diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/100epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/100epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2ff0ee5449b9b8ee13277d3889f14fa2c370a5a3
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/100epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dd0daaefe2355e6b0151b39c8f53b350fc4f531973a9cff2cb5dfb6fffeaddf
+size 398012507
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/64epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/64epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5f08e8016873aaa53a1c44268b4a6fe4c9409647
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/64epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab1a15b413af7c42ce18c259b8fb17d33196e4195a9e2100ecd498a0626ef41a
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/72epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/72epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..cc3ce4ce56744f08093fd4679cc8af81f08bbd74
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/72epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ecbaecace577d8cafd1663294454dc1a1f1e26e2367df846e34411c28514d79
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/74epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/74epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..84107a48250a8fa9ebf441d284094820eb58ec03
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/74epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e219419dd596a9af9ca66d2c07f7313798312122a37a9a3ffd8e931e32b89117
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/81epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/81epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3eb3cadfcdac74338abfa16e6c246a6e644a844f
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/81epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b377bef5802f2e9c3f869171cf63538eb13e2783838e159caaa363396b058288
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/86epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/86epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6aef642321b69c3ed3fd08d29928d027f4ece775
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/86epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cac0e8ed3cb48a764075095dace05f187f8fc060c991231a115b2883ae35bd76
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/92epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/92epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..724e3640a98c2717ba3807dd5441eb72e4d45dac
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/92epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f18d055595cfb7665eca64e5079be6c0e7ed2d06871df4544f43dc9b81abcac6
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/95epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/95epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..7eac26493b8a80fd68d19dcd2e113786d2dd6a48
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/95epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88bb8537bd4761cf34777676b35a826f89ebefe561683f88ad9f7f31c453d310
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/98epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/98epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..105516f7078e49162528fdbceb3372386cede621
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/98epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:323a4273ec9025bdbd076a7cb947c176d25d86ad8dd91c7922833adfcb7e55c0
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/99epoch.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/99epoch.pth
new file mode 100644
index 0000000000000000000000000000000000000000..c92d4a2b5f64720e3eaffec4760ef311358381ae
