-turbo",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50256
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 1280,
17
+ "decoder_attention_heads": 20,
18
+ "decoder_ffn_dim": 5120,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 4,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 20,
24
+ "encoder_ffn_dim": 5120,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 32,
27
+ "eos_token_id": 50257,
28
+ "init_std": 0.02,
29
+ "is_encoder_decoder": true,
30
+ "mask_feature_length": 10,
31
+ "mask_feature_min_masks": 0,
32
+ "mask_feature_prob": 0.0,
33
+ "mask_time_length": 10,
34
+ "mask_time_min_masks": 2,
35
+ "mask_time_prob": 0.05,
36
+ "max_source_positions": 1500,
37
+ "max_target_positions": 448,
38
+ "median_filter_width": 7,
39
+ "model_type": "whisper",
40
+ "num_hidden_layers": 32,
41
+ "num_mel_bins": 128,
42
+ "pad_token_id": 50257,
43
+ "scale_embedding": false,
44
+ "torch_dtype": "float32",
45
+ "transformers_version": "4.44.2",
46
+ "use_cache": true,
47
+ "use_weighted_layer_sum": false,
48
+ "vocab_size": 51866
49
+ }
generation_config.json ADDED
@@ -0,0 +1,249 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alignment_heads": [
3
+ [
4
+ 2,
5
+ 4
6
+ ],
7
+ [
8
+ 2,
9
+ 11
10
+ ],
11
+ [
12
+ 3,
13
+ 3
14
+ ],
15
+ [
16
+ 3,
17
+ 6
18
+ ],
19
+ [
20
+ 3,
21
+ 11
22
+ ],
23
+ [
24
+ 3,
25
+ 14
26
+ ]
27
+ ],
28
+ "begin_suppress_tokens": [
29
+ 220,
30
+ 50257
31
+ ],
32
+ "bos_token_id": 50257,
33
+ "decoder_start_token_id": 50258,
34
+ "eos_token_id": 50257,
35
+ "forced_decoder_ids": [
36
+ [
37
+ 1,
38
+ null
39
+ ],
40
+ [
41
+ 2,
42
+ 50360
43
+ ]
44
+ ],
45
+ "is_multilingual": true,
46
+ "lang_to_id": {
47
+ "<|af|>": 50327,
48
+ "<|am|>": 50334,
49
+ "<|ar|>": 50272,
50
+ "<|as|>": 50350,
51
+ "<|az|>": 50304,
52
+ "<|ba|>": 50355,
53
+ "<|be|>": 50330,
54
+ "<|bg|>": 50292,
55
+ "<|bn|>": 50302,
56
+ "<|bo|>": 50347,
57
+ "<|br|>": 50309,
58
+ "<|bs|>": 50315,
59
+ "<|ca|>": 50270,
60
+ "<|cs|>": 50283,
61
+ "<|cy|>": 50297,
62
+ "<|da|>": 50285,
63
+ "<|de|>": 50261,
64
+ "<|el|>": 50281,
65
+ "<|en|>": 50259,
66
+ "<|es|>": 50262,
67
+ "<|et|>": 50307,
68
+ "<|eu|>": 50310,
69
+ "<|fa|>": 50300,
70
+ "<|fi|>": 50277,
71
+ "<|fo|>": 50338,
72
+ "<|fr|>": 50265,
73
+ "<|gl|>": 50319,
74
+ "<|gu|>": 50333,
75
+ "<|haw|>": 50352,
76
+ "<|ha|>": 50354,
77
+ "<|he|>": 50279,
78
+ "<|hi|>": 50276,
79
+ "<|hr|>": 50291,
80
+ "<|ht|>": 50339,
81
+ "<|hu|>": 50286,
82
+ "<|hy|>": 50312,
83
+ "<|id|>": 50275,
84
+ "<|is|>": 50311,
85
+ "<|it|>": 50274,
86
+ "<|ja|>": 50266,
87
+ "<|jw|>": 50356,
88
+ "<|ka|>": 50329,
89
+ "<|kk|>": 50316,
90
+ "<|km|>": 50323,
91
+ "<|kn|>": 50306,
92
+ "<|ko|>": 50264,
93
+ "<|la|>": 50294,
94
+ "<|lb|>": 50345,
95
+ "<|ln|>": 50353,
96
+ "<|lo|>": 50336,
97
+ "<|lt|>": 50293,
98
+ "<|lv|>": 50301,
99
+ "<|mg|>": 50349,
100
+ "<|mi|>": 50295,
101
+ "<|mk|>": 50308,
102
+ "<|ml|>": 50296,
103
+ "<|mn|>": 50314,
104
+ "<|mr|>": 50320,
105
+ "<|ms|>": 50282,
106
+ "<|mt|>": 50343,
107
+ "<|my|>": 50346,
108
+ "<|ne|>": 50313,
109
+ "<|nl|>": 50271,
110
+ "<|nn|>": 50342,
111
+ "<|no|>": 50288,
112
+ "<|oc|>": 50328,
113
+ "<|pa|>": 50321,
114
+ "<|pl|>": 50269,
115
+ "<|ps|>": 50340,
116
+ "<|pt|>": 50267,
117
+ "<|ro|>": 50284,
118
+ "<|ru|>": 50263,
119
+ "<|sa|>": 50344,
120
+ "<|sd|>": 50332,
121
+ "<|si|>": 50322,
122
+ "<|sk|>": 50298,
123
+ "<|sl|>": 50305,
124
+ "<|sn|>": 50324,
125
+ "<|so|>": 50326,
126
+ "<|sq|>": 50317,
127
+ "<|sr|>": 50303,
128
+ "<|su|>": 50357,
129
+ "<|sv|>": 50273,
130
+ "<|sw|>": 50318,
131
+ "<|ta|>": 50287,
132
+ "<|te|>": 50299,
133
+ "<|tg|>": 50331,
134
+ "<|th|>": 50289,
135
+ "<|tk|>": 50341,
136
+ "<|tl|>": 50348,
137
+ "<|tr|>": 50268,
138
+ "<|tt|>": 50351,
139
+ "<|uk|>": 50280,
140
+ "<|ur|>": 50290,
141
+ "<|uz|>": 50337,
142
+ "<|vi|>": 50278,
143
+ "<|yi|>": 50335,
144
+ "<|yo|>": 50325,
145
+ "<|yue|>": 50358,
146
+ "<|zh|>": 50260
147
+ },
148
+ "max_initial_timestamp_index": 50,
149
+ "max_length": 448,
150
+ "no_timestamps_token_id": 50364,
151
+ "pad_token_id": 50257,
152
+ "prev_sot_token_id": 50362,
153
+ "return_timestamps": false,
154
+ "suppress_tokens": [
155
+ 1,
156
+ 2,
157
+ 7,
158
+ 8,
159
+ 9,
160
+ 10,
161
+ 14,
162
+ 25,
163
+ 26,
164
+ 27,
165
+ 28,
166
+ 29,
167
+ 31,
168
+ 58,
169
+ 59,
170
+ 60,
171
+ 61,
172
+ 62,
173
+ 63,
174
+ 90,
175
+ 91,
176
+ 92,
177
+ 93,
178
+ 359,
179
+ 503,
180
+ 522,
181
+ 542,
182
+ 873,
183
+ 893,
184
+ 902,
185
+ 918,
186
+ 922,
187
+ 931,
188
+ 1350,
189
+ 1853,
190
+ 1982,
191
+ 2460,
192
+ 2627,
193
+ 3246,
194
+ 3253,
195
+ 3268,
196
+ 3536,
197
+ 3846,
198
+ 3961,
199
+ 4183,
200
+ 4667,
201
+ 6585,
202
+ 6647,
203
+ 7273,
204
+ 9061,
205
+ 9383,
206
+ 10428,
207
+ 10929,
208
+ 11938,
209
+ 12033,
210
+ 12331,
211
+ 12562,
212
+ 13793,
213
+ 14157,
214
+ 14635,
215
+ 15265,
216
+ 15618,
217
+ 16553,
218
+ 16604,
219
+ 18362,
220
+ 18956,
221
+ 20075,
222
+ 21675,
223
+ 22520,
224
+ 26130,
225
+ 26161,
226
+ 26435,
227
+ 28279,
228
+ 29464,
229
+ 31650,
230
+ 32302,
231
+ 32470,
232
+ 36865,
233
+ 42863,
234
+ 47425,
235
+ 49870,
236
+ 50254,
237
+ 50258,
238
+ 50359,
239
+ 50360,
240
+ 50361,
241
+ 50362,
242
+ 50363
243
+ ],
244
+ "task_to_id": {
245
+ "transcribe": 50360,
246
+ "translate": 50359
247
+ },
248
+ "transformers_version": "4.44.2"
249
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76b7de8fb3536b10815da091f55a9f88b77e28e9d91111714316da71e01f2227
3
+ size 3235581408
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99973b36b4b8f95cfaa8582fcd299b862cc8eab46daaaf40a78b796fe58edf32
3
+ size 1642743078
preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 128,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f905e6d259feac89b66bb266cc1afab133022b590943f1a7b0ae0c887320645d
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:66da107d9c6c269be6ca1e279bf2b1451a412bfe47928d09c57c302ba4021cb9
3
+ size 1064
trainer_state.json ADDED
@@ -0,0 +1,3693 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.03668233007192612,
3
+ "best_model_checkpoint": "./susurro-model/checkpoint-3500",
4
+ "epoch": 8.0,
5
+ "eval_steps": 250,
6
+ "global_step": 5000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.016,
13
+ "grad_norm": 22.783082962036133,
14
+ "learning_rate": 4.0000000000000003e-07,
15
+ "loss": 1.4707,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.032,
20
+ "grad_norm": 15.918609619140625,
21
+ "learning_rate": 8.000000000000001e-07,
22
+ "loss": 1.1318,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.048,
27
+ "grad_norm": 6.734115123748779,
28
+ "learning_rate": 1.2000000000000002e-06,
29
+ "loss": 0.566,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.064,
34
+ "grad_norm": 4.893101692199707,
35
+ "learning_rate": 1.6000000000000001e-06,
36
+ "loss": 0.3064,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.08,
41
+ "grad_norm": 4.515320777893066,
42
+ "learning_rate": 2.0000000000000003e-06,
43
+ "loss": 0.1913,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.096,
48
+ "grad_norm": 2.93345046043396,
49
+ "learning_rate": 2.4000000000000003e-06,
50
+ "loss": 0.0643,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.112,
55
+ "grad_norm": 1.9217101335525513,
56
+ "learning_rate": 2.8000000000000003e-06,
57
+ "loss": 0.0534,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.128,
62
+ "grad_norm": 2.2198150157928467,
63
+ "learning_rate": 3.2000000000000003e-06,
64
+ "loss": 0.0547,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.144,
69
+ "grad_norm": 1.7477046251296997,
70
+ "learning_rate": 3.6000000000000003e-06,
71
+ "loss": 0.0376,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.16,
76
+ "grad_norm": 2.2940073013305664,
77
+ "learning_rate": 4.000000000000001e-06,
78
+ "loss": 0.0414,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.176,
83
+ "grad_norm": 2.83594012260437,
84
+ "learning_rate": 4.4e-06,
85
+ "loss": 0.0501,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.192,
90
+ "grad_norm": 1.6579861640930176,
91
+ "learning_rate": 4.800000000000001e-06,
92
+ "loss": 0.0483,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.208,
97
+ "grad_norm": 2.768068313598633,
98
+ "learning_rate": 5.2e-06,
99
+ "loss": 0.0504,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.224,
104
+ "grad_norm": 1.8856732845306396,
105
+ "learning_rate": 5.600000000000001e-06,
106
+ "loss": 0.0452,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.24,
111
+ "grad_norm": 2.343675374984741,
112
+ "learning_rate": 6e-06,
113
+ "loss": 0.0527,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.256,
118
+ "grad_norm": 2.380032539367676,
119
+ "learning_rate": 6.4000000000000006e-06,
120
+ "loss": 0.0488,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.272,
125
+ "grad_norm": 2.847942352294922,
126
+ "learning_rate": 6.800000000000001e-06,
127
+ "loss": 0.0481,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.288,
132
+ "grad_norm": 2.505783796310425,
133
+ "learning_rate": 7.2000000000000005e-06,
134
+ "loss": 0.0525,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.304,
139
+ "grad_norm": 2.4004359245300293,
140
+ "learning_rate": 7.600000000000001e-06,
141
+ "loss": 0.0475,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.32,
146
+ "grad_norm": 2.5646846294403076,
147
+ "learning_rate": 8.000000000000001e-06,
148
+ "loss": 0.0541,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.336,
153
+ "grad_norm": 2.2978780269622803,
154
+ "learning_rate": 8.400000000000001e-06,
155
+ "loss": 0.056,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.352,
160
+ "grad_norm": 1.9104993343353271,
161
+ "learning_rate": 8.8e-06,
162
+ "loss": 0.0507,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.368,
167
+ "grad_norm": 3.259106397628784,
168
+ "learning_rate": 9.200000000000002e-06,
169
+ "loss": 0.0599,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.384,
174
+ "grad_norm": 1.9815133810043335,
175
+ "learning_rate": 9.600000000000001e-06,
176
+ "loss": 0.0531,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.4,
181
+ "grad_norm": 2.202136278152466,
182
+ "learning_rate": 1e-05,
183
+ "loss": 0.0494,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.4,
188
+ "eval_loss": 0.0524754524230957,
189
+ "eval_runtime": 1455.6988,
190
+ "eval_samples_per_second": 3.876,
191
+ "eval_steps_per_second": 3.876,
192
+ "step": 250
193
+ },
194
+ {
195
+ "epoch": 0.416,
196
+ "grad_norm": 2.5911672115325928,
197
+ "learning_rate": 9.978947368421053e-06,
198
+ "loss": 0.0512,
199
+ "step": 260
200
+ },
201
+ {
202
+ "epoch": 0.432,
203
+ "grad_norm": 1.6327568292617798,
204
+ "learning_rate": 9.957894736842106e-06,
205
+ "loss": 0.0485,
206
+ "step": 270
207
+ },
208
+ {
209
+ "epoch": 0.448,
210
+ "grad_norm": 2.290238380432129,
211
+ "learning_rate": 9.936842105263159e-06,
212
+ "loss": 0.0586,
213
+ "step": 280
214
+ },
215
+ {
216
+ "epoch": 0.464,
217
+ "grad_norm": 2.7893710136413574,
218
+ "learning_rate": 9.915789473684211e-06,
219
+ "loss": 0.0597,
220
+ "step": 290
221
+ },
222
+ {
223
+ "epoch": 0.48,
224
+ "grad_norm": 2.4395689964294434,
225
+ "learning_rate": 9.894736842105264e-06,
226
+ "loss": 0.0596,
227
+ "step": 300
228
+ },
229
+ {
230
+ "epoch": 0.496,
231
+ "grad_norm": 2.9003653526306152,
232
+ "learning_rate": 9.873684210526317e-06,
233
+ "loss": 0.0551,
234
+ "step": 310
235
+ },
236
+ {
237
+ "epoch": 0.512,
238
+ "grad_norm": 2.7941534519195557,
239
+ "learning_rate": 9.85263157894737e-06,
240
+ "loss": 0.0579,
241
+ "step": 320
242
+ },
243
+ {
244
+ "epoch": 0.528,
245
+ "grad_norm": 2.705819845199585,
246
+ "learning_rate": 9.831578947368422e-06,
247
+ "loss": 0.059,
248
+ "step": 330
249
+ },
250
+ {
251
+ "epoch": 0.544,
252
+ "grad_norm": 3.2800068855285645,
253
+ "learning_rate": 9.810526315789475e-06,
254
+ "loss": 0.0649,
255
+ "step": 340
256
+ },
257
+ {
258
+ "epoch": 0.56,
259
+ "grad_norm": 1.8395154476165771,
260
+ "learning_rate": 9.789473684210527e-06,
261
+ "loss": 0.0487,
262
+ "step": 350
263
+ },
264
+ {
265
