-
Notifications
You must be signed in to change notification settings - Fork 4
/
mt.out
192 lines (192 loc) · 13.3 KB
/
mt.out
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
nohup: ignoring input
| distributed init (rank 4): tcp://localhost:15212
| distributed init (rank 1): tcp://localhost:15212
| distributed init (rank 2): tcp://localhost:15212
| distributed init (rank 3): tcp://localhost:15212
| distributed init (rank 0): tcp://localhost:15212
Namespace(adam_betas='(0.9, 0.98)', adam_eps=1e-08, adaptive_softmax_cutoff=None, arch='transformer', attention_dropout=0.0, clip_norm=0.0, criterion='label_smoothed_cross_entropy', data='/data2/mmyin/data-raw/summ/tsinghua/lcsts-32k-joint', decoder_attention_heads=8, decoder_embed_dim=512, decoder_embed_path=None, decoder_ffn_embed_dim=2048, decoder_layers=6, decoder_learned_pos=False, decoder_normalize_before=False, device_id=0, distributed_backend='nccl', distributed_init_method='tcp://localhost:15212', distributed_port=-1, distributed_rank=0, distributed_world_size=5, dropout=0.3, encoder_attention_heads=8, encoder_embed_dim=512, encoder_embed_path=None, encoder_ffn_embed_dim=2048, encoder_layers=6, encoder_learned_pos=False, encoder_normalize_before=False, fix_dec_emb=False, fix_enc_emb=False, fp16=False, keep_interval_updates=-1, label_smoothing=0.1, left_pad_source='True', left_pad_target='False', log_format=None, log_interval=1000, lr=[0.0005], lr_scheduler='inverse_sqrt', lr_shrink=0.1, max_epoch=0, max_sentences=None, max_sentences_valid=None, max_source_positions=1024, max_target_positions=1024, max_tokens=5500, max_update=0, min_loss_scale=0.0001, min_lr=1e-09, momentum=0.99, no_epoch_checkpoints=False, no_progress_bar=False, no_save=False, no_token_positional_embeddings=False, optimizer='adam', raw_text=False, relu_dropout=0.0, restore_file='checkpoint_last.pt', save_dir='checkpoints/tsinghua/tmp', save_interval=1, save_interval_updates=0, seed=1, sentence_avg=False, share_all_embeddings=True, share_decoder_input_output_embed=False, skip_invalid_size_inputs_valid_test=False, source_lang='src', target_lang='tgt', task='translation', train_subset='train', update_freq=[1], valid_subset='valid', validate_interval=1, warmup_init_lr=1e-07, warmup_updates=5000, weight_decay=0.0)
| [src] dictionary: 48608 types
| [tgt] dictionary: 48608 types
| /data2/mmyin/data-raw/summ/tsinghua/lcsts-32k-joint train 2400591 examples
| /data2/mmyin/data-raw/summ/tsinghua/lcsts-32k-joint valid 8685 examples
TransformerModel(
(encoder): TransformerEncoder(
(embed_tokens): Embedding(48608, 512, padding_idx=1)
(embed_positions): SinusoidalPositionalEmbedding()
(layers): ModuleList(
(0): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
(1): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
(2): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
(3): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
(4): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
(5): TransformerEncoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(layer_norms): ModuleList(
(0): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(1): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
)
)
(decoder): TransformerDecoder(
(embed_tokens): Embedding(48608, 512, padding_idx=1)
(embed_positions): SinusoidalPositionalEmbedding()
(layers): ModuleList(
(0): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
(1): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
(2): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
(3): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
(4): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
(5): TransformerDecoderLayer(
(self_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(self_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(encoder_attn): MultiheadAttention(