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/99epoch.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b060bf5418cbd9020192dbed4490a99a803e47389627057a78d497138b8aa75d
+size 398011377
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/RESULTS.md b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/RESULTS.md
new file mode 100644
index 0000000000000000000000000000000000000000..78e316e2325a9ccefdba7d1616e88d7bdd9bae99
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/RESULTS.md
@@ -0,0 +1,38 @@
+<!-- Generated by scripts/utils/show_asr_result.sh -->
+# RESULTS
+## Environments
+- date: `Thu Jul 18 18:36:12 HKT 2024`
+- python version: `3.8.19 (default, Mar 20 2024, 19:58:24)  [GCC 11.2.0]`
+- espnet version: `espnet 202402`
+- pytorch version: `pytorch 1.13.1+cu117`
+- Git hash: `19787b1793eda2b4007aa5b2c4d03adf6c18abfb`
+  - Commit date: `Fri Jun 14 19:27:35 2024 +0900`
+
+## exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7
+### WER
+
+|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
+|---|---|---|---|---|---|---|---|---|
+|decode_asr_asr_model_valid.acc.ave/dev_clean|2703|54402|97.2|2.5|0.3|0.3|3.1|34.9|
+|decode_asr_asr_model_valid.acc.ave/dev_other|2864|50948|92.6|6.7|0.7|0.9|8.3|56.5|
+|decode_asr_asr_model_valid.acc.ave/test_clean|2620|52576|97.0|2.6|0.4|0.4|3.4|36.1|
+|decode_asr_asr_model_valid.acc.ave/test_other|2939|52343|93.0|6.3|0.7|1.0|7.9|57.4|
+
+### CER
+
+|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
+|---|---|---|---|---|---|---|---|---|
+|decode_asr_asr_model_valid.acc.ave/dev_clean|2703|288456|99.2|0.4|0.4|0.3|1.2|34.9|
+|decode_asr_asr_model_valid.acc.ave/dev_other|2864|265951|97.1|1.7|1.2|1.0|3.8|56.5|
+|decode_asr_asr_model_valid.acc.ave/test_clean|2620|281530|99.0|0.4|0.5|0.4|1.3|36.1|
+|decode_asr_asr_model_valid.acc.ave/test_other|2939|272758|97.5|1.5|1.0|1.0|3.5|57.4|
+
+### TER
+
+|dataset|Snt|Wrd|Corr|Sub|Del|Ins|Err|S.Err|
+|---|---|---|---|---|---|---|---|---|
+|decode_asr_asr_model_valid.acc.ave/dev_clean|2703|68010|96.7|2.4|0.9|0.5|3.9|34.9|
+|decode_asr_asr_model_valid.acc.ave/dev_other|2864|63110|91.3|6.5|2.3|1.4|10.1|56.5|
+|decode_asr_asr_model_valid.acc.ave/test_clean|2620|65818|96.3|2.5|1.2|0.5|4.2|36.1|
+|decode_asr_asr_model_valid.acc.ave/test_other|2939|65101|91.7|5.9|2.3|1.2|9.4|57.4|
+
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/checkpoint.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/checkpoint.pth
new file mode 100644
index 0000000000000000000000000000000000000000..aec20db344efbe12caa2fd93f9fbe47fc27ca4b6
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/checkpoint.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:105b0b5d675fbdb31e2909bf6e275e154fb42673d91052d0a1e331e1e00940a1
+size 1193162349
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..257bc9c27ef69d30b7c5d45ac4b7f4dac1de11ef
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml
@@ -0,0 +1,5211 @@
+config: conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml
+print_config: false
+log_level: INFO
+drop_last_iter: false
+dry_run: false
+iterator_type: sequence
+valid_iterator_type: null
+output_dir: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7
+ngpu: 1
+seed: 0
+num_workers: 1
+num_att_plot: 3
+dist_backend: nccl
+dist_init_method: env://
+dist_world_size: 4
+dist_rank: 0
+local_rank: 0
+dist_master_addr: localhost
+dist_master_port: 38893
+dist_launcher: null
+multiprocessing_distributed: true
+unused_parameters: false
+sharded_ddp: false
+cudnn_enabled: true
+cudnn_benchmark: false
+cudnn_deterministic: true
+collect_stats: false
+write_collected_feats: false
+max_epoch: 100
+patience: null
+val_scheduler_criterion:
+- valid
+- loss
+early_stopping_criterion:
+- valid
+- loss
+- min
+best_model_criterion:
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 10