+ "epoch": 0.576,
266
+ "grad_norm": 2.447082757949829,
267
+ "learning_rate": 9.76842105263158e-06,
268
+ "loss": 0.0552,
269
+ "step": 360
270
+ },
271
+ {
272
+ "epoch": 0.592,
273
+ "grad_norm": 2.2035391330718994,
274
+ "learning_rate": 9.747368421052633e-06,
275
+ "loss": 0.0505,
276
+ "step": 370
277
+ },
278
+ {
279
+ "epoch": 0.608,
280
+ "grad_norm": 3.010380506515503,
281
+ "learning_rate": 9.726315789473685e-06,
282
+ "loss": 0.0717,
283
+ "step": 380
284
+ },
285
+ {
286
+ "epoch": 0.624,
287
+ "grad_norm": 2.2279305458068848,
288
+ "learning_rate": 9.705263157894738e-06,
289
+ "loss": 0.0574,
290
+ "step": 390
291
+ },
292
+ {
293
+ "epoch": 0.64,
294
+ "grad_norm": 2.1078450679779053,
295
+ "learning_rate": 9.68421052631579e-06,
296
+ "loss": 0.0549,
297
+ "step": 400
298
+ },
299
+ {
300
+ "epoch": 0.656,
301
+ "grad_norm": 1.9342560768127441,
302
+ "learning_rate": 9.663157894736843e-06,
303
+ "loss": 0.055,
304
+ "step": 410
305
+ },
306
+ {
307
+ "epoch": 0.672,
308
+ "grad_norm": 2.2525532245635986,
309
+ "learning_rate": 9.642105263157896e-06,
310
+ "loss": 0.0562,
311
+ "step": 420
312
+ },
313
+ {
314
+ "epoch": 0.688,
315
+ "grad_norm": 1.9181652069091797,
316
+ "learning_rate": 9.621052631578947e-06,
317
+ "loss": 0.0561,
318
+ "step": 430
319
+ },
320
+ {
321
+ "epoch": 0.704,
322
+ "grad_norm": 2.088951826095581,
323
+ "learning_rate": 9.600000000000001e-06,
324
+ "loss": 0.0575,
325
+ "step": 440
326
+ },
327
+ {
328
+ "epoch": 0.72,
329
+ "grad_norm": 2.951388120651245,
330
+ "learning_rate": 9.578947368421054e-06,
331
+ "loss": 0.0503,
332
+ "step": 450
333
+ },
334
+ {
335
+ "epoch": 0.736,
336
+ "grad_norm": 2.5841259956359863,
337
+ "learning_rate": 9.557894736842107e-06,
338
+ "loss": 0.0613,
339
+ "step": 460
340
+ },
341
+ {
342
+ "epoch": 0.752,
343
+ "grad_norm": 3.1597883701324463,
344
+ "learning_rate": 9.53684210526316e-06,
345
+ "loss": 0.0594,
346
+ "step": 470
347
+ },
348
+ {
349
+ "epoch": 0.768,
350
+ "grad_norm": 1.4449310302734375,
351
+ "learning_rate": 9.515789473684212e-06,
352
+ "loss": 0.0565,
353
+ "step": 480
354
+ },
355
+ {
356
+ "epoch": 0.784,
357
+ "grad_norm": 1.588442325592041,
358
+ "learning_rate": 9.494736842105265e-06,
359
+ "loss": 0.0505,
360
+ "step": 490
361
+ },
362
+ {
363
+ "epoch": 0.8,
364
+ "grad_norm": 2.315497398376465,
365
+ "learning_rate": 9.473684210526315e-06,
366
+ "loss": 0.0571,
367
+ "step": 500
368
+ },
369
+ {
370
+ "epoch": 0.8,
371
+ "eval_loss": 0.04724394157528877,
372
+ "eval_runtime": 1481.9655,
373
+ "eval_samples_per_second": 3.807,
374
+ "eval_steps_per_second": 3.807,
375
+ "step": 500
376
+ },
377
+ {
378
+ "epoch": 0.816,
379
+ "grad_norm": 3.124136209487915,
380
+ "learning_rate": 9.454736842105265e-06,
381
+ "loss": 0.0439,
382
+ "step": 510
383
+ },
384
+ {
385
+ "epoch": 0.832,
386
+ "grad_norm": 2.687971353530884,
387
+ "learning_rate": 9.433684210526318e-06,
388
+ "loss": 0.0535,
389
+ "step": 520
390
+ },
391
+ {
392
+ "epoch": 0.848,
393
+ "grad_norm": 2.442317485809326,
394
+ "learning_rate": 9.412631578947369e-06,
395
+ "loss": 0.0516,
396
+ "step": 530
397
+ },
398
+ {
399
+ "epoch": 0.864,
400
+ "grad_norm": 2.6689493656158447,
401
+ "learning_rate": 9.391578947368421e-06,
402
+ "loss": 0.0565,
403
+ "step": 540
404
+ },
405
+ {
406
+ "epoch": 0.88,
407
+ "grad_norm": 1.959835171699524,
408
+ "learning_rate": 9.370526315789474e-06,
409
+ "loss": 0.0575,
410
+ "step": 550
411
+ },
412
+ {
413
+ "epoch": 0.896,
414
+ "grad_norm": 3.6005074977874756,
415
+ "learning_rate": 9.349473684210526e-06,
416
+ "loss": 0.0444,
417
+ "step": 560
418
+ },
419
+ {
420
+ "epoch": 0.912,
421
+ "grad_norm": 2.1912620067596436,
422
+ "learning_rate": 9.328421052631579e-06,
423
+ "loss": 0.052,
424
+ "step": 570
425
+ },
426
+ {
427
+ "epoch": 0.928,
428
+ "grad_norm": 2.148808240890503,
429
+ "learning_rate": 9.307368421052634e-06,
430
+ "loss": 0.0555,
431
+ "step": 580
432
+ },
433
+ {
434
+ "epoch": 0.944,
435
+ "grad_norm": 2.872231960296631,
436
+ "learning_rate": 9.286315789473686e-06,
437
+ "loss": 0.054,
438
+ "step": 590
439
+ },
440
+ {
441
+ "epoch": 0.96,
442
+ "grad_norm": 3.3271450996398926,
443
+ "learning_rate": 9.265263157894737e-06,
444
+ "loss": 0.0554,
445
+ "step": 600
446
+ },
447
+ {
448
+ "epoch": 0.976,
449
+ "grad_norm": 2.058793067932129,
450
+ "learning_rate": 9.24421052631579e-06,
451
+ "loss": 0.0418,
452
+ "step": 610
453
+ },
454
+ {
455
+ "epoch": 0.992,
456
+ "grad_norm": 2.0928821563720703,
457
+ "learning_rate": 9.225263157894737e-06,
458
+ "loss": 0.0525,
459
+ "step": 620
460
+ },
461
+ {
462
+ "epoch": 1.008,
463
+ "grad_norm": 1.410493016242981,
464
+ "learning_rate": 9.20421052631579e-06,
465
+ "loss": 0.0448,
466
+ "step": 630
467
+ },
468
+ {
469
+ "epoch": 1.024,
470
+ "grad_norm": 1.4376940727233887,
471
+ "learning_rate": 9.183157894736843e-06,
472
+ "loss": 0.0281,
473
+ "step": 640
474
+ },
475
+ {
476
+ "epoch": 1.04,
477
+ "grad_norm": 1.678579568862915,
478
+ "learning_rate": 9.162105263157895e-06,
479
+ "loss": 0.0207,
480
+ "step": 650
481
+ },
482
+ {
483
+ "epoch": 1.056,
484
+ "grad_norm": 2.18996524810791,
485
+ "learning_rate": 9.141052631578948e-06,
486
+ "loss": 0.0288,
487
+ "step": 660
488
+ },
489
+ {
490
+ "epoch": 1.072,
491
+ "grad_norm": 1.0160465240478516,
492
+ "learning_rate": 9.12e-06,
493
+ "loss": 0.0199,
494
+ "step": 670
495
+ },
496
+ {
497
+ "epoch": 1.088,
498
+ "grad_norm": 1.4741309881210327,
499
+ "learning_rate": 9.098947368421053e-06,
500
+ "loss": 0.0248,
501
+ "step": 680
502
+ },
503
+ {
504
+ "epoch": 1.104,
505
+ "grad_norm": 1.5385643243789673,
506
+ "learning_rate": 9.077894736842106e-06,
507
+ "loss": 0.0286,
508
+ "step": 690
509
+ },
510
+ {
511
+ "epoch": 1.12,
512
+ "grad_norm": 1.65945565700531,
513
+ "learning_rate": 9.056842105263159e-06,
514
+ "loss": 0.0331,
515
+ "step": 700
516
+ },
517
+ {
518
+ "epoch": 1.1360000000000001,
519
+ "grad_norm": 2.443169593811035,
520
+ "learning_rate": 9.035789473684211e-06,
521
+ "loss": 0.0328,
522
+ "step": 710
523
+ },
524
+ {
525
+ "epoch": 1.152,
526
+ "grad_norm": 1.950245976448059,
527
+ "learning_rate": 9.014736842105264e-06,
528
+ "loss": 0.0297,
529
+ "step": 720
530
+ },
531
+ {
532
+ "epoch": 1.168,
533
+ "grad_norm": 1.0469032526016235,
534
+ "learning_rate": 8.993684210526317e-06,
535
+ "loss": 0.0282,
536
+ "step": 730
537
+ },
538
+ {
539
+ "epoch": 1.184,
540
+ "grad_norm": 1.5326778888702393,
541
+ "learning_rate": 8.97263157894737e-06,
542
+ "loss": 0.0307,
543
+ "step": 740
544
+ },
545
+ {
546
+ "epoch": 1.2,
547
+ "grad_norm": 0.976245105266571,
548
+ "learning_rate": 8.951578947368422e-06,
549
+ "loss": 0.0293,
550
+ "step": 750
551
+ },
552
+ {
553
+ "epoch": 1.2,
554
+ "eval_loss": 0.04405752569437027,
555
+ "eval_runtime": 1500.9354,
556
+ "eval_samples_per_second": 3.759,
557
+ "eval_steps_per_second": 3.759,
558
+ "step": 750
559
+ },
560
+ {
561
+ "epoch": 1.216,
562
+ "grad_norm": 1.4788917303085327,
563
+ "learning_rate": 8.930526315789475e-06,
564
+ "loss": 0.0341,
565
+ "step": 760
566
+ },
567
+ {
568
+ "epoch": 1.232,
569
+ "grad_norm": 1.398728847503662,
570
+ "learning_rate": 8.909473684210527e-06,
571
+ "loss": 0.0301,
572
+ "step": 770
573
+ },
574
+ {
575
+ "epoch": 1.248,
576
+ "grad_norm": 1.7809382677078247,
577
+ "learning_rate": 8.88842105263158e-06,
578
+ "loss": 0.0265,
579
+ "step": 780
580
+ },
581
+ {
582
+ "epoch": 1.264,
583
+ "grad_norm": 1.9043110609054565,
584
+ "learning_rate": 8.867368421052633e-06,
585
+ "loss": 0.0277,
586
+ "step": 790
587
+ },
588
+ {
589
+ "epoch": 1.28,
590
+ "grad_norm": 2.286780595779419,
591
+ "learning_rate": 8.846315789473685e-06,
592
+ "loss": 0.0241,
593
+ "step": 800
594
+ },
595
+ {
596
+ "epoch": 1.296,
597
+ "grad_norm": 2.867587089538574,
598
+ "learning_rate": 8.825263157894738e-06,
599
+ "loss": 0.0288,
600
+ "step": 810
601
+ },
602
+ {
603
+ "epoch": 1.312,
604
+ "grad_norm": 1.6300053596496582,
605
+ "learning_rate": 8.80421052631579e-06,
606
+ "loss": 0.0229,
607
+ "step": 820
608
+ },
609
+ {
610
+ "epoch": 1.328,
611
+ "grad_norm": 2.1260311603546143,
612
+ "learning_rate": 8.783157894736842e-06,
613
+ "loss": 0.034,
614
+ "step": 830
615
+ },
616
+ {
617
+ "epoch": 1.3439999999999999,
618
+ "grad_norm": 1.2183325290679932,
619
+ "learning_rate": 8.762105263157896e-06,
620
+ "loss": 0.0261,
621
+ "step": 840
622
+ },
623
+ {
624
+ "epoch": 1.3599999999999999,
625
+ "grad_norm": 1.4144467115402222,
626
+ "learning_rate": 8.741052631578949e-06,
627
+ "loss": 0.0258,
628
+ "step": 850
629
+ },
630
+ {
631
+ "epoch": 1.376,
632
+ "grad_norm": 1.658300757408142,
633
+ "learning_rate": 8.720000000000001e-06,
634
+ "loss": 0.0286,
635
+ "step": 860
636
+ },
637
+ {
638
+ "epoch": 1.392,
639
+ "grad_norm": 1.566178321838379,
640
+ "learning_rate": 8.698947368421054e-06,
641
+ "loss": 0.0259,
642
+ "step": 870
643
+ },
644
+ {
645
+ "epoch": 1.408,
646
+ "grad_norm": 1.7910547256469727,
647
+ "learning_rate": 8.677894736842107e-06,
648
+ "loss": 0.024,
649
+ "step": 880
650
+ },
651
+ {
652
+ "epoch": 1.424,
653
+ "grad_norm": 1.882262110710144,
654
+ "learning_rate": 8.65684210526316e-06,
655
+ "loss": 0.0312,
656
+ "step": 890
657
+ },
658
+ {
659
+ "epoch": 1.44,
660
+ "grad_norm": 1.2737823724746704,
661
+ "learning_rate": 8.63578947368421e-06,
662
+ "loss": 0.0324,
663
+ "step": 900
664
+ },
665
+ {
666
+ "epoch": 1.456,
667
+ "grad_norm": 1.954757809638977,
668
+ "learning_rate": 8.614736842105263e-06,
669
+ "loss": 0.0276,
670
+ "step": 910
671
+ },
672
+ {
673
+ "epoch": 1.472,
674
+ "grad_norm": 2.4381039142608643,
675
+ "learning_rate": 8.593684210526315e-06,
676
+ "loss": 0.0258,
677
+ "step": 920
678
+ },
679
+ {
680
+ "epoch": 1.488,
681
+ "grad_norm": 1.962517499923706,
682
+ "learning_rate": 8.57263157894737e-06,
683
+ "loss": 0.0357,
684
+ "step": 930
685
+ },
686
+ {
687
+ "epoch": 1.504,
688
+ "grad_norm": 1.7667138576507568,
689
+ "learning_rate": 8.551578947368422e-06,
690
+ "loss": 0.0358,
691
+ "step": 940
692
+ },
693
+ {
694
+ "epoch": 1.52,
695
+ "grad_norm": 1.6980652809143066,
696
+ "learning_rate": 8.530526315789475e-06,
697
+ "loss": 0.0247,
698
+ "step": 950
699
+ },
700
+ {
701
+ "epoch": 1.536,
702
+ "grad_norm": 1.765106201171875,
703
+ "learning_rate": 8.509473684210528e-06,
704
+ "loss": 0.0289,
705
+ "step": 960
706
+ },
707
+ {
708
+ "epoch": 1.552,
709
+ "grad_norm": 2.0965826511383057,
710
+ "learning_rate": 8.488421052631579e-06,
711
+ "loss": 0.0293,
712
+ "step": 970
713
+ },
714
+ {
715
+ "epoch": 1.568,
716
+ "grad_norm": 1.2776910066604614,
717
+ "learning_rate": 8.467368421052631e-06,
718
+ "loss": 0.0248,
719
+ "step": 980
720
+ },
721
+ {
722
+ "epoch": 1.584,
723
+ "grad_norm": 1.9160679578781128,
724
+ "learning_rate": 8.446315789473684e-06,
725
+ "loss": 0.0327,
726
+ "step": 990
727
+ },
728
+ {
729
+ "epoch": 1.6,
730
+ "grad_norm": 2.084198474884033,
731
+ "learning_rate": 8.425263157894737e-06,
732
+ "loss": 0.0248,
733
+ "step": 1000
734
+ },
735
+ {
736
+ "epoch": 1.6,
737
+ "eval_loss": 0.04179142042994499,
738
+ "eval_runtime": 1520.1889,
739
+ "eval_samples_per_second": 3.711,
740
+ "eval_steps_per_second": 3.711,
741
+ "step": 1000
742
+ },
743
+ {
744
+ "epoch": 1.616,
745
+ "grad_norm": 1.8233721256256104,
746
+ "learning_rate": 8.404210526315791e-06,
747
+ "loss": 0.0325,
748
+ "step": 1010
749
+ },
750
+ {
751
+ "epoch": 1.6320000000000001,
752
+ "grad_norm": 2.6856882572174072,
753
+ "learning_rate": 8.383157894736844e-06,
754
+ "loss": 0.0345,
755
+ "step": 1020
756
+ },
757
+ {
758
+ "epoch": 1.6480000000000001,
759
+ "grad_norm": 1.6194241046905518,
760
+ "learning_rate": 8.362105263157896e-06,
761
+ "loss": 0.0248,
762
+ "step": 1030
763
+ },
764
+ {
765
+ "epoch": 1.6640000000000001,
766
+ "grad_norm": 2.0623207092285156,
767
+ "learning_rate": 8.341052631578947e-06,
768
+ "loss": 0.0297,
769
+ "step": 1040
770
+ },
771
+ {
772
+ "epoch": 1.6800000000000002,
773
+ "grad_norm": 2.122952461242676,
774
+ "learning_rate": 8.32e-06,
775
+ "loss": 0.0346,
