Arampacha's workspace
Runs
301
Name
20 visualized
Tags
Group
User
Created
Runtime
State
epoch
train_loss
valid_loss
corpus_bleu
accuracy
bpc
perplexity
model_n_hashes
Learner.lr
batch size
exp
lm
lsh
nhashes
n_hashes
tyoc213
1d 3h 59m 6s
Finished
10
1.0408
1.0109
-
0.70157
1.45843
2.74808
16
0.001
2
exp
lm
lsh
nhashes
n_hashes
morgan
17h 45m 46s
Finished
10
1.0404
1.00436
-
0.70288
1.44899
2.73016
8
0.001
2
exp
lm
lsh
nhashes
n_hashes
arampacha
17d 4h 27m 42s
Finished
-
-
-
-
-
-
-
-
-
-
exp
lm
lsh
nhashes
n_hashes
arampacha
17d 4h 30m 39s
Finished
-
-
-
-
-
-
-
-
-
-
exp
lm
lsh
nhashes
n_hashes
tyoc213
1d 3h 41m 45s
Finished
10
1.04108
1.03256
-
0.6954
1.48967
2.80824
16
0.001
2
exp
lm
lsh
nhashes
n_hashes
arampacha
1h 54m 49s
Finished
0.99927
1.70367
-
-
-
-
-
2
0.001
8
exp
lm
lsh
nhashes
n_hashes
tyoc213
2h 43m 30s
Finished
0.99982
1.44983
-
-
-
-
-
16
0.001
2
exp
lm
lsh
nhashes
n_hashes
tyoc213
8m 51s
Failed
0.99836
2.80251
-
-
-
-
-
16
0.001
2
exp
lm
lsh
nhashes
n_hashes
tyoc213
12s
Failed
-
-
-
-
-
-
-
16
0.001
8
exp
lm
lsh
nhashes
n_hashes
tyoc213
4s
Failed
-
-
-
-
-
-
-
16
0.001
8
exp
lm
lsh
nhashes
n_hashes
tyoc213
6s
Failed
-
-
-
-
-
-
-
16
0.001
8
Notes
Learner.opt_func
batch per epoch
Learner.loss_func._name
Learner.moms
input 1 dim 1
input 1 dim 2
eps_0
raw_loss
model parameters
Learner.metrics
ProgressCallback
Recorder.add_time
Recorder.train_metrics
Recorder.valid_metrics
TrainEvalCallback
dataset.tfms
device
dls.after_batch
dls.after_item
dls.before_batch
n_inp
CombineInputOutputCallback
LossTargetShiftCallback
RemoveEOSCallback
loss
lr_0
mom_0
sqr_mom_0
model_d_model
model_n_heads
model_n_layers
model_vocab_sz
model_pos_enc
model_use_lsh
model_attn_bias
model_causal
model_bucket_size
model_prenorm
model_attn_dropout
model_ff_dropout
model_emb_dropout
model_max_seq_len
model_tie_weights
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
10987
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
2
4096
-
1.16752
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7fd16e8c73d0>","<reformer_fastai.metrics.BPC object at 0x7fd167175e50>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.0010492
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (8)
reformer_fastai.optimizers.adafactor
10987
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
2
4096
-
1.17684
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7fa236c60cd0>","<reformer_fastai.metrics.BPC object at 0x7fa234d0d650>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.0010615
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (2)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
performance as function of n_hashes (4)
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
10987
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
2
4096
-
1.18233
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7fddda9434f0>","<reformer_fastai.metrics.BPC object at 0x7fddd31f0fa0>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.00094551
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (2)
reformer_fastai.optimizers.adafactor
2747
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
8
4096
-
1.6654
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7f726ebb7b70>","<reformer_fastai.metrics.BPC object at 0x7f726b7469e8>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.0005972
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
10987
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
2
4096
-
1.36482
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7f2e56fc7490>","<reformer_fastai.metrics.BPC object at 0x7f2e4f873e50>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.00053297
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
611
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
2
4096
-
2.752
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7f2d907de3d0>","<reformer_fastai.metrics.BPC object at 0x7f2d8908adf0>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
0.00027801
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
2747
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
8
4096
-
-
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7f42a618b190>","<reformer_fastai.metrics.BPC object at 0x7f429ea524c0>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
-
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
153
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
8
4096
-
-
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7fa6462a2d60>","<reformer_fastai.metrics.BPC object at 0x7fa63eb50d30>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
-
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
performance as function of n_hashes (16)
reformer_fastai.optimizers.adafactor
153
FlattenedLoss of CrossEntropyLoss()
[0.95,0.85,0.95]
8
4096
-
-
34958592
["fastai.metrics.accuracy","<fastai.metrics.Perplexity object at 0x7f013a4ddd90>","<reformer_fastai.metrics.BPC object at 0x7f0132d8deb0>"]
true
true
false
true
true
Pipeline: attrgetter -> ByteTextTokenizer
cuda
Pipeline:
Pipeline:
Pipeline:
1
-
-
-
-
-
-
-
1024
8
3
256
axial
true
false
true
64
false
0.1
0.1
0.1
4096
true
1-11
of 11