+nbest_averaging_interval: 0
+grad_clip: 5.0
+grad_clip_type: 2.0
+grad_noise: false
+accum_grad: 2
+no_forward_run: false
+resume: true
+train_dtype: float32
+use_amp: true
+log_interval: null
+use_matplotlib: true
+use_tensorboard: true
+create_graph_in_tensorboard: false
+use_wandb: false
+wandb_project: null
+wandb_id: null
+wandb_entity: null
+wandb_name: null
+wandb_model_log_interval: -1
+detect_anomaly: false
+use_adapter: false
+adapter: lora
+save_strategy: all
+adapter_conf: {}
+pretrain_path: null
+init_param: []
+ignore_init_mismatch: false
+freeze_param: []
+num_iters_per_epoch: null
+batch_size: 20
+valid_batch_size: null
+batch_bins: 45000000
+valid_batch_bins: null
+train_shape_file:
+- exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape
+- exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe
+valid_shape_file:
+- exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape
+- exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe
+batch_type: numel
+valid_batch_type: null
+fold_length:
+- 80000
+- 150
+sort_in_batch: descending
+shuffle_within_batch: false
+sort_batch: descending
+multiple_iterator: false
+chunk_length: 500
+chunk_shift_ratio: 0.5
+num_cache_chunks: 1024
+chunk_excluded_key_prefixes: []
+chunk_default_fs: null
+chunk_max_abs_length: null
+chunk_discard_short_samples: true
+train_data_path_and_name_and_type:
+-   - dump/raw/train_960_sp/wav.scp
+    - speech
+    - sound
+-   - dump/raw/train_960_sp/text
+    - text
+    - text
+valid_data_path_and_name_and_type:
+-   - dump/raw/dev/wav.scp
+    - speech
+    - sound
+-   - dump/raw/dev/text
+    - text
+    - text
+allow_variable_data_keys: false
+max_cache_size: 0.0
+max_cache_fd: 32
+allow_multi_rates: false
+valid_max_cache_size: null
+exclude_weight_decay: false
+exclude_weight_decay_conf: {}
+optim: adam
+optim_conf:
+    lr: 0.002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+token_list:
+- <blank>
+- <unk>
+- ▁THE
+- S
+- ▁AND
+- ▁OF
+- ▁TO
+- ▁A
+- ▁IN
+- ▁I
+- ▁HE
+- ▁THAT
+- ▁WAS
+- ED
+- ▁IT
+- ''''
+- ▁HIS
+- ING
+- ▁YOU
+- ▁WITH
+- ▁FOR
+- ▁HAD
+- T
+- ▁AS
+- ▁HER
+- ▁IS
+- ▁BE
+- ▁BUT
+- ▁NOT
+- ▁SHE
+- D
+- ▁AT
+- ▁ON
+- LY
+- ▁HIM
+- ▁THEY
+- ▁ALL
+- ▁HAVE
+- ▁BY
+- ▁SO
+- ▁THIS
+- ▁MY
+- ▁WHICH
+- ▁ME
+- ▁SAID
+- ▁FROM
+- ▁ONE
+- Y
+- E
+- ▁WERE
+- ▁WE
+- ▁NO
+- N
+- ▁THERE
+- ▁OR
+- ER
+- ▁AN
+- ▁WHEN
+- ▁ARE
+- ▁THEIR
+- ▁WOULD
+- ▁IF
+- ▁WHAT
+- ▁THEM
+- ▁WHO
+- ▁OUT
+- M
+- ▁DO
+- ▁WILL
+- ▁UP
+- ▁BEEN
+- P
+- R
+- ▁MAN
+- ▁THEN
+- ▁COULD
+- ▁MORE
+- C
+- ▁INTO
+- ▁NOW
+- ▁VERY
+- ▁YOUR
+- ▁SOME
+- ▁LITTLE
+- ES
+- ▁TIME
+- RE
+- ▁CAN
+- ▁LIKE
+- LL
+- ▁ABOUT
+- ▁HAS
+- ▁THAN
+- ▁DID
+- ▁UPON
+- ▁OVER
+- IN
+- ▁ANY
+- ▁WELL
+- ▁ONLY
+- B
+- ▁SEE
+- ▁GOOD
+- ▁OTHER
+- ▁TWO
+- L
+- ▁KNOW
+- ▁GO
+- ▁DOWN
+- ▁BEFORE
+- A
+- AL
+- ▁OUR
+- ▁OLD
+- ▁SHOULD
+- ▁MADE
+- ▁AFTER
+- ▁GREAT
+- ▁DAY
+- ▁MUST
+- ▁COME
+- ▁HOW
+- ▁SUCH
+- ▁CAME
+- LE
+- ▁WHERE
+- ▁US
+- ▁NEVER
+- ▁THESE
+- ▁MUCH
+- ▁DE
+- ▁MISTER
+- ▁WAY
+- G
+- ▁S
+- ▁MAY
+- ATION
+- ▁LONG
+- OR
+- ▁AM
+- ▁FIRST
+- ▁BACK
+- ▁OWN
+- ▁RE
+- ▁AGAIN
+- ▁SAY
+- ▁MEN
+- ▁WENT
+- ▁HIMSELF
+- ▁HERE
+- NESS
+- ▁THINK
+- V
+- IC
+- ▁EVEN
+- ▁THOUGHT
+- ▁HAND
+- ▁JUST
+- ▁O
+- ▁UN
+- VE
+- ION
+- ▁ITS
+- 'ON'
+- ▁MAKE
+- ▁MIGHT
+- ▁TOO
+- K
+- ▁AWAY
+- ▁LIFE
+- TH
+- ▁WITHOUT
+- ST
+- ▁THROUGH
+- ▁MOST
+- ▁TAKE
+- ▁DON
+- ▁EVERY
+- F
+- O
+- ▁SHALL
+- ▁THOSE
+- ▁EYES
+- AR
+- ▁STILL
+- ▁LAST
+- ▁HOUSE
+- ▁HEAD
+- ABLE
+- ▁NOTHING
+- ▁NIGHT
+- ITY
+- ▁LET
+- ▁MANY
+- ▁OFF
+- ▁BEING
+- ▁FOUND
+- ▁WHILE
+- EN
+- ▁SAW
+- ▁GET
+- ▁PEOPLE
+- ▁FACE
+- ▁YOUNG
+- CH
+- ▁UNDER
+- ▁ONCE
+- ▁TELL
+- AN
+- ▁THREE
+- ▁PLACE
+- ▁ROOM
+- ▁YET
+- ▁SAME
+- IL
+- US
+- U
+- ▁FATHER
+- ▁RIGHT
+- EL
+- ▁THOUGH
+- ▁ANOTHER
+- LI
+- RI