776
+ "step": 1050
777
+ },
778
+ {
779
+ "epoch": 1.696,
780
+ "grad_norm": 1.4452091455459595,
781
+ "learning_rate": 8.298947368421053e-06,
782
+ "loss": 0.03,
783
+ "step": 1060
784
+ },
785
+ {
786
+ "epoch": 1.712,
787
+ "grad_norm": 2.9357564449310303,
788
+ "learning_rate": 8.277894736842105e-06,
789
+ "loss": 0.026,
790
+ "step": 1070
791
+ },
792
+ {
793
+ "epoch": 1.728,
794
+ "grad_norm": 1.0874323844909668,
795
+ "learning_rate": 8.256842105263158e-06,
796
+ "loss": 0.0294,
797
+ "step": 1080
798
+ },
799
+ {
800
+ "epoch": 1.744,
801
+ "grad_norm": 1.1625168323516846,
802
+ "learning_rate": 8.235789473684212e-06,
803
+ "loss": 0.0291,
804
+ "step": 1090
805
+ },
806
+ {
807
+ "epoch": 1.76,
808
+ "grad_norm": 1.3480169773101807,
809
+ "learning_rate": 8.214736842105265e-06,
810
+ "loss": 0.034,
811
+ "step": 1100
812
+ },
813
+ {
814
+ "epoch": 1.776,
815
+ "grad_norm": 1.1702100038528442,
816
+ "learning_rate": 8.193684210526316e-06,
817
+ "loss": 0.029,
818
+ "step": 1110
819
+ },
820
+ {
821
+ "epoch": 1.792,
822
+ "grad_norm": 2.0428359508514404,
823
+ "learning_rate": 8.172631578947369e-06,
824
+ "loss": 0.023,
825
+ "step": 1120
826
+ },
827
+ {
828
+ "epoch": 1.808,
829
+ "grad_norm": 0.7537345886230469,
830
+ "learning_rate": 8.151578947368421e-06,
831
+ "loss": 0.03,
832
+ "step": 1130
833
+ },
834
+ {
835
+ "epoch": 1.8239999999999998,
836
+ "grad_norm": 2.184401273727417,
837
+ "learning_rate": 8.130526315789474e-06,
838
+ "loss": 0.0327,
839
+ "step": 1140
840
+ },
841
+ {
842
+ "epoch": 1.8399999999999999,
843
+ "grad_norm": 2.1545660495758057,
844
+ "learning_rate": 8.109473684210527e-06,
845
+ "loss": 0.0275,
846
+ "step": 1150
847
+ },
848
+ {
849
+ "epoch": 1.8559999999999999,
850
+ "grad_norm": 2.535369396209717,
851
+ "learning_rate": 8.08842105263158e-06,
852
+ "loss": 0.0292,
853
+ "step": 1160
854
+ },
855
+ {
856
+ "epoch": 1.8719999999999999,
857
+ "grad_norm": 1.8030229806900024,
858
+ "learning_rate": 8.067368421052632e-06,
859
+ "loss": 0.0264,
860
+ "step": 1170
861
+ },
862
+ {
863
+ "epoch": 1.888,
864
+ "grad_norm": 1.8990155458450317,
865
+ "learning_rate": 8.046315789473686e-06,
866
+ "loss": 0.0365,
867
+ "step": 1180
868
+ },
869
+ {
870
+ "epoch": 1.904,
871
+ "grad_norm": 1.4508986473083496,
872
+ "learning_rate": 8.025263157894737e-06,
873
+ "loss": 0.0251,
874
+ "step": 1190
875
+ },
876
+ {
877
+ "epoch": 1.92,
878
+ "grad_norm": 1.0793194770812988,
879
+ "learning_rate": 8.00421052631579e-06,
880
+ "loss": 0.0274,
881
+ "step": 1200
882
+ },
883
+ {
884
+ "epoch": 1.936,
885
+ "grad_norm": 1.9805344343185425,
886
+ "learning_rate": 7.983157894736842e-06,
887
+ "loss": 0.0299,
888
+ "step": 1210
889
+ },
890
+ {
891
+ "epoch": 1.952,
892
+ "grad_norm": 2.258500576019287,
893
+ "learning_rate": 7.962105263157895e-06,
894
+ "loss": 0.0266,
895
+ "step": 1220
896
+ },
897
+ {
898
+ "epoch": 1.968,
899
+ "grad_norm": 1.4639638662338257,
900
+ "learning_rate": 7.941052631578948e-06,
901
+ "loss": 0.026,
902
+ "step": 1230
903
+ },
904
+ {
905
+ "epoch": 1.984,
906
+ "grad_norm": 1.3057489395141602,
907
+ "learning_rate": 7.92e-06,
908
+ "loss": 0.0305,
909
+ "step": 1240
910
+ },
911
+ {
912
+ "epoch": 2.0,
913
+ "grad_norm": 1.29533052444458,
914
+ "learning_rate": 7.898947368421053e-06,
915
+ "loss": 0.0265,
916
+ "step": 1250
917
+ },
918
+ {
919
+ "epoch": 2.0,
920
+ "eval_loss": 0.04163273423910141,
921
+ "eval_runtime": 1542.8649,
922
+ "eval_samples_per_second": 3.657,
923
+ "eval_steps_per_second": 3.657,
924
+ "step": 1250
925
+ },
926
+ {
927
+ "epoch": 2.016,
928
+ "grad_norm": 0.5275375843048096,
929
+ "learning_rate": 7.877894736842106e-06,
930
+ "loss": 0.0117,
931
+ "step": 1260
932
+ },
933
+ {
934
+ "epoch": 2.032,
935
+ "grad_norm": 2.0590779781341553,
936
+ "learning_rate": 7.856842105263158e-06,
937
+ "loss": 0.0156,
938
+ "step": 1270
939
+ },
940
+ {
941
+ "epoch": 2.048,
942
+ "grad_norm": 1.2595654726028442,
943
+ "learning_rate": 7.835789473684211e-06,
944
+ "loss": 0.0105,
945
+ "step": 1280
946
+ },
947
+ {
948
+ "epoch": 2.064,
949
+ "grad_norm": 1.605797529220581,
950
+ "learning_rate": 7.814736842105264e-06,
951
+ "loss": 0.0176,
952
+ "step": 1290
953
+ },
954
+ {
955
+ "epoch": 2.08,
956
+ "grad_norm": 1.2974120378494263,
957
+ "learning_rate": 7.793684210526316e-06,
958
+ "loss": 0.0196,
959
+ "step": 1300
960
+ },
961
+ {
962
+ "epoch": 2.096,
963
+ "grad_norm": 3.662148952484131,
964
+ "learning_rate": 7.772631578947369e-06,
965
+ "loss": 0.0144,
966
+ "step": 1310
967
+ },
968
+ {
969
+ "epoch": 2.112,
970
+ "grad_norm": 0.9286357164382935,
971
+ "learning_rate": 7.751578947368422e-06,
972
+ "loss": 0.0126,
973
+ "step": 1320
974
+ },
975
+ {
976
+ "epoch": 2.128,
977
+ "grad_norm": 1.4938029050827026,
978
+ "learning_rate": 7.730526315789474e-06,
979
+ "loss": 0.0181,
980
+ "step": 1330
981
+ },
982
+ {
983
+ "epoch": 2.144,
984
+ "grad_norm": 1.9004080295562744,
985
+ "learning_rate": 7.709473684210527e-06,
986
+ "loss": 0.0192,
987
+ "step": 1340
988
+ },
989
+ {
990
+ "epoch": 2.16,
991
+ "grad_norm": 0.7749639749526978,
992
+ "learning_rate": 7.68842105263158e-06,
993
+ "loss": 0.0146,
994
+ "step": 1350
995
+ },
996
+ {
997
+ "epoch": 2.176,
998
+ "grad_norm": 1.69952392578125,
999
+ "learning_rate": 7.667368421052632e-06,
1000
+ "loss": 0.0169,
1001
+ "step": 1360
1002
+ },
1003
+ {
1004
+ "epoch": 2.192,
1005
+ "grad_norm": 0.6169717311859131,
1006
+ "learning_rate": 7.646315789473685e-06,
1007
+ "loss": 0.0102,
1008
+ "step": 1370
1009
+ },
1010
+ {
1011
+ "epoch": 2.208,
1012
+ "grad_norm": 3.5800750255584717,
1013
+ "learning_rate": 7.6252631578947376e-06,
1014
+ "loss": 0.014,
1015
+ "step": 1380
1016
+ },
1017
+ {
1018
+ "epoch": 2.224,
1019
+ "grad_norm": 0.6337082982063293,
1020
+ "learning_rate": 7.60421052631579e-06,
1021
+ "loss": 0.0161,
1022
+ "step": 1390
1023
+ },
1024
+ {
1025
+ "epoch": 2.24,
1026
+ "grad_norm": 0.7504271268844604,
1027
+ "learning_rate": 7.583157894736842e-06,
1028
+ "loss": 0.0145,
1029
+ "step": 1400
1030
+ },
1031
+ {
1032
+ "epoch": 2.2560000000000002,
1033
+ "grad_norm": 0.6736518740653992,
1034
+ "learning_rate": 7.562105263157895e-06,
1035
+ "loss": 0.0105,
1036
+ "step": 1410
1037
+ },
1038
+ {
1039
+ "epoch": 2.2720000000000002,
1040
+ "grad_norm": 1.1138242483139038,
1041
+ "learning_rate": 7.541052631578948e-06,
1042
+ "loss": 0.0136,
1043
+ "step": 1420
1044
+ },
1045
+ {
1046
+ "epoch": 2.288,
1047
+ "grad_norm": 1.6553155183792114,
1048
+ "learning_rate": 7.520000000000001e-06,
1049
+ "loss": 0.0135,
1050
+ "step": 1430
1051
+ },
1052
+ {
1053
+ "epoch": 2.304,
1054
+ "grad_norm": 1.3883057832717896,
1055
+ "learning_rate": 7.4989473684210535e-06,
1056
+ "loss": 0.0136,
1057
+ "step": 1440
1058
+ },
1059
+ {
1060
+ "epoch": 2.32,
1061
+ "grad_norm": 1.3742883205413818,
1062
+ "learning_rate": 7.477894736842106e-06,
1063
+ "loss": 0.0146,
1064
+ "step": 1450
1065
+ },
1066
+ {
1067
+ "epoch": 2.336,
1068
+ "grad_norm": 2.129136562347412,
1069
+ "learning_rate": 7.456842105263159e-06,
1070
+ "loss": 0.0142,
1071
+ "step": 1460
1072
+ },
1073
+ {
1074
+ "epoch": 2.352,
1075
+ "grad_norm": 1.771249532699585,
1076
+ "learning_rate": 7.435789473684211e-06,
1077
+ "loss": 0.0182,
1078
+ "step": 1470
1079
+ },
1080
+ {
1081
+ "epoch": 2.368,
1082
+ "grad_norm": 0.906812846660614,
1083
+ "learning_rate": 7.414736842105263e-06,
1084
+ "loss": 0.0127,
1085
+ "step": 1480
1086
+ },
1087
+ {
1088
+ "epoch": 2.384,
1089
+ "grad_norm": 1.0035991668701172,
1090
+ "learning_rate": 7.393684210526316e-06,
1091
+ "loss": 0.0148,
1092
+ "step": 1490
1093
+ },
1094
+ {
1095
+ "epoch": 2.4,
1096
+ "grad_norm": 1.0261139869689941,
1097
+ "learning_rate": 7.3726315789473694e-06,
1098
+ "loss": 0.0162,
1099
+ "step": 1500
1100
+ },
1101
+ {
1102
+ "epoch": 2.4,
1103
+ "eval_loss": 0.04060380160808563,
1104
+ "eval_runtime": 1542.3166,
1105
+ "eval_samples_per_second": 3.658,
1106
+ "eval_steps_per_second": 3.658,
1107
+ "step": 1500
1108
+ },
1109
+ {
1110
+ "epoch": 2.416,
1111
+ "grad_norm": 0.9582253098487854,
1112
+ "learning_rate": 7.351578947368422e-06,
1113
+ "loss": 0.0139,
1114
+ "step": 1510
1115
+ },
1116
+ {
1117
+ "epoch": 2.432,
1118
+ "grad_norm": 1.6380131244659424,
1119
+ "learning_rate": 7.330526315789475e-06,
1120
+ "loss": 0.0175,
1121
+ "step": 1520
1122
+ },
1123
+ {
1124
+ "epoch": 2.448,
1125
+ "grad_norm": 2.057828426361084,
1126
+ "learning_rate": 7.309473684210527e-06,
1127
+ "loss": 0.0168,
1128
+ "step": 1530
1129
+ },
1130
+ {
1131
+ "epoch": 2.464,
1132
+ "grad_norm": 1.6444942951202393,
1133
+ "learning_rate": 7.288421052631579e-06,
1134
+ "loss": 0.0151,
1135
+ "step": 1540
1136
+ },
1137
+ {
1138
+ "epoch": 2.48,
1139
+ "grad_norm": 0.5916880369186401,
1140
+ "learning_rate": 7.267368421052632e-06,
1141
+ "loss": 0.0184,
1142
+ "step": 1550
1143
+ },
1144
+ {
1145
+ "epoch": 2.496,
1146
+ "grad_norm": 1.4819905757904053,
1147
+ "learning_rate": 7.2463157894736845e-06,
1148
+ "loss": 0.0215,
1149
+ "step": 1560
1150
+ },
1151
+ {
1152
+ "epoch": 2.512,
1153
+ "grad_norm": 0.944874107837677,
1154
+ "learning_rate": 7.225263157894737e-06,
1155
+ "loss": 0.0146,
1156
+ "step": 1570
1157
+ },
1158
+ {
1159
+ "epoch": 2.528,
1160
+ "grad_norm": 1.1218395233154297,
1161
+ "learning_rate": 7.20421052631579e-06,
1162
+ "loss": 0.0155,
1163
+ "step": 1580
1164
+ },
1165
+ {
1166
+ "epoch": 2.544,
1167
+ "grad_norm": 1.7295416593551636,
1168
+ "learning_rate": 7.183157894736843e-06,
1169
+ "loss": 0.0149,
1170
+ "step": 1590
1171
+ },
1172
+ {
1173
+ "epoch": 2.56,
1174
+ "grad_norm": 2.064664602279663,
1175
+ "learning_rate": 7.162105263157896e-06,
1176
+ "loss": 0.0209,
1177
+ "step": 1600
1178
+ },
1179
+ {
1180
+ "epoch": 2.576,
1181
+ "grad_norm": 1.4919706583023071,
1182
+ "learning_rate": 7.141052631578948e-06,
1183
+ "loss": 0.0174,
1184
+ "step": 1610
1185
+ },
1186
+ {
1187
+ "epoch": 2.592,
1188
+ "grad_norm": 1.8185276985168457,
1189
+ "learning_rate": 7.1200000000000004e-06,
1190
+ "loss": 0.0153,
1191
+ "step": 1620
1192
+ },
1193
+ {
1194
+ "epoch": 2.608,
1195
+ "grad_norm": 1.0406450033187866,
1196
+ "learning_rate": 7.098947368421053e-06,
1197
+ "loss": 0.017,
1198
+ "step": 1630
1199
+ },
1200
+ {
1201
+ "epoch": 2.624,
1202
+ "grad_norm": 2.1542301177978516,
1203
+ "learning_rate": 7.077894736842106e-06,
1204
+ "loss": 0.0171,
1205
+ "step": 1640
1206
+ },
1207
+ {
1208
+ "epoch": 2.64,
1209
+ "grad_norm": 1.3869684934616089,
1210
+ "learning_rate": 7.056842105263158e-06,
1211
+ "loss": 0.0141,
1212
+ "step": 1650
1213
+ },
1214
+ {
1215
+ "epoch": 2.656,
1216
+ "grad_norm": 1.0020642280578613,
1217
+ "learning_rate": 7.035789473684211e-06,
1218
+ "loss": 0.0176,
1219
+ "step": 1660
1220
+ },
1221
+ {
1222
+ "epoch": 2.672,
1223
+ "grad_norm": 2.786790370941162,
1224
+ "learning_rate": 7.0147368421052646e-06,
1225
+ "loss": 0.0132,
1226
+ "step": 1670
1227
+ },
1228
+ {
1229
+ "epoch": 2.6879999999999997,
1230
+ "grad_norm": 1.0302237272262573,
1231
+ "learning_rate": 6.9936842105263155e-06,
1232
+ "loss": 0.0139,
1233
+ "step": 1680
1234
+ },
1235
+ {
1236
+ "epoch": 2.7039999999999997,
1237
+ "grad_norm": 1.0650746822357178,
1238
+ "learning_rate": 6.972631578947369e-06,
1239
+ "loss": 0.0137,
1240
+ "step": 1690
1241
+ },
1242
+ {
1243
+ "epoch": 2.7199999999999998,
1244
+ "grad_norm": 1.5508835315704346,
1245
+ "learning_rate": 6.951578947368422e-06,
1246
+ "loss": 0.0151,
1247
+ "step": 1700
1248
+ },
1249
+ {
1250
+ "epoch": 2.7359999999999998,
1251
+ "grad_norm": 0.7046643495559692,
1252
+ "learning_rate": 6.930526315789474e-06,
1253
+ "loss": 0.0127,
1254
+ "step": 1710
1255
+ },
1256
+ {
1257
+ "epoch": 2.752,
1258
+ "grad_norm": 1.3175015449523926,
1259
+ "learning_rate": 6.909473684210527e-06,
1260
+ "loss": 0.015,
1261
+ "step": 1720
1262
+ },
1263
+ {
1264
+ "epoch": 2.768,
1265
+ "grad_norm": 2.0063388347625732,
1266
+ "learning_rate": 6.88842105263158e-06,
1267
+ "loss": 0.0143,
1268
+ "step": 1730
1269
+ },
1270
+ {
1271
+ "epoch": 2.784,