(out_proj): Linear(in_features=512, out_features=512, bias=True)
)
(encoder_attn_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
(fc1): Linear(in_features=512, out_features=2048, bias=True)
(fc2): Linear(in_features=2048, out_features=512, bias=True)
(final_layer_norm): LayerNorm(torch.Size([512]), eps=1e-05, elementwise_affine=True)
)
)
)
)
| model transformer, criterion LabelSmoothedCrossEntropyCriterion
| num. model params: 69025792
| training on 5 GPUs
| max tokens per GPU = 5500 and max sentences per GPU = None
| epoch 001: 1000 / 6099 loss=12.884, nll_loss=12.427, ppl=5504.91, acc=10.28, wps=14608, ups=3.0, wpb=4691, bsz=394, num_updates=1001, lr=0.00010018, gnorm=1.508, clip=100%, oom=0, wall=338
| epoch 001: 2000 / 6099 loss=12.050, nll_loss=11.462, ppl=2821.60, acc=12.17, wps=14530, ups=3.0, wpb=4682, bsz=394, num_updates=2001, lr=0.00020016, gnorm=1.399, clip=100%, oom=0, wall=661
| epoch 001: 3000 / 6099 loss=11.397, nll_loss=10.707, ppl=1671.91, acc=13.94, wps=14517, ups=3.0, wpb=4684, bsz=394, num_updates=3001, lr=0.00030014, gnorm=1.310, clip=100%, oom=0, wall=985
| epoch 001: 4000 / 6099 loss=10.882, nll_loss=10.110, ppl=1105.33, acc=15.7, wps=14510, ups=3.1, wpb=4681, bsz=394, num_updates=4001, lr=0.00040012, gnorm=1.252, clip=100%, oom=0, wall=1307
| epoch 001: 5000 / 6099 loss=10.466, nll_loss=9.628, ppl=791.18, acc=17.4, wps=14497, ups=3.1, wpb=4679, bsz=394, num_updates=5001, lr=0.00049995, gnorm=1.223, clip=100%, oom=0, wall=1631
| epoch 001: 6000 / 6099 loss=10.109, nll_loss=9.214, ppl=594.00, acc=19.14, wps=14509, ups=3.1, wpb=4680, bsz=394, num_updates=6001, lr=0.000456397, gnorm=1.209, clip=100%, oom=0, wall=1952
| epoch 001 | loss 10.076 | nll_loss 9.177 | ppl 578.67 | acc 19.31 | wps 14506 | ups 3.1 | wpb 4679 | bsz 394 | num_updates 6099 | lr 0.000452716 | gnorm 1.208 | clip 100% | oom 0 | wall 1984
| epoch 001 | valid on 'valid' subset | valid_loss 8.06991 | acc 31.55 | valid_nll_loss 6.73704 | valid_ppl 106.67 | num_updates 6099
| epoch 002: 1000 / 6099 loss=7.792, nll_loss=6.536, ppl=92.82, acc=32.06, wps=14494, ups=3.1, wpb=4655, bsz=394, num_updates=7100, lr=0.000419591, gnorm=1.207, clip=100%, oom=0, wall=2311
| epoch 002: 2000 / 6099 loss=7.664, nll_loss=6.391, ppl=83.90, acc=33.21, wps=14507, ups=3.1, wpb=4668, bsz=393, num_updates=8100, lr=0.000392837, gnorm=1.205, clip=100%, oom=0, wall=2634
| epoch 002: 3000 / 6099 loss=7.551, nll_loss=6.262, ppl=76.76, acc=34.2, wps=14566, ups=3.1, wpb=4687, bsz=394, num_updates=9100, lr=0.000370625, gnorm=1.198, clip=100%, oom=0, wall=2955
| epoch 002: 4000 / 6099 loss=7.459, nll_loss=6.159, ppl=71.44, acc=34.99, wps=14600, ups=3.1, wpb=4686, bsz=394, num_updates=10100, lr=0.000351799, gnorm=1.192, clip=100%, oom=0, wall=3274
| epoch 002: 5000 / 6099 loss=7.383, nll_loss=6.073, ppl=67.30, acc=35.65, wps=14590, ups=3.1, wpb=4685, bsz=394, num_updates=11100, lr=0.000335578, gnorm=1.187, clip=100%, oom=0, wall=3596
| epoch 002: 6000 / 6099 loss=7.317, nll_loss=5.998, ppl=63.92, acc=36.2, wps=14569, ups=3.1, wpb=4680, bsz=394, num_updates=12100, lr=0.000321412, gnorm=1.177, clip=100%, oom=0, wall=3917
| epoch 002 | loss 7.311 | nll_loss 5.992 | ppl 63.63 | acc 36.25 | wps 14564 | ups 3.1 | wpb 4679 | bsz 394 | num_updates 12198 | lr 0.000320118 | gnorm 1.176 | clip 100% | oom 0 | wall 3949
| epoch 002 | valid on 'valid' subset | valid_loss 7.16504 | acc 39.58 | valid_nll_loss 5.67031 | valid_ppl 50.93 | num_updates 12198 | best 7.16504
| epoch 003: 1000 / 6099 loss=6.797, nll_loss=5.409, ppl=42.50, acc=40.34, wps=14602, ups=2.9, wpb=4710, bsz=394, num_updates=13199, lr=0.00030774, gnorm=1.167, clip=100%, oom=0, wall=4290
| epoch 003: 2000 / 6099 loss=6.773, nll_loss=5.382, ppl=41.71, acc=40.57, wps=14562, ups=3.0, wpb=4688, bsz=394, num_updates=14199, lr=0.000296706, gnorm=1.159, clip=100%, oom=0, wall=4611
Killed