+- ▁HEART
+- IT
+- ▁PUT
+- ▁TOOK
+- ▁GIVE
+- ▁EVER
+- ▁E
+- ▁PART
+- ▁WORK
+- ERS
+- ▁LOOK
+- ▁NEW
+- ▁KING
+- ▁MISSUS
+- ▁SIR
+- ▁LOVE
+- ▁MIND
+- ▁LOOKED
+- W
+- RY
+- ▁ASKED
+- ▁LEFT
+- ET
+- ▁LIGHT
+- CK
+- ▁DOOR
+- ▁MOMENT
+- RO
+- ▁WORLD
+- ▁THINGS
+- ▁HOME
+- UL
+- ▁THING
+- LA
+- ▁WHY
+- ▁MOTHER
+- ▁ALWAYS
+- ▁FAR
+- FUL
+- ▁WATER
+- CE
+- IVE
+- UR
+- ▁HEARD
+- ▁SOMETHING
+- ▁SEEMED
+- I
+- LO
+- ▁BECAUSE
+- OL
+- ▁END
+- ▁TOLD
+- ▁CON
+- ▁YES
+- ▁GOING
+- ▁GOT
+- RA
+- IR
+- ▁WOMAN
+- ▁GOD
+- EST
+- TED
+- ▁FIND
+- ▁KNEW
+- ▁SOON
+- ▁EACH
+- ▁SIDE
+- H
+- TON
+- MENT
+- ▁OH
+- NE
+- Z
+- LING
+- ▁AGAINST
+- TER
+- ▁NAME
+- ▁MISS
+- ▁QUITE
+- ▁WANT
+- ▁YEARS
+- ▁FEW
+- ▁BETTER
+- ENT
+- ▁HALF
+- ▁DONE
+- ▁ALSO
+- ▁BEGAN
+- ▁HAVING
+- ▁ENOUGH
+- IS
+- ▁LADY
+- ▁WHOLE
+- LESS
+- ▁BOTH
+- ▁SEEN
+- ▁SET
+- ▁WHITE
+- ▁COURSE
+- IES
+- ▁VOICE
+- ▁CALLED
+- ▁D
+- ▁EX
+- ATE
+- ▁TURNED
+- ▁GAVE
+- ▁C
+- ▁POOR
+- MAN
+- UT
+- NA
+- ▁DEAR
+- ISH
+- ▁GIRL
+- ▁MORNING
+- ▁BETWEEN
+- LED
+- ▁NOR
+- IA
+- ▁AMONG
+- MA
+- ▁
+- ▁SMALL
+- ▁REST
+- ▁WHOM
+- ▁FELT
+- ▁HANDS
+- ▁MYSELF
+- ▁HIGH
+- ▁M
+- ▁HOWEVER
+- ▁HERSELF
+- ▁P
+- CO
+- ▁STOOD
+- ID
+- ▁KIND
+- ▁HUNDRED
+- AS
+- ▁ROUND
+- ▁ALMOST
+- TY
+- ▁SINCE
+- ▁G
+- AM
+- ▁LA
+- SE
+- ▁BOY
+- ▁MA
+- ▁PERHAPS
+- ▁WORDS
+- ATED
+- ▁HO
+- X
+- ▁MO
+- ▁SAT
+- ▁REPLIED
+- ▁FOUR
+- ▁ANYTHING
+- ▁TILL
+- ▁UNTIL
+- ▁BLACK
+- TION
+- ▁CRIED
+- RU
+- TE
+- ▁FACT
+- ▁HELP
+- ▁NEXT
+- ▁LOOKING
+- ▁DOES
+- ▁FRIEND
+- ▁LAY
+- ANCE
+- ▁POWER
+- ▁BROUGHT
+- VER
+- ▁FIRE
+- ▁KEEP
+- PO
+- FF
+- ▁COUNTRY
+- ▁SEA
+- ▁WORD
+- ▁CAR
+- ▁DAYS
+- ▁TOGETHER
+- ▁IMP
+- ▁REASON
+- KE
+- ▁INDEED
+- TING
+- ▁MATTER
+- ▁FULL
+- ▁TEN
+- TIC
+- ▁LAND
+- ▁RATHER
+- ▁AIR
+- ▁HOPE
+- ▁DA
+- ▁OPEN
+- ▁FEET
+- ▁EN
+- ▁FIVE
+- ▁POINT
+- ▁CO
+- OM
+- ▁LARGE
+- ▁B
+- ▁CL
+- ME
+- ▁GONE
+- ▁CHILD
+- INE
+- GG
+- ▁BEST
+- ▁DIS
+- UM
+- ▁HARD
+- ▁LORD
+- OUS
+- ▁WIFE
+- ▁SURE
+- ▁FORM
+- DE
+- ▁DEATH
+- ANT
+- ▁NATURE
+- ▁BA
+- ▁CARE
+- ▁BELIEVE
+- PP
+- ▁NEAR
+- ▁RO
+- ▁RED
+- ▁WAR
+- IE
+- ▁SPEAK
+- ▁FEAR
+- ▁CASE
+- ▁TAKEN
+- ▁ALONG
+- ▁CANNOT
+- ▁HEAR
+- ▁THEMSELVES
+- CI
+- ▁PRESENT
+- AD
+- ▁MASTER
+- ▁SON
+- ▁THUS
+- ▁LI
+- ▁LESS
+- ▁SUN
+- ▁TRUE
+- IM
+- IOUS
+- ▁THOUSAND
+- ▁MONEY
+- ▁W
+- ▁BEHIND
+- ▁CHILDREN
+- ▁DOCTOR
+- AC
+- ▁TWENTY
+- ▁WISH
+- ▁SOUND
+- ▁WHOSE
+- ▁LEAVE
+- ▁ANSWERED
+- ▁THOU
+- ▁DUR
+- ▁HA
+- ▁CERTAIN
+- ▁PO
+- ▁PASSED
+- GE
+- TO
+- ▁ARM
+- ▁LO
+- ▁STATE
+- ▁ALONE
+- TA
+- ▁SHOW
+- ▁NEED
+- ▁LIVE
+- ND
+- ▁DEAD
+- ENCE
+- ▁STRONG
+- ▁PRE
+- ▁TI
+- ▁GROUND
+- SH
+- TI
+- ▁SHORT
+- IAN
+- UN
+- ▁PRO
+- ▁HORSE
+- MI
+- ▁PRINCE
+- ARD
+- ▁FELL
+- ▁ORDER
+- ▁CALL
+- AT
+- ▁GIVEN
+- ▁DARK
+- ▁THEREFORE
+- ▁CLOSE
+- ▁BODY
+- ▁OTHERS
+- ▁SENT
+- ▁SECOND
+- ▁OFTEN
+- ▁CA
+- ▁MANNER
+- MO
+- NI
+- ▁BRING
+- ▁QUESTION
+- ▁HOUR
+- ▁BO
+- AGE
+- ▁ST
+- ▁TURN
+- ▁TABLE
+- ▁GENERAL
+- ▁EARTH
+- ▁BED
+- ▁REALLY
+- ▁SIX
+- 'NO'
+- IST
+- ▁BECOME
+- ▁USE
+- ▁READ
+- ▁SE
+- ▁VI
+- ▁COMING
+- ▁EVERYTHING
+- ▁EM
+- ▁ABOVE
+- ▁EVENING
+- ▁BEAUTIFUL
+- ▁FEEL
+- ▁RAN
+- ▁LEAST
+- ▁LAW
+- ▁ALREADY
+- ▁MEAN
+- ▁ROSE
+- WARD
+- ▁ITSELF
+- ▁SOUL
+- ▁SUDDENLY
+- ▁AROUND
+- RED
+- ▁ANSWER
+- ICAL
+- ▁RA
+- ▁WIND
+- ▁FINE
+- ▁WON
+- ▁WHETHER
+- ▁KNOWN
+- BER
+- NG
+- ▁TA
+- ▁CAPTAIN
+- ▁EYE
+- ▁PERSON
+- ▁WOMEN
+- ▁SORT
+- ▁ASK
+- ▁BROTHER
+- ▁USED
+- ▁HELD
+- ▁BIG
+- ▁RETURNED
+- ▁STRANGE
+- ▁BU
+- ▁PER
+- ▁FREE
+- ▁EITHER
+- ▁WITHIN
+- ▁DOUBT
+- ▁YEAR
+- ▁CLEAR
+- ▁SIGHT
+- ▁GRA
+- ▁LOST
+- ▁KEPT
+- ▁F
+- PE
+- ▁BAR
+- ▁TOWN
+- ▁SLEEP
+- ARY
+- ▁HAIR
+- ▁FRIENDS
+- ▁DREAM
+- ▁FELLOW
+- PER
+- ▁DEEP
+- QUE
+- ▁BECAME
+- ▁REAL
+- ▁PAST
+- ▁MAKING
+- RING
+- ▁COMP
+- ▁ACT
+- ▁BAD
+- HO
+- STER
+- ▁YE
+- ▁MEANS
+- ▁RUN
+- MEN
+- ▁DAUGHTER
+- ▁SENSE
+- ▁CITY
+- ▁SOMETIMES
+- ▁TOWARDS
+- ▁ROAD