1272
+ "grad_norm": 1.5595343112945557,
1273
+ "learning_rate": 6.867368421052632e-06,
1274
+ "loss": 0.0175,
1275
+ "step": 1740
1276
+ },
1277
+ {
1278
+ "epoch": 2.8,
1279
+ "grad_norm": 1.6928199529647827,
1280
+ "learning_rate": 6.846315789473684e-06,
1281
+ "loss": 0.0147,
1282
+ "step": 1750
1283
+ },
1284
+ {
1285
+ "epoch": 2.8,
1286
+ "eval_loss": 0.03860897570848465,
1287
+ "eval_runtime": 9405.2396,
1288
+ "eval_samples_per_second": 0.6,
1289
+ "eval_steps_per_second": 0.6,
1290
+ "step": 1750
1291
+ },
1292
+ {
1293
+ "epoch": 2.816,
1294
+ "grad_norm": 2.1976370811462402,
1295
+ "learning_rate": 6.825263157894737e-06,
1296
+ "loss": 0.0158,
1297
+ "step": 1760
1298
+ },
1299
+ {
1300
+ "epoch": 2.832,
1301
+ "grad_norm": 1.3669664859771729,
1302
+ "learning_rate": 6.804210526315789e-06,
1303
+ "loss": 0.0213,
1304
+ "step": 1770
1305
+ },
1306
+ {
1307
+ "epoch": 2.848,
1308
+ "grad_norm": 1.8074935674667358,
1309
+ "learning_rate": 6.783157894736843e-06,
1310
+ "loss": 0.018,
1311
+ "step": 1780
1312
+ },
1313
+ {
1314
+ "epoch": 2.864,
1315
+ "grad_norm": 0.8959200382232666,
1316
+ "learning_rate": 6.7621052631578956e-06,
1317
+ "loss": 0.0137,
1318
+ "step": 1790
1319
+ },
1320
+ {
1321
+ "epoch": 2.88,
1322
+ "grad_norm": 1.1795583963394165,
1323
+ "learning_rate": 6.741052631578948e-06,
1324
+ "loss": 0.0157,
1325
+ "step": 1800
1326
+ },
1327
+ {
1328
+ "epoch": 2.896,
1329
+ "grad_norm": 1.4148980379104614,
1330
+ "learning_rate": 6.720000000000001e-06,
1331
+ "loss": 0.0174,
1332
+ "step": 1810
1333
+ },
1334
+ {
1335
+ "epoch": 2.912,
1336
+ "grad_norm": 0.8638339042663574,
1337
+ "learning_rate": 6.698947368421053e-06,
1338
+ "loss": 0.0209,
1339
+ "step": 1820
1340
+ },
1341
+ {
1342
+ "epoch": 2.928,
1343
+ "grad_norm": 0.9411534667015076,
1344
+ "learning_rate": 6.677894736842105e-06,
1345
+ "loss": 0.0114,
1346
+ "step": 1830
1347
+ },
1348
+ {
1349
+ "epoch": 2.944,
1350
+ "grad_norm": 0.6806014180183411,
1351
+ "learning_rate": 6.656842105263158e-06,
1352
+ "loss": 0.02,
1353
+ "step": 1840
1354
+ },
1355
+ {
1356
+ "epoch": 2.96,
1357
+ "grad_norm": 0.9796069264411926,
1358
+ "learning_rate": 6.635789473684211e-06,
1359
+ "loss": 0.0116,
1360
+ "step": 1850
1361
+ },
1362
+ {
1363
+ "epoch": 2.976,
1364
+ "grad_norm": 3.364715814590454,
1365
+ "learning_rate": 6.614736842105264e-06,
1366
+ "loss": 0.0132,
1367
+ "step": 1860
1368
+ },
1369
+ {
1370
+ "epoch": 2.992,
1371
+ "grad_norm": 0.729645848274231,
1372
+ "learning_rate": 6.593684210526317e-06,
1373
+ "loss": 0.0147,
1374
+ "step": 1870
1375
+ },
1376
+ {
1377
+ "epoch": 3.008,
1378
+ "grad_norm": 2.3798277378082275,
1379
+ "learning_rate": 6.5726315789473695e-06,
1380
+ "loss": 0.0174,
1381
+ "step": 1880
1382
+ },
1383
+ {
1384
+ "epoch": 3.024,
1385
+ "grad_norm": 0.6777742505073547,
1386
+ "learning_rate": 6.551578947368421e-06,
1387
+ "loss": 0.0105,
1388
+ "step": 1890
1389
+ },
1390
+ {
1391
+ "epoch": 3.04,
1392
+ "grad_norm": 1.1942986249923706,
1393
+ "learning_rate": 6.530526315789474e-06,
1394
+ "loss": 0.0098,
1395
+ "step": 1900
1396
+ },
1397
+ {
1398
+ "epoch": 3.056,
1399
+ "grad_norm": 0.5124061107635498,
1400
+ "learning_rate": 6.509473684210527e-06,
1401
+ "loss": 0.0122,
1402
+ "step": 1910
1403
+ },
1404
+ {
1405
+ "epoch": 3.072,
1406
+ "grad_norm": 0.9624982476234436,
1407
+ "learning_rate": 6.488421052631579e-06,
1408
+ "loss": 0.0075,
1409
+ "step": 1920
1410
+ },
1411
+ {
1412
+ "epoch": 3.088,
1413
+ "grad_norm": 0.6776851415634155,
1414
+ "learning_rate": 6.467368421052632e-06,
1415
+ "loss": 0.0078,
1416
+ "step": 1930
1417
+ },
1418
+ {
1419
+ "epoch": 3.104,
1420
+ "grad_norm": 0.4030074179172516,
1421
+ "learning_rate": 6.4463157894736845e-06,
1422
+ "loss": 0.0078,
1423
+ "step": 1940
1424
+ },
1425
+ {
1426
+ "epoch": 3.12,
1427
+ "grad_norm": 0.5822834372520447,
1428
+ "learning_rate": 6.425263157894738e-06,
1429
+ "loss": 0.0058,
1430
+ "step": 1950
1431
+ },
1432
+ {
1433
+ "epoch": 3.136,
1434
+ "grad_norm": 0.23138459026813507,
1435
+ "learning_rate": 6.404210526315791e-06,
1436
+ "loss": 0.0062,
1437
+ "step": 1960
1438
+ },
1439
+ {
1440
+ "epoch": 3.152,
1441
+ "grad_norm": 1.4345389604568481,
1442
+ "learning_rate": 6.3831578947368425e-06,
1443
+ "loss": 0.009,
1444
+ "step": 1970
1445
+ },
1446
+ {
1447
+ "epoch": 3.168,
1448
+ "grad_norm": 1.3896571397781372,
1449
+ "learning_rate": 6.362105263157895e-06,
1450
+ "loss": 0.0091,
1451
+ "step": 1980
1452
+ },
1453
+ {
1454
+ "epoch": 3.184,
1455
+ "grad_norm": 0.7762678861618042,
1456
+ "learning_rate": 6.341052631578948e-06,
1457
+ "loss": 0.0078,
1458
+ "step": 1990
1459
+ },
1460
+ {
1461
+ "epoch": 3.2,
1462
+ "grad_norm": 0.7915733456611633,
1463
+ "learning_rate": 6.3200000000000005e-06,
1464
+ "loss": 0.0061,
1465
+ "step": 2000
1466
+ },
1467
+ {
1468
+ "epoch": 3.2,
1469
+ "eval_loss": 0.039765432476997375,
1470
+ "eval_runtime": 1647.5775,
1471
+ "eval_samples_per_second": 3.424,
1472
+ "eval_steps_per_second": 3.424,
1473
+ "step": 2000
1474
+ },
1475
+ {
1476
+ "epoch": 3.216,
1477
+ "grad_norm": 0.6915295124053955,
1478
+ "learning_rate": 6.298947368421053e-06,
1479
+ "loss": 0.0072,
1480
+ "step": 2010
1481
+ },
1482
+ {
1483
+ "epoch": 3.232,
1484
+ "grad_norm": 1.2451772689819336,
1485
+ "learning_rate": 6.277894736842106e-06,
1486
+ "loss": 0.0085,
1487
+ "step": 2020
1488
+ },
1489
+ {
1490
+ "epoch": 3.248,
1491
+ "grad_norm": 0.7011457085609436,
1492
+ "learning_rate": 6.256842105263159e-06,
1493
+ "loss": 0.0095,
1494
+ "step": 2030
1495
+ },
1496
+ {
1497
+ "epoch": 3.2640000000000002,
1498
+ "grad_norm": 0.3362152874469757,
1499
+ "learning_rate": 6.23578947368421e-06,
1500
+ "loss": 0.0068,
1501
+ "step": 2040
1502
+ },
1503
+ {
1504
+ "epoch": 3.2800000000000002,
1505
+ "grad_norm": 2.8105270862579346,
1506
+ "learning_rate": 6.214736842105264e-06,
1507
+ "loss": 0.0065,
1508
+ "step": 2050
1509
+ },
1510
+ {
1511
+ "epoch": 3.296,
1512
+ "grad_norm": 0.5386505722999573,
1513
+ "learning_rate": 6.193684210526316e-06,
1514
+ "loss": 0.0064,
1515
+ "step": 2060
1516
+ },
1517
+ {
1518
+ "epoch": 3.312,
1519
+ "grad_norm": 1.7395727634429932,
1520
+ "learning_rate": 6.172631578947369e-06,
1521
+ "loss": 0.0054,
1522
+ "step": 2070
1523
+ },
1524
+ {
1525
+ "epoch": 3.328,
1526
+ "grad_norm": 0.7333295345306396,
1527
+ "learning_rate": 6.151578947368422e-06,
1528
+ "loss": 0.0074,
1529
+ "step": 2080
1530
+ },
1531
+ {
1532
+ "epoch": 3.344,
1533
+ "grad_norm": 0.36293238401412964,
1534
+ "learning_rate": 6.130526315789474e-06,
1535
+ "loss": 0.0083,
1536
+ "step": 2090
1537
+ },
1538
+ {
1539
+ "epoch": 3.36,
1540
+ "grad_norm": 0.7955753207206726,
1541
+ "learning_rate": 6.109473684210527e-06,
1542
+ "loss": 0.0057,
1543
+ "step": 2100
1544
+ },
1545
+ {
1546
+ "epoch": 3.376,
1547
+ "grad_norm": 1.8287686109542847,
1548
+ "learning_rate": 6.088421052631579e-06,
1549
+ "loss": 0.0071,
1550
+ "step": 2110
1551
+ },
1552
+ {
1553
+ "epoch": 3.392,
1554
+ "grad_norm": 0.6369631886482239,
1555
+ "learning_rate": 6.0673684210526315e-06,
1556
+ "loss": 0.0073,
1557
+ "step": 2120
1558
+ },
1559
+ {
1560
+ "epoch": 3.408,
1561
+ "grad_norm": 0.9539582133293152,
1562
+ "learning_rate": 6.046315789473685e-06,
1563
+ "loss": 0.0114,
1564
+ "step": 2130
1565
+ },
1566
+ {
1567
+ "epoch": 3.424,
1568
+ "grad_norm": 0.8086543083190918,
1569
+ "learning_rate": 6.025263157894738e-06,
1570
+ "loss": 0.0099,
1571
+ "step": 2140
1572
+ },
1573
+ {
1574
+ "epoch": 3.44,
1575
+ "grad_norm": 0.40894851088523865,
1576
+ "learning_rate": 6.00421052631579e-06,
1577
+ "loss": 0.0066,
1578
+ "step": 2150
1579
+ },
1580
+ {
1581
+ "epoch": 3.456,
1582
+ "grad_norm": 0.39193499088287354,
1583
+ "learning_rate": 5.983157894736843e-06,
1584
+ "loss": 0.0044,
1585
+ "step": 2160
1586
+ },
1587
+ {
1588
+ "epoch": 3.472,
1589
+ "grad_norm": 1.0509319305419922,
1590
+ "learning_rate": 5.962105263157896e-06,
1591
+ "loss": 0.0081,
1592
+ "step": 2170
1593
+ },
1594
+ {
1595
+ "epoch": 3.488,
1596
+ "grad_norm": 2.615633964538574,
1597
+ "learning_rate": 5.941052631578947e-06,
1598
+ "loss": 0.0076,
1599
+ "step": 2180
1600
+ },
1601
+ {
1602
+ "epoch": 3.504,
1603
+ "grad_norm": 0.9589405059814453,
1604
+ "learning_rate": 5.92e-06,
1605
+ "loss": 0.0078,
1606
+ "step": 2190
1607
+ },
1608
+ {
1609
+ "epoch": 3.52,
1610
+ "grad_norm": 1.0844931602478027,
1611
+ "learning_rate": 5.898947368421053e-06,
1612
+ "loss": 0.0064,
1613
+ "step": 2200
1614
+ },
1615
+ {
1616
+ "epoch": 3.536,
1617
+ "grad_norm": 0.5187249183654785,
1618
+ "learning_rate": 5.8800000000000005e-06,
1619
+ "loss": 0.0094,
1620
+ "step": 2210
1621
+ },
1622
+ {
1623
+ "epoch": 3.552,
1624
+ "grad_norm": 1.4108637571334839,
1625
+ "learning_rate": 5.858947368421053e-06,
1626
+ "loss": 0.0075,
1627
+ "step": 2220
1628
+ },
1629
+ {
1630
+ "epoch": 3.568,
1631
+ "grad_norm": 0.7219350337982178,
1632
+ "learning_rate": 5.837894736842106e-06,
1633
+ "loss": 0.0091,
1634
+ "step": 2230
1635
+ },
1636
+ {
1637
+ "epoch": 3.584,
1638
+ "grad_norm": 0.8116509914398193,
1639
+ "learning_rate": 5.816842105263158e-06,
1640
+ "loss": 0.008,
1641
+ "step": 2240
1642
+ },
1643
+ {
1644
+ "epoch": 3.6,
1645
+ "grad_norm": 0.27414098381996155,
1646
+ "learning_rate": 5.795789473684211e-06,
1647
+ "loss": 0.0094,
1648
+ "step": 2250
1649
+ },
1650
+ {
1651
+ "epoch": 3.6,
1652
+ "eval_loss": 0.03814266622066498,
1653
+ "eval_runtime": 1450.6262,
1654
+ "eval_samples_per_second": 3.889,
1655
+ "eval_steps_per_second": 3.889,
1656
+ "step": 2250
1657
+ },
1658
+ {
1659
+ "epoch": 3.616,
1660
+ "grad_norm": 0.7917847633361816,
1661
+ "learning_rate": 5.774736842105264e-06,
1662
+ "loss": 0.0074,
1663
+ "step": 2260
1664
+ },
1665
+ {
1666
+ "epoch": 3.632,
1667
+ "grad_norm": 0.7660414576530457,
1668
+ "learning_rate": 5.7536842105263155e-06,
1669
+ "loss": 0.0081,
1670
+ "step": 2270
1671
+ },
1672
+ {
1673
+ "epoch": 3.648,
1674
+ "grad_norm": 0.9440973401069641,
1675
+ "learning_rate": 5.732631578947368e-06,
1676
+ "loss": 0.0072,
1677
+ "step": 2280
1678
+ },
1679
+ {
1680
+ "epoch": 3.664,
1681
+ "grad_norm": 1.79617440700531,
1682
+ "learning_rate": 5.711578947368422e-06,
1683
+ "loss": 0.0094,
1684
+ "step": 2290
1685
+ },
1686
+ {
1687
+ "epoch": 3.68,
1688
+ "grad_norm": 0.6188564896583557,
1689
+ "learning_rate": 5.690526315789474e-06,
1690
+ "loss": 0.0101,
1691
+ "step": 2300
1692
+ },
1693
+ {
1694
+ "epoch": 3.6959999999999997,
1695
+ "grad_norm": 1.2464138269424438,
1696
+ "learning_rate": 5.669473684210527e-06,
1697
+ "loss": 0.0073,
1698
+ "step": 2310
1699
+ },
1700
+ {
1701
+ "epoch": 3.7119999999999997,
1702
+ "grad_norm": 0.35902971029281616,
1703
+ "learning_rate": 5.64842105263158e-06,
1704
+ "loss": 0.007,
1705
+ "step": 2320
1706
+ },
1707
+ {
1708
+ "epoch": 3.7279999999999998,
1709
+ "grad_norm": 0.5012713670730591,
1710
+ "learning_rate": 5.627368421052632e-06,
1711
+ "loss": 0.0046,
1712
+ "step": 2330
1713
+ },
1714
+ {
1715
+ "epoch": 3.7439999999999998,
1716
+ "grad_norm": 1.280382752418518,
1717
+ "learning_rate": 5.606315789473684e-06,
1718
+ "loss": 0.0105,
1719
+ "step": 2340
1720
+ },
1721
+ {
1722
+ "epoch": 3.76,
1723
+ "grad_norm": 0.7401492595672607,
1724
+ "learning_rate": 5.585263157894737e-06,
1725
+ "loss": 0.0079,
1726
+ "step": 2350
1727
+ },
1728
+ {
1729
+ "epoch": 3.776,
1730
+ "grad_norm": 0.9203608632087708,
1731
+ "learning_rate": 5.5642105263157894e-06,
1732
+ "loss": 0.0052,
1733
+ "step": 2360
1734
+ },
1735
+ {
1736
+ "epoch": 3.792,
1737
+ "grad_norm": 1.1353886127471924,
1738
+ "learning_rate": 5.543157894736843e-06,
1739
+ "loss": 0.0087,
1740
+ "step": 2370
1741
+ },
1742
+ {
1743
+ "epoch": 3.808,
1744
+ "grad_norm": 1.4035215377807617,
1745
+ "learning_rate": 5.522105263157896e-06,
1746
+ "loss": 0.0077,
1747
+ "step": 2380
1748
+ },
1749
+ {
1750
+ "epoch": 3.824,
1751
+ "grad_norm": 0.429685115814209,
1752
+ "learning_rate": 5.501052631578948e-06,
1753
+ "loss": 0.0084,
1754
+ "step": 2390
1755
+ },
1756
+ {
1757
+ "epoch": 3.84,
1758
+ "grad_norm": 0.868377685546875,
1759
+ "learning_rate": 5.480000000000001e-06,
1760
+ "loss": 0.0114,
1761
+ "step": 2400
1762
+ },
1763
+ {
1764
+ "epoch": 3.856,
1765
+ "grad_norm": 0.9295462369918823,
1766
+ "learning_rate": 5.458947368421053e-06,
1767
+ "loss": 0.0066,
1768
+ "step": 2410
1769
+ },
1770
+ {
1771
+ "epoch": 3.872,
1772
+ "grad_norm": 1.1884615421295166,
1773
+ "learning_rate": 5.437894736842105e-06,
1774
+ "loss": 0.0083,
1775
+ "step": 2420
1776
+ },
1777
+ {
1778
+ "epoch": 3.888,
1779
+ "grad_norm": 0.9172056913375854,
1780
+ "learning_rate": 5.416842105263158e-06,
1781
+ "loss": 0.0071,
1782
+ "step": 2430
1783
+ },
1784
+ {
1785
+ "epoch": 3.904,
1786
+ "grad_norm": 0.7041633725166321,
1787
+ "learning_rate": 5.395789473684211e-06,
1788
+ "loss": 0.0069,
1789
+ "step": 2440
1790
+ },
1791
+ {
1792
+ "epoch": 3.92,
1793
+ "grad_norm": 0.7116276621818542,
1794
+ "learning_rate": 5.374736842105263e-06,
1795
+ "loss": 0.0076,
1796
+ "step": 2450
1797
+ },
1798
+ {
1799
+ "epoch": 3.936,
1800
+ "grad_norm": 0.7440093159675598,
1801
+ "learning_rate": 5.353684210526317e-06,
1802
+ "loss": 0.0076,
1803
+ "step": 2460
1804
+ },
1805
+ {
1806
+ "epoch": 3.952,
1807
+ "grad_norm": 1.1006033420562744,
1808
+ "learning_rate": 5.3326315789473695e-06,
1809
+ "loss": 0.0077,
1810
+ "step": 2470
1811
+ },
1812
+ {
1813
+ "epoch": 3.968,
1814
+ "grad_norm": 1.046329140663147,
1815
+ "learning_rate": 5.311578947368421e-06,
1816
+ "loss": 0.0089,
1817
+ "step": 2480
1818
+ },
1819
+ {
1820
+ "epoch": 3.984,
1821
+ "grad_norm": 0.9022756814956665,
1822
+ "learning_rate": 5.290526315789474e-06,
1823
+ "loss": 0.0073,
1824
+ "step": 2490
1825
+ },
1826
+ {
1827
+ "epoch": 4.0,
1828
+ "grad_norm": 1.0019030570983887,
1829
+ "learning_rate": 5.269473684210527e-06,
1830
+ "loss": 0.0074,
1831
+ "step": 2500
1832
+ },
1833
+ {
1834
+ "epoch": 4.0,
1835
+ "eval_loss": 0.03700412064790726,
1836
+ "eval_runtime": 1477.9736,
1837
+ "eval_samples_per_second": 3.817,
1838
+ "eval_steps_per_second": 3.817,
1839
+ "step": 2500
1840
+ },
1841
+ {
1842
+ "epoch": 4.016,
1843
+ "grad_norm": 1.2217806577682495,
1844
+ "learning_rate": 5.248421052631579e-06,
1845
+ "loss": 0.0037,
1846
+ "step": 2510
1847
+ },
1848
+ {
1849
+ "epoch": 4.032,
1850
+ "grad_norm": 0.7005161643028259,
1851
+ "learning_rate": 5.227368421052632e-06,
1852
+ "loss": 0.0047,
1853
+ "step": 2520
1854
+ },
1855
+ {
1856
+ "epoch": 4.048,
1857
+ "grad_norm": 0.3553329110145569,
1858
+ "learning_rate": 5.2063157894736846e-06,
1859
+ "loss": 0.0038,
1860
+ "step": 2530
1861
+ },
1862
+ {
1863
+ "epoch": 4.064,
1864
+ "grad_norm": 0.4705675542354584,
1865
+ "learning_rate": 5.185263157894738e-06,
1866
+ "loss": 0.0038,
1867
+ "step": 2540
1868
+ },
1869
+ {
1870
+ "epoch": 4.08,
1871
+ "grad_norm": 0.9150580763816833,
1872
+ "learning_rate": 5.164210526315791e-06,
1873
+ "loss": 0.0055,
1874
+ "step": 2550
1875
+ },
1876
+ {
1877
+ "epoch": 4.096,
1878
+ "grad_norm": 0.7440536618232727,
1879
+ "learning_rate": 5.1431578947368425e-06,
1880
+ "loss": 0.0064,
1881
+ "step": 2560
1882
+ },
1883
+ {
1884
+ "epoch": 4.112,
1885
+ "grad_norm": 0.16911202669143677,
1886
+ "learning_rate": 5.122105263157895e-06,
1887
+ "loss": 0.0033,
1888
+ "step": 2570
1889
+ },
1890
+ {
1891
+ "epoch": 4.128,
1892
+ "grad_norm": 0.20625616610050201,
1893
+ "learning_rate": 5.101052631578948e-06,
1894
+ "loss": 0.0029,
1895
+ "step": 2580
1896
+ },
1897
+ {
1898
+ "epoch": 4.144,
1899
+ "grad_norm": 1.1767489910125732,
1900
+ "learning_rate": 5.0800000000000005e-06,
1901
+ "loss": 0.0043,
1902
+ "step": 2590
1903
+ },
1904
+ {
1905
+ "epoch": 4.16,
1906
+ "grad_norm": 0.7702798247337341,
1907
+ "learning_rate": 5.058947368421053e-06,
1908
+ "loss": 0.002,
1909
+ "step": 2600
1910
+ },
1911
+ {
1912
+ "epoch": 4.176,
1913
+ "grad_norm": 0.9442148804664612,
1914
+ "learning_rate": 5.037894736842106e-06,
1915
+ "loss": 0.0049,
1916
+ "step": 2610
1917
+ },
1918
+ {
1919
+ "epoch": 4.192,
1920
+ "grad_norm": 0.439223974943161,
1921
+ "learning_rate": 5.0168421052631585e-06,
1922
+ "loss": 0.0033,
1923
+ "step": 2620
1924
+ },
1925
+ {
1926
+ "epoch": 4.208,
1927
+ "grad_norm": 0.2672136127948761,
1928
+ "learning_rate": 4.995789473684211e-06,
1929
+ "loss": 0.0055,
1930
+ "step": 2630
1931
+ },
1932
+ {
1933
+ "epoch": 4.224,
1934
+ "grad_norm": 0.1722344309091568,
1935
+ "learning_rate": 4.974736842105263e-06,
1936
+ "loss": 0.0049,
1937
+ "step": 2640
1938
+ },
1939
+ {
1940
+ "epoch": 4.24,
1941
+ "grad_norm": 0.8054834604263306,
1942
+ "learning_rate": 4.953684210526316e-06,
1943
+ "loss": 0.0028,
1944
+ "step": 2650
1945
+ },
1946
+ {
1947
+ "epoch": 4.256,
1948
+ "grad_norm": 0.38357770442962646,
1949
+ "learning_rate": 4.932631578947369e-06,
1950
+ "loss": 0.0022,
1951
+ "step": 2660
1952
+ },
1953
+ {
1954
+ "epoch": 4.272,
1955
+ "grad_norm": 0.7503048777580261,
1956
+ "learning_rate": 4.911578947368422e-06,
1957
+ "loss": 0.0021,
1958
+ "step": 2670
1959
+ },
1960
+ {
1961
+ "epoch": 4.288,
1962
+ "grad_norm": 0.95183265209198,
1963
+ "learning_rate": 4.8905263157894735e-06,
1964
+ "loss": 0.0035,
1965
+ "step": 2680
1966
+ },
1967
+ {
1968
+ "epoch": 4.304,
1969
+ "grad_norm": 0.9550835490226746,
1970
+ "learning_rate": 4.869473684210527e-06,
1971
+ "loss": 0.003,
1972
+ "step": 2690
1973
+ },
1974
+ {
1975
+ "epoch": 4.32,
1976
+ "grad_norm": 1.5608851909637451,
1977
+ "learning_rate": 4.84842105263158e-06,
1978
+ "loss": 0.0042,
1979
+ "step": 2700
1980
+ },
1981
+ {
1982
+ "epoch": 4.336,
1983
+ "grad_norm": 1.4856270551681519,
1984
+ "learning_rate": 4.8273684210526315e-06,
1985
+ "loss": 0.0048,
1986
+ "step": 2710
1987
+ },
1988
+ {
1989
+ "epoch": 4.352,
1990
+ "grad_norm": 0.11427481472492218,
1991
+ "learning_rate": 4.806315789473684e-06,
1992
+ "loss": 0.0022,
1993
+ "step": 2720
1994
+ },
1995
+ {
1996
+ "epoch": 4.368,
1997
+ "grad_norm": 1.5165557861328125,
1998
+ "learning_rate": 4.785263157894738e-06,
1999
+ "loss": 0.0037,
2000
+ "step": 2730
2001
+ },
2002
+ {
2003
+ "epoch": 4.384,
2004
+ "grad_norm": 0.30141228437423706,
2005
+ "learning_rate": 4.76421052631579e-06,
2006
+ "loss": 0.0026,
2007
+ "step": 2740
2008
+ },
2009
+ {
2010
+ "epoch": 4.4,
2011
+ "grad_norm": 0.4358047842979431,
2012
+ "learning_rate": 4.743157894736842e-06,
2013
+ "loss": 0.0046,
2014
+ "step": 2750
2015
+ },
2016
+ {
2017
+ "epoch": 4.4,
2018
+ "eval_loss": 0.0372794084250927,
2019
+ "eval_runtime": 1452.4236,
2020
+ "eval_samples_per_second": 3.885,
2021
+ "eval_steps_per_second": 3.885,
2022
+ "step": 2750
2023
+ },
2024
+ {
2025
+ "epoch": 4.416,
2026
+ "grad_norm": 0.8224077820777893,
2027
+ "learning_rate": 4.72421052631579e-06,
2028
+ "loss": 0.0046,
2029
+ "step": 2760
2030
+ },
2031
+ {
2032
+ "epoch": 4.432,
2033
+ "grad_norm": 1.4491019248962402,
2034
+ "learning_rate": 4.7031578947368425e-06,
2035
+ "loss": 0.0029,
2036
+ "step": 2770
2037
+ },
2038
+ {
2039
+ "epoch": 4.448,
2040
+ "grad_norm": 0.2398335337638855,
2041
+ "learning_rate": 4.682105263157895e-06,
2042
+ "loss": 0.0038,
2043
+ "step": 2780
2044
+ },
2045
+ {
2046
+ "epoch": 4.464,
2047
+ "grad_norm": 0.4416545629501343,
2048
+ "learning_rate": 4.661052631578948e-06,
2049
+ "loss": 0.0059,
2050
+ "step": 2790
2051
+ },
2052
+ {
2053
+ "epoch": 4.48,
2054
+ "grad_norm": 0.12480773776769638,
2055
+ "learning_rate": 4.6400000000000005e-06,
2056
+ "loss": 0.0036,
2057
+ "step": 2800
2058
+ },
2059
+ {
2060
+ "epoch": 4.496,
2061
+ "grad_norm": 0.9009574055671692,
2062
+ "learning_rate": 4.618947368421053e-06,
2063
+ "loss": 0.0044,
2064
+ "step": 2810
2065
+ },
2066
+ {
2067
+ "epoch": 4.5120000000000005,
2068
+ "grad_norm": 0.5740355253219604,
2069
+ "learning_rate": 4.597894736842106e-06,
2070
+ "loss": 0.0032,
2071
+ "step": 2820
2072
+ },
2073
+ {
2074
+ "epoch": 4.5280000000000005,
2075
+ "grad_norm": 0.6319159865379333,
2076
+ "learning_rate": 4.5768421052631584e-06,
2077
+ "loss": 0.0026,
2078
+ "step": 2830
2079
+ },
2080
+ {
2081
+ "epoch": 4.5440000000000005,
2082
+ "grad_norm": 0.710137665271759,
2083
+ "learning_rate": 4.55578947368421e-06,
2084
+ "loss": 0.0025,
2085
+ "step": 2840
2086
+ },
2087
+ {
2088
+ "epoch": 4.5600000000000005,
2089
+ "grad_norm": 0.3798495829105377,
2090
+ "learning_rate": 4.534736842105264e-06,
2091
+ "loss": 0.0032,
2092
+ "step": 2850
2093
+ },
2094
+ {
2095
+ "epoch": 4.576,
2096
+ "grad_norm": 0.09057649224996567,
2097
+ "learning_rate": 4.513684210526316e-06,
2098
+ "loss": 0.0019,
2099
+ "step": 2860
2100
+ },
2101
+ {
2102
+ "epoch": 4.592,
2103
+ "grad_norm": 0.10224127769470215,
2104
+ "learning_rate": 4.492631578947369e-06,
2105
+ "loss": 0.0024,
2106
+ "step": 2870
2107
+ },
2108
+ {
2109
+ "epoch": 4.608,
2110
+ "grad_norm": 0.45868900418281555,
2111
+ "learning_rate": 4.471578947368421e-06,
2112
+ "loss": 0.0033,
2113
+ "step": 2880
2114
+ },
2115
+ {
2116
+ "epoch": 4.624,
2117
+ "grad_norm": 0.12800952792167664,
2118
+ "learning_rate": 4.450526315789474e-06,
2119
+ "loss": 0.006,
2120
+ "step": 2890
2121
+ },
2122
+ {
2123
+ "epoch": 4.64,
2124
+ "grad_norm": 0.12955500185489655,
2125
+ "learning_rate": 4.429473684210527e-06,
2126
+ "loss": 0.0028,
2127
+ "step": 2900
2128
+ },
2129
+ {
2130
+ "epoch": 4.656,
2131
+ "grad_norm": 0.3282037675380707,
2132
+ "learning_rate": 4.408421052631579e-06,
2133
+ "loss": 0.0032,
2134
+ "step": 2910
2135
+ },
2136
+ {
2137
+ "epoch": 4.672,
2138
+ "grad_norm": 0.480395644903183,
2139
+ "learning_rate": 4.3873684210526315e-06,
2140
+ "loss": 0.0039,
2141
+ "step": 2920
2142
+ },
2143
+ {
2144
+ "epoch": 4.688,
2145
+ "grad_norm": 1.337486743927002,
2146
+ "learning_rate": 4.366315789473685e-06,
2147
+ "loss": 0.0053,
2148
+ "step": 2930
2149
+ },
2150
+ {
2151
+ "epoch": 4.704,
2152
+ "grad_norm": 0.39609357714653015,
2153
+ "learning_rate": 4.345263157894738e-06,
2154
+ "loss": 0.0026,
2155
+ "step": 2940
2156
+ },
2157
+ {
2158
+ "epoch": 4.72,
2159
+ "grad_norm": 0.2566475570201874,
2160
+ "learning_rate": 4.3242105263157894e-06,
2161
+ "loss": 0.0041,
2162
+ "step": 2950
2163
+ },
2164
+ {
2165
+ "epoch": 4.736,
2166
+ "grad_norm": 0.6467047929763794,
2167
+ "learning_rate": 4.303157894736842e-06,
2168
+ "loss": 0.0052,
2169
+ "step": 2960
2170
+ },
2171
+ {
2172
+ "epoch": 4.752,
2173
+ "grad_norm": 1.1100505590438843,
2174
+ "learning_rate": 4.282105263157896e-06,
2175
+ "loss": 0.0062,
2176
+ "step": 2970
2177
+ },
2178
+ {
2179
+ "epoch": 4.768,
2180
+ "grad_norm": 0.6272072196006775,
2181
+ "learning_rate": 4.261052631578947e-06,
2182
+ "loss": 0.0042,
2183
+ "step": 2980
2184
+ },
2185
+ {
2186
+ "epoch": 4.784,
2187
+ "grad_norm": 0.34918469190597534,
2188
+ "learning_rate": 4.24e-06,
2189
+ "loss": 0.0036,
2190
+ "step": 2990
2191
+ },
2192
+ {
2193
+ "epoch": 4.8,
2194
+ "grad_norm": 0.37302619218826294,
2195
+ "learning_rate": 4.218947368421053e-06,
2196
+ "loss": 0.0036,
2197
+ "step": 3000
2198
+ },
2199
+ {
2200
+ "epoch": 4.8,
2201
+ "eval_loss": 0.03739345446228981,
2202
+ "eval_runtime": 1465.397,
2203
+ "eval_samples_per_second": 3.85,
2204
+ "eval_steps_per_second": 3.85,
2205
+ "step": 3000
2206
+ },
2207
+ {
2208
+ "epoch": 4.816,
2209
+ "grad_norm": 0.5744602084159851,
2210
+ "learning_rate": 4.197894736842106e-06,
2211
+ "loss": 0.0061,
2212
+ "step": 3010
2213
+ },
2214
+ {
2215
+ "epoch": 4.832,
2216
+ "grad_norm": 1.2474104166030884,
2217
+ "learning_rate": 4.176842105263158e-06,
2218
+ "loss": 0.0035,
2219
+ "step": 3020
2220
+ },
2221
+ {
2222
+ "epoch": 4.848,
2223
+ "grad_norm": 0.25358426570892334,
2224
+ "learning_rate": 4.155789473684211e-06,
2225
+ "loss": 0.0039,
2226
+ "step": 3030
2227
+ },
2228
+ {
2229
+ "epoch": 4.864,
2230
+ "grad_norm": 1.7587385177612305,
2231
+ "learning_rate": 4.134736842105263e-06,
2232
+ "loss": 0.0043,
2233
+ "step": 3040
2234
+ },
2235
+ {
2236
+ "epoch": 4.88,
2237
+ "grad_norm": 0.5396108627319336,
2238
+ "learning_rate": 4.113684210526316e-06,
2239
+ "loss": 0.0039,
2240
+ "step": 3050
2241
+ },
2242
+ {
2243
+ "epoch": 4.896,
2244
+ "grad_norm": 0.2963917553424835,