+- ▁SP
+- ▁LU
+- ▁READY
+- ▁FOOT
+- ▁COLD
+- ▁SA
+- ▁LETTER
+- ▁ELSE
+- ▁MAR
+- ▁STA
+- BE
+- ▁TRUTH
+- ▁LE
+- BO
+- ▁BUSINESS
+- CHE
+- ▁JOHN
+- ▁SUBJECT
+- ▁COURT
+- ▁IDEA
+- ILY
+- ▁RIVER
+- ATING
+- ▁FAMILY
+- HE
+- ▁DIDN
+- ▁GLAD
+- ▁SEVERAL
+- IAL
+- ▁UNDERSTAND
+- ▁SC
+- ▁POSSIBLE
+- ▁DIFFERENT
+- ▁RETURN
+- ▁ARMS
+- ▁LOW
+- ▁HOLD
+- ▁TALK
+- ▁RU
+- ▁WINDOW
+- ▁INTEREST
+- ▁SISTER
+- SON
+- ▁SH
+- ▁BLOOD
+- ▁SAYS
+- ▁CAP
+- ▁DI
+- ▁HUMAN
+- ▁CAUSE
+- NCE
+- ▁THANK
+- ▁LATE
+- GO
+- ▁CUT
+- ▁ACROSS
+- ▁STORY
+- NT
+- ▁COUNT
+- ▁ABLE
+- DY
+- LEY
+- ▁NUMBER
+- ▁STAND
+- ▁CHURCH
+- ▁THY
+- ▁SUPPOSE
+- LES
+- BLE
+- OP
+- ▁EFFECT
+- BY
+- ▁K
+- ▁NA
+- ▁SPOKE
+- ▁MET
+- ▁GREEN
+- ▁HUSBAND
+- ▁RESPECT
+- ▁PA
+- ▁FOLLOWED
+- ▁REMEMBER
+- ▁LONGER
+- ▁AGE
+- ▁TAKING
+- ▁LINE
+- ▁SEEM
+- ▁HAPPY
+- LAND
+- EM
+- ▁STAY
+- ▁PLAY
+- ▁COMMON
+- ▁GA
+- ▁BOOK
+- ▁TIMES
+- ▁OBJECT
+- ▁SEVEN
+- QUI
+- DO
+- UND
+- ▁FL
+- ▁PRETTY
+- ▁FAIR
+- WAY
+- ▁WOOD
+- ▁REACHED
+- ▁APPEARED
+- ▁SWEET
+- ▁FALL
+- BA
+- ▁PASS
+- ▁SIGN
+- ▁TREE
+- IONS
+- ▁GARDEN
+- ▁ILL
+- ▁ART
+- ▁REMAIN
+- ▁OPENED
+- ▁BRIGHT
+- ▁STREET
+- ▁TROUBLE
+- ▁PAIN
+- ▁CONTINUED
+- ▁SCHOOL
+- OUR
+- ▁CARRIED
+- ▁SAYING
+- HA
+- ▁CHANGE
+- ▁FOLLOW
+- ▁GOLD
+- ▁SW
+- ▁FEELING
+- ▁COMMAND
+- ▁BEAR
+- ▁CERTAINLY
+- ▁BLUE
+- ▁NE
+- CA
+- ▁WILD
+- ▁ACCOUNT
+- ▁OUGHT
+- UD
+- ▁T
+- ▁BREATH
+- ▁WANTED
+- ▁RI
+- ▁HEAVEN
+- ▁PURPOSE
+- ▁CHARACTER
+- ▁RICH
+- ▁PE
+- ▁DRESS
+- OS
+- FA
+- ▁TH
+- ▁ENGLISH
+- ▁CHANCE
+- ▁SHIP
+- ▁VIEW
+- ▁TOWARD
+- AK
+- ▁JOY
+- ▁JA
+- ▁HAR
+- ▁NEITHER
+- ▁FORCE
+- ▁UNCLE
+- DER
+- ▁PLAN
+- ▁PRINCESS
+- DI
+- ▁CHIEF
+- ▁HAT
+- ▁LIVED
+- ▁AB
+- ▁VISIT
+- ▁MOR
+- TEN
+- ▁WALL
+- UC
+- ▁MINE
+- ▁PLEASURE
+- ▁SMILE
+- ▁FRONT
+- ▁HU
+- ▁DEAL
+- OW
+- ▁FURTHER
+- GED
+- ▁TRIED
+- DA
+- VA
+- ▁NONE
+- ▁ENTERED
+- ▁QUEEN
+- ▁PAY
+- ▁EL
+- ▁EXCEPT
+- ▁SHA
+- ▁FORWARD
+- ▁EIGHT
+- ▁ADDED
+- ▁PUBLIC
+- ▁EIGHTEEN
+- ▁STAR
+- ▁HAPPENED
+- ▁LED
+- ▁WALKED
+- ▁ALTHOUGH
+- ▁LATER
+- ▁SPIRIT
+- ▁WALK
+- ▁BIT
+- ▁MEET
+- LIN
+- ▁FI
+- LT
+- ▁MOUTH
+- ▁WAIT
+- ▁HOURS
+- ▁LIVING
+- ▁YOURSELF
+- ▁FAST
+- ▁CHA
+- ▁HALL
+- ▁BEYOND
+- ▁BOAT
+- ▁SECRET
+- ENS
+- ▁CHAIR
+- RN
+- ▁RECEIVED
+- ▁CAT
+- RESS
+- ▁DESIRE
+- ▁GENTLEMAN
+- UGH
+- ▁LAID
+- EVER
+- ▁OCCASION
+- ▁WONDER
+- ▁GU
+- ▁PARTY
+- DEN
+- ▁FISH
+- ▁SEND
+- ▁NEARLY
+- ▁TRY
+- CON
+- ▁SEEMS
+- RS
+- ▁BELL
+- ▁BRA
+- ▁SILENCE
+- IG
+- ▁GUARD
+- ▁DIE
+- ▁DOING
+- ▁TU
+- ▁COR
+- ▁EARLY
+- ▁BANK
+- ▁FIGURE
+- IF
+- ▁ENGLAND
+- ▁MARY
+- ▁AFRAID
+- LER
+- ▁FO
+- ▁WATCH
+- ▁FA
+- ▁VA
+- ▁GRE
+- ▁AUNT
+- PED
+- ▁SERVICE
+- ▁JE
+- ▁PEN
+- ▁MINUTES
+- ▁PAN
+- ▁TREES
+- NED
+- ▁GLASS
+- ▁TONE
+- ▁PLEASE
+- ▁FORTH
+- ▁CROSS
+- ▁EXCLAIMED
+- ▁DREW
+- ▁EAT
+- ▁AH
+- ▁GRAVE
+- ▁CUR
+- PA
+- URE
+- CENT
+- ▁MILES
+- ▁SOFT
+- ▁AGO
+- ▁POSITION
+- ▁WARM
+- ▁LENGTH
+- ▁NECESSARY
+- ▁THINKING
+- ▁PICTURE
+- ▁PI
+- SHIP
+- IBLE
+- ▁HEAVY
+- ▁ATTENTION
+- ▁DOG
+- ABLY
+- ▁STANDING
+- ▁NATURAL
+- ▁APPEAR
+- OV
+- ▁CAUGHT
+- VO
+- ISM
+- ▁SPRING
+- ▁EXPERIENCE
+- ▁PAT
+- OT
+- ▁STOPPED
+- ▁REGARD
+- ▁HARDLY
+- ▁SELF
+- ▁STRENGTH
+- ▁GREW
+- ▁KNIGHT
+- ▁OPINION
+- ▁WIDE
+- ▁INSTEAD
+- ▁SOUTH
+- ▁TRANS
+- ▁CORNER
+- ▁LEARN
+- ▁ISLAND
+- ▁MI
+- ▁THIRD
+- ▁STE
+- ▁STRAIGHT
+- ▁TEA
+- ▁BOUND
+- ▁SEEING
+- ▁JU
+- ▁DINNER
+- ▁BEAUTY
+- ▁PEACE
+- AH
+- ▁REP
+- ▁SILENT
+- ▁CRE
+- ALLY
+- RIC
+- ▁STEP
+- ▁VER
+- ▁JO
+- GER
+- ▁SITTING
+- ▁THIRTY
+- ▁SAVE
+- ENED
+- ▁GLANCE
+- ▁REACH
+- ▁ACTION
+- ▁SAL
+- ▁SAD
+- ▁STONE
+- ITIES
+- ▁FRENCH
+- ▁STRUCK
+- ▁PAPER
+- ▁WHATEVER
+- ▁SUB
+- ▁DISTANCE
+- ▁WRONG
+- ▁KNOWLEDGE
+- ▁SAFE
+- ▁SNOW
+- ▁MUSIC
+- ▁FIFTY
+- RON
+- ▁ATTEMPT
+- ▁GOVERNMENT
+- TU
+- ▁CROWD
+- ▁BESIDES
+- ▁LOVED
+- ▁BOX
+- ▁DIRECTION
+- ▁TRAIN
+- ▁NORTH
+- ▁THICK
+- ▁GETTING
+- AV
+- ▁FLOOR
+- ▁COMPANY
+- ▁BLOW
+- ▁PLAIN