2245
+ "learning_rate": 4.092631578947369e-06,
2246
+ "loss": 0.0037,
2247
+ "step": 3060
2248
+ },
2249
+ {
2250
+ "epoch": 4.912,
2251
+ "grad_norm": 0.6284441947937012,
2252
+ "learning_rate": 4.071578947368421e-06,
2253
+ "loss": 0.0042,
2254
+ "step": 3070
2255
+ },
2256
+ {
2257
+ "epoch": 4.928,
2258
+ "grad_norm": 0.1446247100830078,
2259
+ "learning_rate": 4.050526315789474e-06,
2260
+ "loss": 0.0054,
2261
+ "step": 3080
2262
+ },
2263
+ {
2264
+ "epoch": 4.944,
2265
+ "grad_norm": 0.5421654582023621,
2266
+ "learning_rate": 4.029473684210527e-06,
2267
+ "loss": 0.0021,
2268
+ "step": 3090
2269
+ },
2270
+ {
2271
+ "epoch": 4.96,
2272
+ "grad_norm": 0.10596631467342377,
2273
+ "learning_rate": 4.008421052631579e-06,
2274
+ "loss": 0.0032,
2275
+ "step": 3100
2276
+ },
2277
+ {
2278
+ "epoch": 4.976,
2279
+ "grad_norm": 0.36356231570243835,
2280
+ "learning_rate": 3.987368421052632e-06,
2281
+ "loss": 0.0029,
2282
+ "step": 3110
2283
+ },
2284
+ {
2285
+ "epoch": 4.992,
2286
+ "grad_norm": 0.13546526432037354,
2287
+ "learning_rate": 3.9663157894736846e-06,
2288
+ "loss": 0.0026,
2289
+ "step": 3120
2290
+ },
2291
+ {
2292
+ "epoch": 5.008,
2293
+ "grad_norm": 0.09022821485996246,
2294
+ "learning_rate": 3.945263157894737e-06,
2295
+ "loss": 0.0021,
2296
+ "step": 3130
2297
+ },
2298
+ {
2299
+ "epoch": 5.024,
2300
+ "grad_norm": 0.11343013495206833,
2301
+ "learning_rate": 3.92421052631579e-06,
2302
+ "loss": 0.0035,
2303
+ "step": 3140
2304
+ },
2305
+ {
2306
+ "epoch": 5.04,
2307
+ "grad_norm": 0.39134401082992554,
2308
+ "learning_rate": 3.9031578947368425e-06,
2309
+ "loss": 0.0018,
2310
+ "step": 3150
2311
+ },
2312
+ {
2313
+ "epoch": 5.056,
2314
+ "grad_norm": 0.05532646179199219,
2315
+ "learning_rate": 3.882105263157895e-06,
2316
+ "loss": 0.0015,
2317
+ "step": 3160
2318
+ },
2319
+ {
2320
+ "epoch": 5.072,
2321
+ "grad_norm": 0.8032190203666687,
2322
+ "learning_rate": 3.861052631578948e-06,
2323
+ "loss": 0.0011,
2324
+ "step": 3170
2325
+ },
2326
+ {
2327
+ "epoch": 5.088,
2328
+ "grad_norm": 0.05971115827560425,
2329
+ "learning_rate": 3.8400000000000005e-06,
2330
+ "loss": 0.0015,
2331
+ "step": 3180
2332
+ },
2333
+ {
2334
+ "epoch": 5.104,
2335
+ "grad_norm": 0.690596342086792,
2336
+ "learning_rate": 3.818947368421053e-06,
2337
+ "loss": 0.0024,
2338
+ "step": 3190
2339
+ },
2340
+ {
2341
+ "epoch": 5.12,
2342
+ "grad_norm": 0.15326976776123047,
2343
+ "learning_rate": 3.7978947368421054e-06,
2344
+ "loss": 0.0014,
2345
+ "step": 3200
2346
+ },
2347
+ {
2348
+ "epoch": 5.136,
2349
+ "grad_norm": 0.05565618351101875,
2350
+ "learning_rate": 3.776842105263158e-06,
2351
+ "loss": 0.0014,
2352
+ "step": 3210
2353
+ },
2354
+ {
2355
+ "epoch": 5.152,
2356
+ "grad_norm": 0.18008044362068176,
2357
+ "learning_rate": 3.755789473684211e-06,
2358
+ "loss": 0.0053,
2359
+ "step": 3220
2360
+ },
2361
+ {
2362
+ "epoch": 5.168,
2363
+ "grad_norm": 0.3844296932220459,
2364
+ "learning_rate": 3.7347368421052634e-06,
2365
+ "loss": 0.0018,
2366
+ "step": 3230
2367
+ },
2368
+ {
2369
+ "epoch": 5.184,
2370
+ "grad_norm": 0.15020981431007385,
2371
+ "learning_rate": 3.713684210526316e-06,
2372
+ "loss": 0.0013,
2373
+ "step": 3240
2374
+ },
2375
+ {
2376
+ "epoch": 5.2,
2377
+ "grad_norm": 0.2373928427696228,
2378
+ "learning_rate": 3.6926315789473687e-06,
2379
+ "loss": 0.0013,
2380
+ "step": 3250
2381
+ },
2382
+ {
2383
+ "epoch": 5.2,
2384
+ "eval_loss": 0.03558925539255142,
2385
+ "eval_runtime": 3288.1466,
2386
+ "eval_samples_per_second": 1.716,
2387
+ "eval_steps_per_second": 1.716,
2388
+ "step": 3250
2389
+ },
2390
+ {
2391
+ "epoch": 5.216,
2392
+ "grad_norm": 0.6427579522132874,
2393
+ "learning_rate": 3.6715789473684217e-06,
2394
+ "loss": 0.0024,
2395
+ "step": 3260
2396
+ },
2397
+ {
2398
+ "epoch": 5.232,
2399
+ "grad_norm": 0.09459923207759857,
2400
+ "learning_rate": 3.650526315789474e-06,
2401
+ "loss": 0.0014,
2402
+ "step": 3270
2403
+ },
2404
+ {
2405
+ "epoch": 5.248,
2406
+ "grad_norm": 0.4201672077178955,
2407
+ "learning_rate": 3.6294736842105266e-06,
2408
+ "loss": 0.0013,
2409
+ "step": 3280
2410
+ },
2411
+ {
2412
+ "epoch": 5.264,
2413
+ "grad_norm": 2.386017322540283,
2414
+ "learning_rate": 3.6084210526315793e-06,
2415
+ "loss": 0.0014,
2416
+ "step": 3290
2417
+ },
2418
+ {
2419
+ "epoch": 5.28,
2420
+ "grad_norm": 0.05354594066739082,
2421
+ "learning_rate": 3.5873684210526315e-06,
2422
+ "loss": 0.0012,
2423
+ "step": 3300
2424
+ },
2425
+ {
2426
+ "epoch": 5.296,
2427
+ "grad_norm": 0.4693075120449066,
2428
+ "learning_rate": 3.5663157894736846e-06,
2429
+ "loss": 0.0011,
2430
+ "step": 3310
2431
+ },
2432
+ {
2433
+ "epoch": 5.312,
2434
+ "grad_norm": 0.23063229024410248,
2435
+ "learning_rate": 3.5452631578947372e-06,
2436
+ "loss": 0.0011,
2437
+ "step": 3320
2438
+ },
2439
+ {
2440
+ "epoch": 5.328,
2441
+ "grad_norm": 0.6210919618606567,
2442
+ "learning_rate": 3.52421052631579e-06,
2443
+ "loss": 0.002,
2444
+ "step": 3330
2445
+ },
2446
+ {
2447
+ "epoch": 5.344,
2448
+ "grad_norm": 0.19046130776405334,
2449
+ "learning_rate": 3.503157894736842e-06,
2450
+ "loss": 0.0011,
2451
+ "step": 3340
2452
+ },
2453
+ {
2454
+ "epoch": 5.36,
2455
+ "grad_norm": 0.8910662531852722,
2456
+ "learning_rate": 3.482105263157895e-06,
2457
+ "loss": 0.0037,
2458
+ "step": 3350
2459
+ },
2460
+ {
2461
+ "epoch": 5.376,
2462
+ "grad_norm": 1.3482168912887573,
2463
+ "learning_rate": 3.461052631578948e-06,
2464
+ "loss": 0.0018,
2465
+ "step": 3360
2466
+ },
2467
+ {
2468
+ "epoch": 5.392,
2469
+ "grad_norm": 0.053552981466054916,
2470
+ "learning_rate": 3.44e-06,
2471
+ "loss": 0.001,
2472
+ "step": 3370
2473
+ },
2474
+ {
2475
+ "epoch": 5.408,
2476
+ "grad_norm": 0.165148064494133,
2477
+ "learning_rate": 3.4189473684210528e-06,
2478
+ "loss": 0.0018,
2479
+ "step": 3380
2480
+ },
2481
+ {
2482
+ "epoch": 5.424,
2483
+ "grad_norm": 0.1561366617679596,
2484
+ "learning_rate": 3.3978947368421054e-06,
2485
+ "loss": 0.0017,
2486
+ "step": 3390
2487
+ },
2488
+ {
2489
+ "epoch": 5.44,
2490
+ "grad_norm": 0.0557156503200531,
2491
+ "learning_rate": 3.3768421052631585e-06,
2492
+ "loss": 0.0011,
2493
+ "step": 3400
2494
+ },
2495
+ {
2496
+ "epoch": 5.456,
2497
+ "grad_norm": 0.045138657093048096,
2498
+ "learning_rate": 3.3557894736842107e-06,
2499
+ "loss": 0.0008,
2500
+ "step": 3410
2501
+ },
2502
+ {
2503
+ "epoch": 5.4719999999999995,
2504
+ "grad_norm": 0.6297252178192139,
2505
+ "learning_rate": 3.3347368421052634e-06,
2506
+ "loss": 0.0023,
2507
+ "step": 3420
2508
+ },
2509
+ {
2510
+ "epoch": 5.4879999999999995,
2511
+ "grad_norm": 0.431007444858551,
2512
+ "learning_rate": 3.313684210526316e-06,
2513
+ "loss": 0.0014,
2514
+ "step": 3430
2515
+ },
2516
+ {
2517
+ "epoch": 5.504,
2518
+ "grad_norm": 0.09312240779399872,
2519
+ "learning_rate": 3.292631578947369e-06,
2520
+ "loss": 0.002,
2521
+ "step": 3440
2522
+ },
2523
+ {
2524
+ "epoch": 5.52,
2525
+ "grad_norm": 0.19819211959838867,
2526
+ "learning_rate": 3.2715789473684213e-06,
2527
+ "loss": 0.001,
2528
+ "step": 3450
2529
+ },
2530
+ {
2531
+ "epoch": 5.536,
2532
+ "grad_norm": 0.09475424885749817,
2533
+ "learning_rate": 3.250526315789474e-06,
2534
+ "loss": 0.0019,
2535
+ "step": 3460
2536
+ },
2537
+ {
2538
+ "epoch": 5.552,
2539
+ "grad_norm": 0.034463245421648026,
2540
+ "learning_rate": 3.2294736842105266e-06,
2541
+ "loss": 0.0024,
2542
+ "step": 3470
2543
+ },
2544
+ {
2545
+ "epoch": 5.568,
2546
+ "grad_norm": 0.037909191101789474,
2547
+ "learning_rate": 3.208421052631579e-06,
2548
+ "loss": 0.0019,
2549
+ "step": 3480
2550
+ },
2551
+ {
2552
+ "epoch": 5.584,
2553
+ "grad_norm": 0.14103959500789642,
2554
+ "learning_rate": 3.187368421052632e-06,
2555
+ "loss": 0.0023,
2556
+ "step": 3490
2557
+ },
2558
+ {
2559
+ "epoch": 5.6,
2560
+ "grad_norm": 0.6614345908164978,
2561
+ "learning_rate": 3.1663157894736846e-06,
2562
+ "loss": 0.0014,
2563
+ "step": 3500
2564
+ },
2565
+ {
2566
+ "epoch": 5.6,
2567
+ "eval_loss": 0.03668233007192612,
2568
+ "eval_runtime": 1484.1952,
2569
+ "eval_samples_per_second": 3.801,
2570
+ "eval_steps_per_second": 3.801,
2571
+ "step": 3500
2572
+ },
2573
+ {
2574
+ "epoch": 5.616,
2575
+ "grad_norm": 0.051937129348516464,
2576
+ "learning_rate": 3.1452631578947373e-06,
2577
+ "loss": 0.0021,
2578
+ "step": 3510
2579
+ },
2580
+ {
2581
+ "epoch": 5.632,
2582
+ "grad_norm": 0.07696125656366348,
2583
+ "learning_rate": 3.1242105263157895e-06,
2584
+ "loss": 0.0017,
2585
+ "step": 3520
2586
+ },
2587
+ {
2588
+ "epoch": 5.648,
2589
+ "grad_norm": 0.2881506681442261,
2590
+ "learning_rate": 3.1031578947368426e-06,
2591
+ "loss": 0.0014,
2592
+ "step": 3530
2593
+ },
2594
+ {
2595
+ "epoch": 5.664,
2596
+ "grad_norm": 0.06252899765968323,
2597
+ "learning_rate": 3.0821052631578952e-06,
2598
+ "loss": 0.0017,
2599
+ "step": 3540
2600
+ },
2601
+ {
2602
+ "epoch": 5.68,
2603
+ "grad_norm": 0.4536062180995941,
2604
+ "learning_rate": 3.0610526315789475e-06,
2605
+ "loss": 0.0015,
2606
+ "step": 3550
2607
+ },
2608
+ {
2609
+ "epoch": 5.696,
2610
+ "grad_norm": 0.06475075334310532,
2611
+ "learning_rate": 3.04e-06,
2612
+ "loss": 0.0007,
2613
+ "step": 3560
2614
+ },
2615
+ {
2616
+ "epoch": 5.712,
2617
+ "grad_norm": 0.10876597464084625,
2618
+ "learning_rate": 3.0189473684210528e-06,
2619
+ "loss": 0.0012,
2620
+ "step": 3570
2621
+ },
2622
+ {
2623
+ "epoch": 5.728,
2624
+ "grad_norm": 0.04506557062268257,
2625
+ "learning_rate": 2.997894736842106e-06,
2626
+ "loss": 0.0012,
2627
+ "step": 3580
2628
+ },
2629
+ {
2630
+ "epoch": 5.744,
2631
+ "grad_norm": 0.08606194704771042,
2632
+ "learning_rate": 2.976842105263158e-06,
2633
+ "loss": 0.0008,
2634
+ "step": 3590
2635
+ },
2636
+ {
2637
+ "epoch": 5.76,
2638
+ "grad_norm": 0.10132581740617752,
2639
+ "learning_rate": 2.9557894736842107e-06,
2640
+ "loss": 0.0008,
2641
+ "step": 3600
2642
+ },
2643
+ {
2644
+ "epoch": 5.776,
2645
+ "grad_norm": 0.07705429196357727,
2646
+ "learning_rate": 2.9347368421052634e-06,
2647
+ "loss": 0.0006,
2648
+ "step": 3610
2649
+ },
2650
+ {
2651
+ "epoch": 5.792,
2652
+ "grad_norm": 0.09106363356113434,
2653
+ "learning_rate": 2.9136842105263156e-06,
2654
+ "loss": 0.0018,
2655
+ "step": 3620
2656
+ },
2657
+ {
2658
+ "epoch": 5.808,
2659
+ "grad_norm": 0.23617351055145264,
2660
+ "learning_rate": 2.8926315789473687e-06,
2661
+ "loss": 0.0019,
2662
+ "step": 3630
2663
+ },
2664
+ {
2665
+ "epoch": 5.824,
2666
+ "grad_norm": 0.20716822147369385,
2667
+ "learning_rate": 2.8715789473684214e-06,
2668
+ "loss": 0.0013,
2669
+ "step": 3640
2670
+ },
2671
+ {
2672
+ "epoch": 5.84,
2673
+ "grad_norm": 0.21701906621456146,
2674
+ "learning_rate": 2.850526315789474e-06,
2675
+ "loss": 0.0014,
2676
+ "step": 3650
2677
+ },
2678
+ {
2679
+ "epoch": 5.856,
2680
+ "grad_norm": 0.611148476600647,
2681
+ "learning_rate": 2.8294736842105262e-06,
2682
+ "loss": 0.0016,
2683
+ "step": 3660
2684
+ },
2685
+ {
2686
+ "epoch": 5.872,
2687
+ "grad_norm": 0.28972306847572327,
2688
+ "learning_rate": 2.8084210526315793e-06,
2689
+ "loss": 0.0021,
2690
+ "step": 3670
2691
+ },
2692
+ {
2693
+ "epoch": 5.888,
2694
+ "grad_norm": 0.13573560118675232,
2695
+ "learning_rate": 2.787368421052632e-06,
2696
+ "loss": 0.0023,
2697
+ "step": 3680
2698
+ },
2699
+ {
2700
+ "epoch": 5.904,
2701
+ "grad_norm": 0.15231332182884216,
2702
+ "learning_rate": 2.766315789473684e-06,
2703
+ "loss": 0.0017,
2704
+ "step": 3690
2705
+ },
2706
+ {
2707
+ "epoch": 5.92,
2708
+ "grad_norm": 0.24314993619918823,
2709
+ "learning_rate": 2.745263157894737e-06,
2710
+ "loss": 0.0015,
2711
+ "step": 3700
2712
+ },
2713
+ {
2714
+ "epoch": 5.936,
2715
+ "grad_norm": 0.2743528485298157,
2716
+ "learning_rate": 2.72421052631579e-06,
2717
+ "loss": 0.0011,
2718
+ "step": 3710
2719
+ },
2720
+ {
2721
+ "epoch": 5.952,
2722
+ "grad_norm": 0.20800799131393433,
2723
+ "learning_rate": 2.7031578947368426e-06,
2724
+ "loss": 0.0009,
2725
+ "step": 3720
2726
+ },
2727
+ {
2728
+ "epoch": 5.968,
2729
+ "grad_norm": 0.04552097246050835,
2730
+ "learning_rate": 2.682105263157895e-06,