+- TRO
+- ▁BESIDE
+- ▁ROCK
+- ▁IMMEDIATELY
+- FI
+- ▁SHADOW
+- ▁SIT
+- ORS
+- ILE
+- ▁DRINK
+- ▁SPOT
+- ▁DANGER
+- ▁AL
+- ▁SAINT
+- ▁SLOWLY
+- ▁PALACE
+- IER
+- ▁RESULT
+- ▁PETER
+- ▁FOREST
+- ▁BELONG
+- ▁SU
+- ▁PAR
+- RIS
+- ▁TEARS
+- ▁APPEARANCE
+- ▁GATE
+- BU
+- ITION
+- ▁QUICKLY
+- ▁QUIET
+- ▁LONDON
+- ▁START
+- ▁BROWN
+- TRA
+- KIN
+- ▁CONSIDER
+- ▁BATTLE
+- ▁ANNE
+- ▁PIECE
+- ▁DIED
+- ▁SUCCESS
+- ▁LIPS
+- ▁FILLED
+- ▁FORGET
+- ▁POST
+- IFIED
+- ▁MARGARET
+- ▁FOOD
+- HAM
+- ▁PLEASANT
+- ▁FE
+- ▁EXPRESSION
+- ▁POCKET
+- ▁FRESH
+- ▁WEAR
+- TRI
+- ▁BROKEN
+- ▁LAUGHED
+- GING
+- ▁FOLLOWING
+- WN
+- IP
+- ▁TOUCH
+- ▁YOUTH
+- ATIVE
+- ▁LEG
+- ▁WEEK
+- ▁REMAINED
+- ▁EASY
+- NER
+- RK
+- ▁ENTER
+- ▁FIGHT
+- ▁PLACED
+- ▁TRAVEL
+- ▁SIMPLE
+- ▁GIRLS
+- ▁WAITING
+- ▁STOP
+- ▁WAVE
+- AU
+- ▁WISE
+- ▁CAMP
+- TURE
+- UB
+- ▁VE
+- ▁OFFICE
+- ▁GRAND
+- ▁FIT
+- ▁JUDGE
+- UP
+- MENTS
+- ▁QUICK
+- HI
+- ▁FLO
+- RIES
+- VAL
+- ▁COMFORT
+- ▁PARTICULAR
+- ▁STARTED
+- ▁SUIT
+- ▁NI
+- ▁PALE
+- ▁IMPOSSIBLE
+- ▁HOT
+- ▁CONVERSATION
+- ▁SCENE
+- ▁BOYS
+- ▁WIN
+- ▁BRE
+- ▁SOCIETY
+- ▁OUTSIDE
+- ▁WRITE
+- ▁EFFORT
+- ▁TALKING
+- ▁FORTUNE
+- ▁NINE
+- ▁WA
+- ▁SINGLE
+- ▁RULE
+- ▁PORT
+- ▁WINTER
+- ▁CAST
+- ▁CRA
+- ▁HAPPEN
+- ▁CRO
+- ▁SHUT
+- NING
+- ▁GUN
+- ▁NOBLE
+- ▁BEGIN
+- ▁PATH
+- ▁SKY
+- ▁WONDERFUL
+- ▁SUDDEN
+- ▁ARMY
+- ▁CHE
+- ▁WORTH
+- ▁MOUNTAIN
+- ▁MIN
+- AG
+- ▁FLU
+- ▁GRACE
+- ▁CHAPTER
+- ▁BELOW
+- ▁RING
+- ▁TURNING
+- ▁IRON
+- ▁TOP
+- ▁AFTERNOON
+- ORY
+- ▁EVIL
+- ▁TRUST
+- ▁BOW
+- ▁TRI
+- ▁SAIL
+- ▁CONTENT
+- ▁HORSES
+- ITE
+- ▁SILVER
+- AP
+- ▁LAD
+- ▁RUNNING
+- ▁HILL
+- ▁BEGINNING
+- ▁MAD
+- ▁HABIT
+- GRA
+- ▁CLOTHES
+- ▁MORROW
+- ▁CRY
+- ▁FASHION
+- ▁PRESENCE
+- ▁Z
+- FE
+- ▁ARRIVED
+- ▁QUARTER
+- ▁PERFECT
+- ▁WO
+- ▁TRA
+- ▁USUAL
+- ▁NECK
+- ▁MARRIED
+- ▁SEAT
+- ▁WI
+- ▁GAR
+- ▁SAND
+- ▁SHORE
+- ▁GIVING
+- NY
+- ▁PROBABLY
+- ▁MINUTE
+- ▁EXPECT
+- ▁DU
+- ▁SHOT
+- ▁INSTANT
+- ▁DEGREE
+- ▁COLOR
+- ▁WEST
+- RT
+- ▁MARCH
+- ▁BIRD
+- ▁SHOWED
+- ▁GREATER
+- ▁SERIOUS
+- ▁CARRY
+- ▁COVERED
+- ▁FORMER
+- ▁LOUD
+- ▁MOVED
+- ▁MASS
+- ▁SEEK
+- ▁CHO
+- GEN
+- ▁ROMAN
+- IB
+- ▁MOON
+- ▁BOARD
+- ▁STREAM
+- ▁EASILY
+- ▁WISHED
+- ▁SEARCH
+- ▁COULDN
+- ▁MONTHS
+- ▁SICK
+- LIE
+- ▁DUTY
+- ▁TWELVE
+- ▁FAINT
+- ▁STRANGER
+- ▁SURPRISE
+- ▁KILL
+- ▁LEAVING
+- ▁JOURNEY
+- ▁SCARCELY
+- ▁RAISED
+- ▁SPEAKING
+- ▁TERRIBLE
+- ▁TOM
+- ▁FIELD
+- ▁GAME
+- ▁QUA
+- ▁PROMISE
+- ▁LIE
+- ▁CONDITION
+- ▁TRO
+- ▁PERSONAL
+- ▁TALL
+- ▁STICK
+- ▁THREW
+- ▁MARRY
+- ▁VAN
+- ▁BURN
+- ▁ACCORDING
+- ▁RISE
+- ▁ATTACK
+- ▁SWORD
+- ▁GUESS
+- ▁THOUGHTS
+- ▁THIN
+- ▁THROW
+- ▁CALM
+- SIDE
+- ▁VILLAGE
+- ▁DEN
+- ▁ANXIOUS
+- ▁MER
+- GI
+- ▁EXPECTED
+- ▁BALL
+- ▁ESPECIALLY
+- ▁CHARGE
+- ▁MEASURE
+- ISE
+- ▁NICE
+- ▁TRYING
+- ▁ALLOW
+- ▁SHARP
+- ▁BREAD
+- ▁HONOUR
+- ▁HONOR
+- ▁ENTIRELY
+- ▁BILL
+- ▁BRI
+- ▁WRITTEN
+- ▁AR
+- ▁BROKE
+- ▁KILLED
+- ▁MARK
+- ▁VEN
+- ▁LADIES
+- ▁LEARNED
+- ▁FLOWERS
+- PLE
+- ▁FORTY
+- ▁OFFER
+- ▁HAPPINESS
+- ▁PRAY
+- ▁CLASS
+- ▁FER
+- ▁PRINCIPLE
+- GU
+- ▁BOOKS
+- ▁SHAPE
+- ▁SUMMER
+- ▁JACK
+- ▁DRAW
+- ▁GOLDEN
+- ▁DECIDED
+- ▁LEAD
+- ▁UNLESS
+- ▁HARM
+- ▁LISTEN
+- HER
+- ▁SHOOK
+- ▁INFLUENCE
+- ▁PERFECTLY
+- ▁MARRIAGE
+- ▁BROAD
+- ▁ESCAPE
+- ▁STATES
+- ▁MIDDLE
+- ▁PLANT
+- ▁MIL
+- ▁MOVEMENT
+- ▁NOISE
+- ▁ENEMY
+- ▁HISTORY
+- ▁BREAK
+- ROUS
+- ▁UNDERSTOOD
+- ▁LATTER
+- FER
+- ▁COMES
+- ▁MERELY
+- ▁SIMPLY
+- WI
+- ▁IMAGINE
+- ▁LOWER
+- ▁CONDUCT
+- ▁BORN
+- WA
+- ▁YARD
+- ▁KA
+- ▁CLOSED
+- ▁NOTE
+- GA
+- ▁STRA
+- RAN
+- ▁EXIST
+- EV
+- ▁SPEECH
+- ▁BITTER
+- JO
+- ▁MAKES
+- ▁GRASS
+- ▁REPLY
+- ▁CHANGED
+- ▁MON
+- ▁LYING
+- ▁DANCE
+- ▁FINALLY
+- ▁AMERICAN
+- ▁ENJOY
+- ▁CONTAIN
+- ▁MEANT
+- USE
+- ▁OBSERVED
+- THER
+- ▁LAUGH
+- ▁AFTERWARDS
+- ▁BEAT
+- ▁RACE
+- ▁EQUAL
+- ▁RAIN
+- PS
+- ▁STEPS
+- ▁BENEATH
+- ▁TAIL
+- ▁TASTE
+- IO
+- EY
+- ▁CHAR
+- ▁GE
+- GN
+- TIN
+- ▁GROW
+- ▁TE
+- IANS
+- ▁MOVE
+- ▁REPEATED
+- ▁DRIVE
+- TUR
+- ▁SI
+- CLOCK
+- ▁BRAVE
+- ▁MADAME
+- ▁LOT
+- ▁CASTLE
+- ▁HI
+- AND
+- ▁FUTURE
+- ▁RELATION
+- ▁SORRY
+- ▁HEALTH
+- ▁DICK
+- ▁R
+- ▁BUILDING
+- ▁EDGE
+- ▁BLESS
+- ▁SPITE
+- WE
+- ▁MIS
+- ▁PRISONER
+- ▁ALLOWED
+- ▁PH
+- ▁CATCH
+- MER
+- ETH
+- ▁COAT
+- ▁COMPLETE
+- ▁WOULDN
+- ▁CREATURE
+- ▁YELLOW
+- ▁IMPORTANT
+- ▁ADD
+- ▁PASSING
+- ▁DARKNESS
+- ▁CARRIAGE
+- ▁MILL
+- ▁FIFTEEN
+- NCY
+- ▁HUNG
+- ▁OB
+- ▁PLEASED
+- ▁SPREAD
+- ▁CURIOUS
+- ▁WORSE
+- ▁CIRCUMSTANCES
+- ▁GI
+- LAR
+- ▁CAL
+- ▁HY
+- ▁MERE
+- ▁JANE
+- ▁EAST
+- BI
+- ▁CUP
+- ▁BLIND
+- ▁PASSION
+- ▁DISCOVERED
+- ▁NOTICE
+- ▁REPORT
+- ▁SPACE
+- ▁PRESENTLY
+- ▁SORROW
+- ▁PACK
+- ▁DIN
+- CY
+- ▁DRY
+- ▁ANCIENT
+- ▁DRESSED
+- ▁COVER
+- ▁VO
+- ▁EXISTENCE
+- ▁EXACTLY
+- ▁BEAST
+- ▁PROPER
+- ▁DROPPED
+- ▁CLEAN
+- ▁COLOUR
+- ▁HOST
+- ▁CHAMBER
+- ▁FAITH
+- LET
+- ▁DETERMINED
+- ▁PRIEST
+- ▁STORM
+- ▁SKIN
+- ▁DARE
+- ▁PERSONS
+- ▁PICK
+- ▁NARROW
+- ▁SUPPORT
+- ▁PRIVATE
+- ▁SMILED
+- ▁COUSIN
+- ▁DRAWING
+- ▁ATTEND
+- ▁COOK
+- ▁PREVENT
+- ▁VARIOUS
+- ▁BLA
+- ▁FIXED
+- ▁WEAK
+- THE
+- ▁HOLE
+- ▁BOTTOM
+- ▁NOBODY
+- ADE
+- ▁LEGS
+- ITCH
+- ▁INDIVIDUAL
+- ▁EARS
+- LIKE
+- ▁ADVANTAGE
+- ▁FRANCE
+- ▁BON
+- ▁WINE
+- ▁LIVES
+- OD
+- ▁WALLS
+- ▁TIRED
+- ▁SHOP
+- ▁ANIMAL
+- ▁CRU
+- ▁WROTE
+- ▁ROYAL
+- ▁CONSIDERED
+- ▁MORAL
+- ▁COMPANION
+- ▁LOSE
+- ▁ISN
+- ▁BAG
+- ▁LAKE
+- ▁INTER
+- ▁COM
+- ▁LETTERS
+- ▁LUCK
+- ▁EAR
+- ▁GERMAN
+- ▁PET
+- ▁SAKE
+- ▁DROP
+- ▁PAID
+- ▁BREAKFAST
+- ▁LABOR
+- ▁DESERT
+- ▁DECLARED
+- ▁HUM
+- ▁STUDY
+- ▁INSTANCE
+- ONE
+- ▁SOMEWHAT
+- ▁CLOTH
+- ▁SPECIAL
+- ▁COLONEL
+- ▁SONG
+- ▁MAIN
+- ▁VALUE
+- ▁PROUD
+- ▁EXPRESS
+- ▁NATION
+- ▁HANDSOME
+- ▁CONFESS
+- ▁PU
+- ▁PASSAGE
+- ▁PERIOD
+- ▁CUSTOM
+- ▁HURT
+- ▁SHOULDER
+- ▁CHRIST
+- ZA
+- ▁RECEIVE
+- ▁DIFFICULT
+- ▁DEPEND
+- ▁MEETING
+- ▁CHI
+- ▁GEN
+- LIGHT
+- ▁BELIEVED
+- ▁SOCIAL
+- ▁DIFFICULTY
+- ▁GREATEST
+- ▁DRAWN
+- ▁GRANT
+- ▁BIRDS
+- ▁ANGRY
+- ▁HEAT
+- UFF
+- ▁DUE
+- ▁PLACES
+- ▁SIN
+- ▁COURAGE
+- ▁EVIDENTLY
+- ▁GENTLE
+- ▁CRUEL
+- ▁GEORGE
+- ▁GRI
+- ▁SERVANT
+- ▁U
+- ▁PURE
+- OOK
+- ▁KNOWS
+- ▁KNOWING
+- LF
+- ▁WRITING
+- ▁REMEMBERED
+- ▁CU
+- ▁HOLDING
+- ▁TENDER
+- ▁QUI
+- ▁BURST
+- ▁SURELY
+- IGN
+- ▁VALLEY
+- ▁FU
+- ▁BUTTER
+- ▁SPOKEN
+- ▁STORE
+- ▁DISC
+- ▁CHRISTIAN
+- ▁PARIS
+- ▁HENRY
+- ▁FINISHED
+- ▁PROVE
+- ▁FOOL
+- ▁SOLDIERS
+- ▁LANGUAGE
+- ▁INSIDE
+- ▁BAN
+- ▁FALLEN
+- ROW
+- ▁MAL
+- ▁BABY
+- ▁SITUATION
+- ▁WATCHED
+- ANS
+- ▁RUIN
+- ▁GENTLEMEN
+- ▁FRO
+- ▁FANCY
+- ▁ACCEPT
+- ▁SEASON
+- ▁OURSELVES
+- ▁SAN
+- ▁SPEED
+- IZED
+- ▁COOL
+- ▁SERVE
+- ▁VESSEL
+- ▁WILLIAM
+- ▁OBLIGED
+- ▁GROUP
+- FORM
+- ▁GOES
+- UOUS
+- ▁LEAVES
+- ▁PECULIAR
+- ▁NEWS
+- ▁VAIN
+- ▁EVERYBODY
+- ▁PIN
+- UG
+- ▁FORGOTTEN
+- ▁FRA
+- GAN
+- ▁CAREFULLY
+- ▁FLASH
+- UCH
+- ▁FUR
+- ▁MURDER
+- ▁DELIGHT
+- ▁WAITED
+- ▁RENDER
+- ▁PROPERTY
+- ▁NOTICED
+- ▁ROLL
+- ▁KNOCK
+- ▁EARNEST
+- KI
+- ▁HONEST
+- ▁PROMISED
+- ▁BAL
+- AW
+- ▁WALKING
+- ANG
+- ▁SQUARE
+- ▁QUIETLY
+- ▁CLOUD
+- WOOD
+- ▁FORMED
+- ▁HIGHER
+- ▁BUILT
+- ▁FATE
+- ▁TEACH
+- MY
+- ▁FALSE
+- ▁YORK
+- ▁DUST
+- ▁CLIMB
+- ▁FOND
+- ▁GROWN
+- ▁DESCEND
+- ▁RAG
+- ▁FRUIT
+- ▁GENERALLY
+- ▁OFFERED
+- ▁ER
+- ▁NURSE
+- POSE
+- ▁SPENT
+- ▁JOIN
+- ▁STATION
+- ▁MEANING
+- ▁SMOKE
+- HOOD
+- ▁ROUGH
+- JU
+- ▁LIKELY
+- ▁SURFACE
+- ▁KE
+- ▁MONTH
+- ▁POSSESSION
+- ▁TONGUE
+- ▁DUKE
+- ▁NOSE
+- ▁LAUGHING
+- ▁WEATHER
+- ▁WHISPERED
+- ▁SYSTEM
+- ▁LAWS
+- DDLE
+- ▁TOUCHED
+- ▁TRADE
+- LD
+- ▁SURPRISED
+- RIN
+- ▁ARCH
+- ▁WEALTH
+- FOR
+- ▁TEMPER
+- ▁FRANK
+- ▁GAL
+- ▁BARE
+- ▁OPPORTUNITY
+- ▁CLAIM
+- ▁ANIMALS
+- ▁REV
+- ▁COST
+- ▁WASH
+- ZE
+- ▁CORN
+- ▁OPPOSITE
+- ▁POLICE
+- ▁IDEAS
+- LON
+- ▁KEY
+- ▁READING
+- ▁COLLECT
+- CHED
+- ▁H
+- ▁CROWN
+- ▁TAR
+- ▁SWIFT
+- ▁SHOULDERS
+- ▁ICE
+- ▁GRAY