2731
+ "loss": 0.0014,
2732
+ "step": 3730
2733
+ },
2734
+ {
2735
+ "epoch": 5.984,
2736
+ "grad_norm": 0.7857608199119568,
2737
+ "learning_rate": 2.6610526315789475e-06,
2738
+ "loss": 0.0022,
2739
+ "step": 3740
2740
+ },
2741
+ {
2742
+ "epoch": 6.0,
2743
+ "grad_norm": 0.8046532273292542,
2744
+ "learning_rate": 2.64e-06,
2745
+ "loss": 0.0018,
2746
+ "step": 3750
2747
+ },
2748
+ {
2749
+ "epoch": 6.0,
2750
+ "eval_loss": 0.037111204117536545,
2751
+ "eval_runtime": 1495.9727,
2752
+ "eval_samples_per_second": 3.771,
2753
+ "eval_steps_per_second": 3.771,
2754
+ "step": 3750
2755
+ },
2756
+ {
2757
+ "epoch": 6.016,
2758
+ "grad_norm": 0.09064371883869171,
2759
+ "learning_rate": 2.6189473684210524e-06,
2760
+ "loss": 0.0005,
2761
+ "step": 3760
2762
+ },
2763
+ {
2764
+ "epoch": 6.032,
2765
+ "grad_norm": 0.148435577750206,
2766
+ "learning_rate": 2.5978947368421054e-06,
2767
+ "loss": 0.0008,
2768
+ "step": 3770
2769
+ },
2770
+ {
2771
+ "epoch": 6.048,
2772
+ "grad_norm": 0.03715546801686287,
2773
+ "learning_rate": 2.576842105263158e-06,
2774
+ "loss": 0.0007,
2775
+ "step": 3780
2776
+ },
2777
+ {
2778
+ "epoch": 6.064,
2779
+ "grad_norm": 0.024771861732006073,
2780
+ "learning_rate": 2.5557894736842108e-06,
2781
+ "loss": 0.0005,
2782
+ "step": 3790
2783
+ },
2784
+ {
2785
+ "epoch": 6.08,
2786
+ "grad_norm": 0.08074558526277542,
2787
+ "learning_rate": 2.534736842105263e-06,
2788
+ "loss": 0.0013,
2789
+ "step": 3800
2790
+ },
2791
+ {
2792
+ "epoch": 6.096,
2793
+ "grad_norm": 0.054898303002119064,
2794
+ "learning_rate": 2.513684210526316e-06,
2795
+ "loss": 0.0007,
2796
+ "step": 3810
2797
+ },
2798
+ {
2799
+ "epoch": 6.112,
2800
+ "grad_norm": 0.3437640070915222,
2801
+ "learning_rate": 2.4926315789473687e-06,
2802
+ "loss": 0.0007,
2803
+ "step": 3820
2804
+ },
2805
+ {
2806
+ "epoch": 6.128,
2807
+ "grad_norm": 0.12333539873361588,
2808
+ "learning_rate": 2.4715789473684214e-06,
2809
+ "loss": 0.0004,
2810
+ "step": 3830
2811
+ },
2812
+ {
2813
+ "epoch": 6.144,
2814
+ "grad_norm": 0.027326831594109535,
2815
+ "learning_rate": 2.4505263157894736e-06,
2816
+ "loss": 0.0005,
2817
+ "step": 3840
2818
+ },
2819
+ {
2820
+ "epoch": 6.16,
2821
+ "grad_norm": 0.025773270055651665,
2822
+ "learning_rate": 2.4294736842105267e-06,
2823
+ "loss": 0.0005,
2824
+ "step": 3850
2825
+ },
2826
+ {
2827
+ "epoch": 6.176,
2828
+ "grad_norm": 0.04546317830681801,
2829
+ "learning_rate": 2.408421052631579e-06,
2830
+ "loss": 0.0009,
2831
+ "step": 3860
2832
+ },
2833
+ {
2834
+ "epoch": 6.192,
2835
+ "grad_norm": 0.054939109832048416,
2836
+ "learning_rate": 2.387368421052632e-06,
2837
+ "loss": 0.0005,
2838
+ "step": 3870
2839
+ },
2840
+ {
2841
+ "epoch": 6.208,
2842
+ "grad_norm": 0.17456945776939392,
2843
+ "learning_rate": 2.3663157894736842e-06,
2844
+ "loss": 0.0004,
2845
+ "step": 3880
2846
+ },
2847
+ {
2848
+ "epoch": 6.224,
2849
+ "grad_norm": 0.04125133529305458,
2850
+ "learning_rate": 2.3452631578947373e-06,
2851
+ "loss": 0.0004,
2852
+ "step": 3890
2853
+ },
2854
+ {
2855
+ "epoch": 6.24,
2856
+ "grad_norm": 0.2721328139305115,
2857
+ "learning_rate": 2.3242105263157895e-06,
2858
+ "loss": 0.0005,
2859
+ "step": 3900
2860
+ },
2861
+ {
2862
+ "epoch": 6.256,
2863
+ "grad_norm": 0.20841915905475616,
2864
+ "learning_rate": 2.3031578947368426e-06,
2865
+ "loss": 0.0004,
2866
+ "step": 3910
2867
+ },
2868
+ {
2869
+ "epoch": 6.272,
2870
+ "grad_norm": 0.024725254625082016,
2871
+ "learning_rate": 2.282105263157895e-06,
2872
+ "loss": 0.0011,
2873
+ "step": 3920
2874
+ },
2875
+ {
2876
+ "epoch": 6.288,
2877
+ "grad_norm": 0.04614824429154396,
2878
+ "learning_rate": 2.2610526315789475e-06,
2879
+ "loss": 0.0004,
2880
+ "step": 3930
2881
+ },
2882
+ {
2883
+ "epoch": 6.304,
2884
+ "grad_norm": 0.02417997643351555,
2885
+ "learning_rate": 2.24e-06,
2886
+ "loss": 0.0004,
2887
+ "step": 3940
2888
+ },
2889
+ {
2890
+ "epoch": 6.32,
2891
+ "grad_norm": 0.025578554719686508,
2892
+ "learning_rate": 2.218947368421053e-06,
2893
+ "loss": 0.0005,
2894
+ "step": 3950
2895
+ },
2896
+ {
2897
+ "epoch": 6.336,
2898
+ "grad_norm": 0.024059904739260674,
2899
+ "learning_rate": 2.1978947368421055e-06,
2900
+ "loss": 0.0004,
2901
+ "step": 3960
2902
+ },
2903
+ {
2904
+ "epoch": 6.352,
2905
+ "grad_norm": 0.01979261450469494,
2906
+ "learning_rate": 2.176842105263158e-06,
2907
+ "loss": 0.0006,
2908
+ "step": 3970
2909
+ },
2910
+ {
2911
+ "epoch": 6.368,
2912
+ "grad_norm": 0.0502333827316761,
2913
+ "learning_rate": 2.1557894736842108e-06,
2914
+ "loss": 0.0007,
2915
+ "step": 3980
2916
+ },
2917
+ {
2918
+ "epoch": 6.384,
2919
+ "grad_norm": 0.390522837638855,
2920
+ "learning_rate": 2.1347368421052634e-06,
2921
+ "loss": 0.0005,
2922
+ "step": 3990
2923
+ },
2924
+ {
2925
+ "epoch": 6.4,
2926
+ "grad_norm": 0.21075351536273956,
2927
+ "learning_rate": 2.1136842105263157e-06,
2928
+ "loss": 0.0008,
2929
+ "step": 4000
2930
+ },
2931
+ {
2932
+ "epoch": 6.4,
2933
+ "eval_loss": 0.03717663884162903,
2934
+ "eval_runtime": 1496.5829,
2935
+ "eval_samples_per_second": 3.77,
2936
+ "eval_steps_per_second": 3.77,
2937
+ "step": 4000
2938
+ },
2939
+ {
2940
+ "epoch": 6.416,
2941
+ "grad_norm": 0.08990900218486786,
2942
+ "learning_rate": 2.0926315789473687e-06,
2943
+ "loss": 0.0004,
2944
+ "step": 4010
2945
+ },
2946
+ {
2947
+ "epoch": 6.432,
2948
+ "grad_norm": 0.11204410344362259,
2949
+ "learning_rate": 2.071578947368421e-06,
2950
+ "loss": 0.0003,
2951
+ "step": 4020
2952
+ },
2953
+ {
2954
+ "epoch": 6.448,
2955
+ "grad_norm": 0.038647376000881195,
2956
+ "learning_rate": 2.050526315789474e-06,
2957
+ "loss": 0.0007,
2958
+ "step": 4030
2959
+ },
2960
+ {
2961
+ "epoch": 6.464,
2962
+ "grad_norm": 0.01824771985411644,
2963
+ "learning_rate": 2.0294736842105263e-06,
2964
+ "loss": 0.0003,
2965
+ "step": 4040
2966
+ },
2967
+ {
2968
+ "epoch": 6.48,
2969
+ "grad_norm": 0.02977917715907097,
2970
+ "learning_rate": 2.0084210526315794e-06,
2971
+ "loss": 0.0004,
2972
+ "step": 4050
2973
+ },
2974
+ {
2975
+ "epoch": 6.496,
2976
+ "grad_norm": 0.05292543023824692,
2977
+ "learning_rate": 1.9873684210526316e-06,
2978
+ "loss": 0.0005,
2979
+ "step": 4060
2980
+ },
2981
+ {
2982
+ "epoch": 6.5120000000000005,
2983
+ "grad_norm": 0.011542899534106255,
2984
+ "learning_rate": 1.9663157894736842e-06,
2985
+ "loss": 0.0005,
2986
+ "step": 4070
2987
+ },
2988
+ {
2989
+ "epoch": 6.5280000000000005,
2990
+ "grad_norm": 0.5990496277809143,
2991
+ "learning_rate": 1.945263157894737e-06,
2992
+ "loss": 0.0005,
2993
+ "step": 4080
2994
+ },
2995
+ {
2996
+ "epoch": 6.5440000000000005,
2997
+ "grad_norm": 0.8402637839317322,
2998
+ "learning_rate": 1.9242105263157896e-06,
2999
+ "loss": 0.0004,
3000
+ "step": 4090
3001
+ },
3002
+ {
3003
+ "epoch": 6.5600000000000005,
3004
+ "grad_norm": 0.02681225724518299,
3005
+ "learning_rate": 1.9031578947368424e-06,
3006
+ "loss": 0.0005,
3007
+ "step": 4100
3008
+ },
3009
+ {
3010
+ "epoch": 6.576,
3011
+ "grad_norm": 0.05136146768927574,
3012
+ "learning_rate": 1.8821052631578949e-06,
3013
+ "loss": 0.0009,
3014
+ "step": 4110
3015
+ },
3016
+ {
3017
+ "epoch": 6.592,
3018
+ "grad_norm": 0.02351960726082325,
3019
+ "learning_rate": 1.8610526315789475e-06,
3020
+ "loss": 0.0004,
3021
+ "step": 4120
3022
+ },
3023
+ {
3024
+ "epoch": 6.608,
3025
+ "grad_norm": 0.0421082004904747,
3026
+ "learning_rate": 1.8400000000000002e-06,
3027
+ "loss": 0.0005,
3028
+ "step": 4130
3029
+ },
3030
+ {
3031
+ "epoch": 6.624,
3032
+ "grad_norm": 0.039992865175008774,
3033
+ "learning_rate": 1.8189473684210528e-06,
3034
+ "loss": 0.0007,
3035
+ "step": 4140
3036
+ },
3037
+ {
3038
+ "epoch": 6.64,
3039
+ "grad_norm": 0.044527649879455566,
3040
+ "learning_rate": 1.7978947368421055e-06,
3041
+ "loss": 0.0014,
3042
+ "step": 4150
3043
+ },
3044
+ {
3045
+ "epoch": 6.656,
3046
+ "grad_norm": 0.3133164942264557,
3047
+ "learning_rate": 1.776842105263158e-06,
3048
+ "loss": 0.0011,
3049
+ "step": 4160
3050
+ },
3051
+ {
3052
+ "epoch": 6.672,
3053
+ "grad_norm": 0.013667403720319271,
3054
+ "learning_rate": 1.7557894736842108e-06,
3055
+ "loss": 0.0018,
3056
+ "step": 4170
3057
+ },
3058
+ {
3059
+ "epoch": 6.688,
3060
+ "grad_norm": 0.06877803057432175,
3061
+ "learning_rate": 1.7347368421052632e-06,
3062
+ "loss": 0.0003,
3063
+ "step": 4180
3064
+ },
3065
+ {
3066
+ "epoch": 6.704,
3067
+ "grad_norm": 0.03892038017511368,
3068
+ "learning_rate": 1.713684210526316e-06,
3069
+ "loss": 0.0003,
3070
+ "step": 4190
3071
+ },
3072
+ {
3073
+ "epoch": 6.72,
3074
+ "grad_norm": 0.05400446802377701,
3075
+ "learning_rate": 1.6926315789473685e-06,
3076
+ "loss": 0.0003,
3077
+ "step": 4200
3078
+ },
3079
+ {
3080
+ "epoch": 6.736,
3081
+ "grad_norm": 0.04348497465252876,
3082
+ "learning_rate": 1.6715789473684212e-06,
3083
+ "loss": 0.0012,
3084
+ "step": 4210
3085
+ },
3086
+ {
3087
+ "epoch": 6.752,
3088
+ "grad_norm": 0.1150202602148056,
3089
+ "learning_rate": 1.6505263157894739e-06,
3090
+ "loss": 0.0004,
3091
+ "step": 4220
3092
+ },
3093
+ {
3094
+ "epoch": 6.768,
3095
+ "grad_norm": 0.022777294740080833,
3096
+ "learning_rate": 1.6294736842105263e-06,
3097
+ "loss": 0.0003,
3098
+ "step": 4230
3099
+ },
3100
+ {
3101
+ "epoch": 6.784,
3102
+ "grad_norm": 0.05009898915886879,
3103
+ "learning_rate": 1.6084210526315792e-06,
3104
+ "loss": 0.0004,
3105
+ "step": 4240
3106
+ },
3107
+ {
3108
+ "epoch": 6.8,
3109
+ "grad_norm": 0.012213055975735188,
3110
+ "learning_rate": 1.5873684210526316e-06,
3111
+ "loss": 0.0003,
3112
+ "step": 4250
3113
+ },
3114
+ {
3115
+ "epoch": 6.8,
3116
+ "eval_loss": 0.03691783547401428,
3117
+ "eval_runtime": 1507.6111,
3118
+ "eval_samples_per_second": 3.742,
3119
+ "eval_steps_per_second": 3.742,
3120
+ "step": 4250
3121
+ },
3122
+ {
3123
+ "epoch": 6.816,
3124
+ "grad_norm": 0.23179693520069122,
3125
+ "learning_rate": 1.5663157894736845e-06,
3126
+ "loss": 0.0005,
3127
+ "step": 4260
3128
+ },
3129
+ {
3130
+ "epoch": 6.832,
3131
+ "grad_norm": 0.03861673176288605,
3132
+ "learning_rate": 1.545263157894737e-06,
3133
+ "loss": 0.0004,
3134
+ "step": 4270
3135
+ },
3136
+ {
3137
+ "epoch": 6.848,
3138
+ "grad_norm": 0.1452307105064392,
3139
+ "learning_rate": 1.5242105263157898e-06,
3140
+ "loss": 0.0003,
3141
+ "step": 4280
3142
+ },
3143
+ {
3144
+ "epoch": 6.864,
3145
+ "grad_norm": 0.050570085644721985,
3146
+ "learning_rate": 1.5031578947368422e-06,
3147
+ "loss": 0.0003,
3148
+ "step": 4290
3149
+ },
3150
+ {
3151
+ "epoch": 6.88,
3152
+ "grad_norm": 0.08024540543556213,
3153
+ "learning_rate": 1.482105263157895e-06,
3154
+ "loss": 0.0004,
3155
+ "step": 4300
3156
+ },
3157
+ {
3158
+ "epoch": 6.896,
3159
+ "grad_norm": 0.02591465599834919,
3160
+ "learning_rate": 1.4610526315789475e-06,
3161
+ "loss": 0.0005,
3162
+ "step": 4310
3163
+ },
3164
+ {
3165
+ "epoch": 6.912,
3166
+ "grad_norm": 0.01832268387079239,
3167
+ "learning_rate": 1.44e-06,
3168
+ "loss": 0.0004,
3169
+ "step": 4320
3170
+ },
3171
+ {
3172
+ "epoch": 6.928,
3173
+ "grad_norm": 0.044344332069158554,
3174
+ "learning_rate": 1.4189473684210528e-06,
3175
+ "loss": 0.0004,
3176
+ "step": 4330
3177
+ },
3178
+ {
3179
+ "epoch": 6.944,
3180
+ "grad_norm": 0.03570440039038658,
3181
+ "learning_rate": 1.3978947368421053e-06,
3182
+ "loss": 0.0003,
3183
+ "step": 4340
3184
+ },
3185
+ {
3186
+ "epoch": 6.96,
3187
+ "grad_norm": 0.18772557377815247,
3188
+ "learning_rate": 1.3768421052631582e-06,
3189
+ "loss": 0.0004,
3190
+ "step": 4350
3191
+ },
3192
+ {
3193
+ "epoch": 6.976,
3194
+ "grad_norm": 0.05984479933977127,
3195
+ "learning_rate": 1.3557894736842106e-06,
3196
+ "loss": 0.0003,
3197
+ "step": 4360
3198
+ },
3199
+ {
3200
+ "epoch": 6.992,
3201
+ "grad_norm": 0.3860021233558655,
3202
+ "learning_rate": 1.3347368421052635e-06,
3203
+ "loss": 0.0006,
3204
+ "step": 4370
3205
+ },
3206
+ {
3207
+ "epoch": 7.008,
3208
+ "grad_norm": 0.015999499708414078,
3209
+ "learning_rate": 1.313684210526316e-06,
3210
+ "loss": 0.0003,
3211
+ "step": 4380
3212
+ },
3213
+ {
3214
+ "epoch": 7.024,
3215
+ "grad_norm": 0.027498619630932808,
3216
+ "learning_rate": 1.2926315789473683e-06,
3217
+ "loss": 0.0002,
3218
+ "step": 4390
3219
+ },
3220
+ {
3221
+ "epoch": 7.04,
3222
+ "grad_norm": 0.24971920251846313,
3223
+ "learning_rate": 1.2715789473684212e-06,
3224
+ "loss": 0.0004,
3225
+ "step": 4400
3226
+ },
3227
+ {
3228
+ "epoch": 7.056,
3229
+ "grad_norm": 0.011065910570323467,
3230
+ "learning_rate": 1.2505263157894737e-06,
3231
+ "loss": 0.0003,
3232
+ "step": 4410
3233
+ },
3234
+ {
3235
+ "epoch": 7.072,
3236
+ "grad_norm": 0.01698007434606552,
3237
+ "learning_rate": 1.2294736842105263e-06,
3238
+ "loss": 0.0008,
3239
+ "step": 4420
3240
+ },
3241
+ {
3242
+ "epoch": 7.088,
3243
+ "grad_norm": 0.02532590739428997,
3244
+ "learning_rate": 1.208421052631579e-06,
3245
+ "loss": 0.0003,
3246
+ "step": 4430
3247
+ },
3248
+ {
3249
+ "epoch": 7.104,
3250
+ "grad_norm": 0.01716386154294014,
3251
+ "learning_rate": 1.1873684210526316e-06,
3252
+ "loss": 0.0003,
3253
+ "step": 4440
3254
+ },
3255
+ {
3256
+ "epoch": 7.12,
3257
+ "grad_norm": 0.016450321301817894,
3258
+ "learning_rate": 1.1663157894736843e-06,
3259
+ "loss": 0.0002,
3260
+ "step": 4450
3261
+ },
3262
+ {
3263
+ "epoch": 7.136,
3264
+ "grad_norm": 0.012627181597054005,
3265
+ "learning_rate": 1.145263157894737e-06,
3266
+ "loss": 0.0003,
3267
+ "step": 4460
3268
+ },
3269
+ {
3270
+ "epoch": 7.152,
3271
+ "grad_norm": 0.05683906376361847,
3272
+ "learning_rate": 1.1242105263157896e-06,
3273
+ "loss": 0.0003,
3274
+ "step": 4470
3275
+ },
3276
+ {
3277
+ "epoch": 7.168,
3278
+ "grad_norm": 0.03171288222074509,
3279
+ "learning_rate": 1.1031578947368422e-06,
3280
+ "loss": 0.0002,
3281
+ "step": 4480
3282
+ },
3283
+ {
3284
+ "epoch": 7.184,
3285
+ "grad_norm": 0.02577533759176731,
3286
+ "learning_rate": 1.082105263157895e-06,
3287
+ "loss": 0.0003,
3288
+ "step": 4490
3289
+ },
3290
+ {
3291
+ "epoch": 7.2,
3292
+ "grad_norm": 0.019202925264835358,
3293
+ "learning_rate": 1.0610526315789473e-06,
3294
+ "loss": 0.0003,
3295
+ "step": 4500
3296
+ },
3297
+ {
3298
+ "epoch": 7.2,
3299
+ "eval_loss": 0.037218254059553146,
3300
+ "eval_runtime": 1593.609,
3301
+ "eval_samples_per_second": 3.54,
3302
+ "eval_steps_per_second": 3.54,
3303
+ "step": 4500
3304
+ },
3305
+ {
3306
+ "epoch": 7.216,
3307
+ "grad_norm": 0.01943960040807724,
3308
+ "learning_rate": 1.04e-06,
3309
+ "loss": 0.0003,
3310
+ "step": 4510
3311
+ },
3312
+ {
3313
+ "epoch": 7.232,
3314
+ "grad_norm": 0.012748222798109055,
3315
+ "learning_rate": 1.0189473684210527e-06,
3316
+ "loss": 0.0002,
3317
+ "step": 4520
3318
+ },
3319
+ {
3320
+ "epoch": 7.248,
3321
+ "grad_norm": 0.017005812376737595,
3322
+ "learning_rate": 9.978947368421053e-07,
3323
+ "loss": 0.0003,
3324
+ "step": 4530
3325
+ },
3326
+ {
3327
+ "epoch": 7.264,
3328
+ "grad_norm": 0.024691853672266006,
3329
+ "learning_rate": 9.76842105263158e-07,
3330
+ "loss": 0.0002,
3331
+ "step": 4540
3332
+ },
3333
+ {
3334
+ "epoch": 7.28,
3335
+ "grad_norm": 0.3427879512310028,
3336
+ "learning_rate": 9.557894736842106e-07,
3337
+ "loss": 0.0004,
3338
+ "step": 4550
3339
+ },
3340
+ {
3341
+ "epoch": 7.296,
3342
+ "grad_norm": 0.03403201699256897,
3343
+ "learning_rate": 9.347368421052633e-07,
3344
+ "loss": 0.0002,
3345
+ "step": 4560
3346
+ },
3347
+ {
3348
+ "epoch": 7.312,
3349
+ "grad_norm": 0.022941039875149727,
3350
+ "learning_rate": 9.136842105263159e-07,
3351
+ "loss": 0.0003,
3352
+ "step": 4570
3353
+ },
3354
+ {
3355
+ "epoch": 7.328,
3356
+ "grad_norm": 0.032287679612636566,
3357
+ "learning_rate": 8.926315789473685e-07,
3358
+ "loss": 0.0002,
3359
+ "step": 4580
3360
+ },
3361
+ {
3362
+ "epoch": 7.344,
3363
+ "grad_norm": 0.013065003789961338,
3364
+ "learning_rate": 8.715789473684211e-07,
3365
+ "loss": 0.0002,
3366
+ "step": 4590
3367
+ },
3368
+ {
3369
+ "epoch": 7.36,
3370
+ "grad_norm": 0.026764079928398132,
3371
+ "learning_rate": 8.505263157894737e-07,
3372
+ "loss": 0.0002,
3373
+ "step": 4600
3374
+ },
3375
+ {
3376
+ "epoch": 7.376,
3377
+ "grad_norm": 0.013478124514222145,
3378
+ "learning_rate": 8.294736842105263e-07,
3379
+ "loss": 0.0002,
3380
+ "step": 4610
3381
+ },
3382
+ {
3383
+ "epoch": 7.392,
3384
+ "grad_norm": 0.014998686499893665,
3385
+ "learning_rate": 8.08421052631579e-07,
3386
+ "loss": 0.0003,
3387
+ "step": 4620
3388
+ },
3389
+ {
3390
+ "epoch": 7.408,
3391
+ "grad_norm": 0.015325380489230156,
3392
+ "learning_rate": 7.873684210526316e-07,
3393
+ "loss": 0.0002,
3394
+ "step": 4630
3395
+ },
3396
+ {
3397
+ "epoch": 7.424,
3398
+ "grad_norm": 0.018431052565574646,
3399
+ "learning_rate": 7.663157894736843e-07,
3400
+ "loss": 0.0002,
3401
+ "step": 4640
3402
+ },
3403
+ {
3404
+ "epoch": 7.44,
3405
+ "grad_norm": 0.01385441143065691,
3406
+ "learning_rate": 7.45263157894737e-07,
3407
+ "loss": 0.0003,
3408
+ "step": 4650
3409
+ },
3410
+ {
3411
+ "epoch": 7.456,
3412
+ "grad_norm": 0.018942181020975113,
3413
+ "learning_rate": 7.242105263157895e-07,
3414
+ "loss": 0.0002,
3415
+ "step": 4660
3416
+ },
3417
+ {
3418
+ "epoch": 7.4719999999999995,
3419
+ "grad_norm": 0.036915648728609085,
3420
+ "learning_rate": 7.031578947368422e-07,
3421
+ "loss": 0.0002,
3422
+ "step": 4670
3423
+ },
3424
+ {
3425
+ "epoch": 7.4879999999999995,
3426
+ "grad_norm": 0.053086813539266586,
3427
+ "learning_rate": 6.821052631578948e-07,
3428
+ "loss": 0.0002,
3429
+ "step": 4680
3430
+ },
3431
+ {
3432
+ "epoch": 7.504,
3433
+ "grad_norm": 0.028192859143018723,
3434
+ "learning_rate": 6.610526315789475e-07,
3435
+ "loss": 0.0003,
3436
+ "step": 4690
3437
+ },
3438
+ {
3439
+ "epoch": 7.52,
3440
+ "grad_norm": 0.6953186988830566,
3441
+ "learning_rate": 6.4e-07,
3442
+ "loss": 0.0005,
3443
+ "step": 4700
3444
+ },
3445
+ {
3446
+ "epoch": 7.536,
3447
+ "grad_norm": 0.016007578000426292,
3448
+ "learning_rate": 6.189473684210527e-07,
3449
+ "loss": 0.0002,
3450
+ "step": 4710
3451
+ },
3452
+ {
3453
+ "epoch": 7.552,
3454
+ "grad_norm": 0.029922867193818092,
3455
+ "learning_rate": 5.978947368421053e-07,
3456
+ "loss": 0.0003,
3457
+ "step": 4720
3458
+ },
3459
+ {
3460
+ "epoch": 7.568,
3461
+ "grad_norm": 0.01892116479575634,
3462
+ "learning_rate": 5.76842105263158e-07,
3463
+ "loss": 0.0003,
3464
+ "step": 4730
3465
+ },
3466
+ {
3467
+ "epoch": 7.584,
3468
+ "grad_norm": 0.03329944983124733,
3469
+ "learning_rate": 5.557894736842106e-07,
3470
+ "loss": 0.0002,
3471
+ "step": 4740
3472
+ },
3473
+ {
3474
+ "epoch": 7.6,
3475
+ "grad_norm": 0.025351159274578094,
3476
+ "learning_rate": 5.347368421052633e-07,
3477
+ "loss": 0.0002,
3478
+ "step": 4750
3479
+ },
3480
+ {
3481
+ "epoch": 7.6,
3482
+ "eval_loss": 0.03762885555624962,
3483
+ "eval_runtime": 1462.6796,
3484
+ "eval_samples_per_second": 3.857,
3485
+ "eval_steps_per_second": 3.857,
3486
+ "step": 4750
3487
+ },
3488
+ {
3489
+ "epoch": 7.616,
3490
+ "grad_norm": 0.017339129000902176,
3491
+ "learning_rate": 5.136842105263158e-07,
3492
+ "loss": 0.0002,
3493
+ "step": 4760
3494
+ },
3495
+ {
3496
+ "epoch": 7.632,
3497
+ "grad_norm": 0.016879908740520477,
3498
+ "learning_rate": 4.926315789473685e-07,
3499
+ "loss": 0.0004,
3500
+ "step": 4770
3501
+ },
3502
+ {
3503
+ "epoch": 7.648,
3504
+ "grad_norm": 0.02509094960987568,
3505
+ "learning_rate": 4.715789473684211e-07,
3506
+ "loss": 0.0002,
3507
+ "step": 4780
3508
+ },
3509
+ {
3510
+ "epoch": 7.664,
3511
+ "grad_norm": 0.017198877409100533,
3512
+ "learning_rate": 4.505263157894737e-07,
3513
+ "loss": 0.0003,
3514
+ "step": 4790
3515
+ },
3516
+ {
3517
+ "epoch": 7.68,
3518
+ "grad_norm": 0.014106789603829384,
3519
+ "learning_rate": 4.2947368421052635e-07,
3520
+ "loss": 0.0002,
3521
+ "step": 4800
3522
+ },
3523
+ {
3524
+ "epoch": 7.696,
3525
+ "grad_norm": 0.03618380054831505,
3526
+ "learning_rate": 4.08421052631579e-07,
3527
+ "loss": 0.0002,
3528
+ "step": 4810
3529
+ },
3530
+ {
3531
+ "epoch": 7.712,
3532
+ "grad_norm": 0.013972431421279907,
3533
+ "learning_rate": 3.873684210526316e-07,
3534
+ "loss": 0.0002,
3535
+ "step": 4820
3536
+ },
3537
+ {
3538
+ "epoch": 7.728,
3539
+ "grad_norm": 0.01894882321357727,
3540
+ "learning_rate": 3.663157894736842e-07,
3541
+ "loss": 0.0002,
3542
+ "step": 4830
3543
+ },
3544
+ {
3545
+ "epoch": 7.744,
3546
+ "grad_norm": 0.01931534893810749,
3547
+ "learning_rate": 3.4526315789473686e-07,
3548
+ "loss": 0.0003,
3549
+ "step": 4840
3550
+ },
3551
+ {
3552
+ "epoch": 7.76,
3553
+ "grad_norm": 0.015359962359070778,
3554
+ "learning_rate": 3.242105263157895e-07,
3555
+ "loss": 0.0002,
3556
+ "step": 4850
3557
+ },
3558
+ {
3559
+ "epoch": 7.776,
3560
+ "grad_norm": 0.0260631013661623,
3561
+ "learning_rate": 3.031578947368421e-07,
3562
+ "loss": 0.0002,
3563
+ "step": 4860
3564
+ },
3565
+ {
3566
+ "epoch": 7.792,
3567
+ "grad_norm": 0.06113347038626671,
3568
+ "learning_rate": 2.821052631578948e-07,
3569
+ "loss": 0.0002,
3570
+ "step": 4870
3571
+ },
3572
+ {
3573
+ "epoch": 7.808,
3574
+ "grad_norm": 0.0257665254175663,
3575
+ "learning_rate": 2.610526315789474e-07,
3576
+ "loss": 0.0003,
3577
+ "step": 4880
3578
+ },
3579
+ {
3580
+ "epoch": 7.824,
3581
+ "grad_norm": 0.01588517054915428,
3582
+ "learning_rate": 2.4000000000000003e-07,
3583
+ "loss": 0.0002,
3584
+ "step": 4890
3585
+ },
3586
+ {
3587
+ "epoch": 7.84,
3588
+ "grad_norm": 0.012915355153381824,
3589
+ "learning_rate": 2.1894736842105263e-07,
3590
+ "loss": 0.0002,
3591
+ "step": 4900
3592
+ },
3593
+ {
3594
+ "epoch": 7.856,
3595
+ "grad_norm": 0.01570785976946354,
3596
+ "learning_rate": 1.978947368421053e-07,
3597
+ "loss": 0.0002,
3598
+ "step": 4910
3599
+ },
3600
+ {
3601
+ "epoch": 7.872,
3602
+ "grad_norm": 0.020730208605527878,
3603
+ "learning_rate": 1.768421052631579e-07,
3604
+ "loss": 0.0002,
3605
+ "step": 4920
3606
+ },
3607
+ {
3608
+ "epoch": 7.888,
3609
+ "grad_norm": 0.03537014126777649,
3610
+ "learning_rate": 1.5578947368421054e-07,
3611
+ "loss": 0.0003,
3612
+ "step": 4930
3613
+ },
3614
+ {
3615
+ "epoch": 7.904,
3616
+ "grad_norm": 0.022489851340651512,
3617
+ "learning_rate": 1.3473684210526317e-07,
3618
+ "loss": 0.0002,
3619
+ "step": 4940
3620
+ },
3621
+ {
3622
+ "epoch": 7.92,
3623
+ "grad_norm": 0.017718670889735222,
3624
+ "learning_rate": 1.136842105263158e-07,
3625
+ "loss": 0.0002,
3626
+ "step": 4950
3627
+ },
3628
+ {
3629
+ "epoch": 7.936,
3630
+ "grad_norm": 0.016573410481214523,
3631
+ "learning_rate": 9.263157894736844e-08,
3632
+ "loss": 0.0007,
3633
+ "step": 4960
3634
+ },
3635
+ {
3636
+ "epoch": 7.952,
3637
+ "grad_norm": 0.008749530650675297,
3638
+ "learning_rate": 7.157894736842106e-08,
3639
+ "loss": 0.0003,
3640
+ "step": 4970
3641
+ },
3642
+ {
3643
+ "epoch": 7.968,
3644
+ "grad_norm": 0.04577890783548355,
3645
+ "learning_rate": 5.0526315789473687e-08,
3646
+ "loss": 0.0003,
3647
+ "step": 4980
3648
+ },
3649
+ {
3650
+ "epoch": 7.984,
3651
+ "grad_norm": 0.061525970697402954,
3652
+ "learning_rate": 2.9473684210526318e-08,
3653
+ "loss": 0.0002,
3654
+ "step": 4990
3655
+ },
3656
+ {
3657
+ "epoch": 8.0,
3658
+ "grad_norm": 0.029113447293639183,
3659
+ "learning_rate": 8.421052631578948e-09,
3660
+ "loss": 0.0002,
3661
+ "step": 5000
3662
+ },
3663
+ {
3664
+ "epoch": 8.0,
3665
+ "eval_loss": 0.03780149295926094,
3666
+ "eval_runtime": 1480.7553,
3667
+ "eval_samples_per_second": 3.81,
3668
+ "eval_steps_per_second": 3.81,
3669
+ "step": 5000
3670
+ }
3671
+ ],
3672
+ "logging_steps": 10,
3673
+ "max_steps": 5000,
3674
+ "num_input_tokens_seen": 0,
3675
+ "num_train_epochs": 8,
3676
+ "save_steps": 500,
3677
+ "stateful_callbacks": {
3678
+ "TrainerControl": {
3679
+ "args": {
3680
+ "should_epoch_stop": false,
3681
+ "should_evaluate": false,
3682
+ "should_log": false,
3683
+ "should_save": true,
3684
+ "should_training_stop": true
3685
+ },
3686
+ "attributes": {}
3687
+ }
3688
+ },
3689
+ "total_flos": 2.727921844224e+20,
3690
+ "train_batch_size": 2,
3691
+ "trial_name": null,
3692
+ "trial_params": null
3693
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f73fd03d1966e401cedcf380fde8f9ebf7254af0885e9c8340e4adb16363e329
3
+ size 5304