+- ▁SHARE
+- ▁PREPARED
+- ▁GRO
+- ▁UND
+- ▁TER
+- ▁EMPTY
+- CING
+- ▁SMILING
+- ▁AVOID
+- ▁DIFFERENCE
+- ▁EXPLAIN
+- ▁POUR
+- ▁ATTRACT
+- ▁OPENING
+- ▁WHEEL
+- ▁MATERIAL
+- ▁BREAST
+- ▁SUFFERING
+- ▁DISTINCT
+- ▁BOOT
+- ▁ROW
+- ▁FINGERS
+- HAN
+- ▁ALTOGETHER
+- ▁FAT
+- ▁PAPA
+- ▁BRAIN
+- ▁ASLEEP
+- ▁GREY
+- ▁SUM
+- ▁GAS
+- ▁WINDOWS
+- ▁ALIVE
+- ▁PROCEED
+- ▁FLOWER
+- ▁LEAP
+- ▁PUR
+- ▁PIECES
+- ▁ALTER
+- ▁MEMORY
+- IENT
+- ▁FILL
+- ▁CLO
+- ▁THROWN
+- ▁KINGDOM
+- ▁RODE
+- IUS
+- ▁MAID
+- ▁DIM
+- ▁BAND
+- ▁VIRTUE
+- ▁DISH
+- ▁GUEST
+- ▁LOSS
+- ▁CAUSED
+- ▁MOTION
+- ▁POT
+- ▁MILLION
+- ▁FAULT
+- ▁LOVELY
+- ▁HERO
+- PPING
+- ▁UNITED
+- ▁SPI
+- SOME
+- BRA
+- ▁MOUNTAINS
+- ▁NU
+- ▁SATISFIED
+- ▁DOLLARS
+- ▁LOVER
+- ▁CONCEAL
+- ▁VAST
+- ▁PULL
+- ▁HATH
+- ▁RUSH
+- ▁J
+- ▁DESPAIR
+- EX
+- ▁HEIGHT
+- ▁CE
+- ▁BENT
+- ▁PITY
+- ▁RISING
+- ATH
+- ▁PRIDE
+- ▁HURRY
+- KA
+- ▁SETTLED
+- ▁JUSTICE
+- ▁LIFTED
+- PEN
+- ▁SOLDIER
+- ▁FINDING
+- ▁REMARK
+- ▁REGULAR
+- ▁STRUGGLE
+- ▁MACHINE
+- ▁SING
+- ▁HURRIED
+- ▁SUFFICIENT
+- ▁REPRESENT
+- ▁DOUBLE
+- ▁ALARM
+- ▁SUPPER
+- ▁DREADFUL
+- ▁FORE
+- ATOR
+- ▁STOCK
+- ▁TIN
+- ▁EXAMPLE
+- ▁ROOF
+- ▁FLOW
+- ▁SUPPOSED
+- ▁PRESERV
+- ▁L
+- ▁LISTENED
+- OC
+- ▁STO
+- ▁SECURE
+- ▁FRIGHTENED
+- ▁DISTURB
+- ▁EMOTION
+- ▁SERVANTS
+- ▁YO
+- ▁BUY
+- ▁FORCED
+- ▁KITCHEN
+- ▁TERROR
+- ▁STAIRS
+- ▁SIXTY
+- KER
+- ▁ORDINARY
+- ▁DIRECTLY
+- ▁HEADS
+- ▁METHOD
+- ▁FORGIVE
+- ▁AWFUL
+- ▁REFLECT
+- ▁GREATLY
+- ▁TALKED
+- ▁RIDE
+- STONE
+- ▁FAVOUR
+- ▁WELCOME
+- ▁SEIZED
+- OU
+- ▁CONTROL
+- ▁ORDERED
+- ▁ANGEL
+- ▁USUALLY
+- ▁POET
+- ▁BOLD
+- LINE
+- ▁ADVENTURE
+- ▁WATCHING
+- ▁FOLK
+- ▁MISTRESS
+- IZE
+- ▁GROWING
+- ▁CAVE
+- ▁EVIDENCE
+- ▁FINGER
+- ▁SEVENTEEN
+- ▁MOVING
+- EOUS
+- ▁DOESN
+- ▁COW
+- ▁TYPE
+- ▁BOIL
+- ▁TALE
+- ▁DELIVER
+- ▁FARM
+- ▁MONSIEUR
+- ▁GATHERED
+- ▁FEELINGS
+- ▁RATE
+- ▁REMARKED
+- ▁PUTTING
+- ▁MAT
+- ▁CONTRARY
+- ▁CRIME
+- ▁PLA
+- ▁COL
+- ▁NEARER
+- TES
+- ▁CIVIL
+- ▁SHAME
+- ▁LOOSE
+- ▁DISCOVER
+- ▁FLAT
+- ▁TWICE
+- ▁FAIL
+- VIS
+- ▁UNC
+- EA
+- ▁EUROPE
+- ▁PATIENT
+- ▁UNTO
+- ▁SUFFER
+- ▁PAIR
+- ▁TREASURE
+- OSE
+- ▁EAGER
+- ▁FLY
+- ▁N
+- ▁VAL
+- ▁DAN
+- ▁SALT
+- ▁BORE
+- BBE
+- ▁ARTHUR
+- ▁AFFAIRS
+- ▁SLOW
+- ▁CONSIST
+- ▁DEVIL
+- LAN
+- ▁AFFECTION
+- ▁ENGAGED
+- ▁KISS
+- ▁YA
+- ▁OFFICER
+- IFICATION
+- ▁LAMP
+- ▁PARTS
+- HEN
+- ▁MILK
+- ▁PROCESS
+- ▁GIFT
+- ▁PULLED
+- ▁HID
+- ▁RAY
+- ▁EXCELLENT
+- ▁IMPRESSION
+- ▁AUTHORITY
+- ▁PROVED
+- ▁TELLING
+- TTE
+- ▁TOWER
+- ▁CONSEQUENCE
+- ▁FAVOR
+- ▁FLEW
+- ▁CHARLES
+- ISTS
+- ▁ADDRESS
+- ▁FAMILIAR
+- ▁LIMIT
+- ▁CONFIDENCE
+- ▁RARE
+- ▁WEEKS
+- ▁WOODS
+- ▁INTENTION
+- ▁DIRECT
+- ▁PERFORM
+- ▁SOLEMN
+- ▁DISTANT
+- ▁IMAGE
+- ▁PRESIDENT
+- ▁FIRM
+- ▁INDIAN
+- ▁RANK
+- ▁LIKED
+- ▁AGREE
+- ▁HOUSES
+- ▁WIL
+- ▁MATTERS
+- ▁PRISON
+- ▁MODE
+- ▁MAJOR
+- ▁WORKING
+- ▁SLIP
+- ▁WEIGHT
+- ▁AWARE
+- ▁BUSY
+- ▁LOOKS
+- ▁WOUND
+- ▁THOR
+- ▁BATH
+- ▁EXERCISE
+- ▁SIMILAR
+- ▁WORE
+- ▁AMOUNT
+- ▁QUESTIONS
+- ▁VIOLENT
+- ▁EXCUSE
+- ▁ASIDE
+- ▁TUR
+- ▁DULL
+- OF
+- ▁EMPEROR
+- ▁NEVERTHELESS
+- ▁SHOUT
+- ▁EXPLAINED
+- ▁SIZE
+- ▁ACCOMPLISH
+- FORD
+- CAN
+- ▁MISTAKE
+- ▁INSTANTLY
+- ▁SMOOTH
+- ▁STRIKE
+- ▁BOB
+- ISED
+- ▁HORROR
+- ▁SCIENCE
+- ▁PROTEST
+- ▁MANAGE
+- ▁OBEY
+- ▁NECESSITY
+- ▁SPLENDID
+- ▁PRESS
+- ▁INTERESTING
+- ▁RELIGION
+- ▁UNKNOWN
+- ▁FIERCE
+- ▁DISAPPEARED
+- ▁HOLY
+- ▁HATE
+- ▁PLAYED
+- ▁LIN
+- ▁NATURALLY
+- ▁DROVE
+- ▁LOUIS
+- TIES
+- ▁BRAND
+- INESS
+- RIE
+- ▁SHOOT
+- ▁CONSENT
+- ▁SEATED
+- ▁LINES
+- GUE
+- ▁AGREED
+- ▁CIRCLE
+- ▁STIR
+- ▁STREETS
+- ▁TASK
+- ▁RID
+- ▁PRODUCED
+- ▁ACCIDENT
+- ▁WITNESS
+- ▁LIBERTY
+- ▁DETAIL
+- ▁MINISTER
+- ▁POWERFUL
+- ▁SAVAGE
+- ▁SIXTEEN
+- ▁PRETEND
+- ▁COAST
+- ▁SQU
+- ▁UTTER
+- ▁NAMED
+- ▁CLEVER
+- ▁ADMIT
+- ▁COUPLE
+- ▁WICKED
+- ▁MESSAGE
+- ▁TEMPLE
+- ▁STONES
+- ▁YESTERDAY
+- ▁HILLS
+- DAY
+- ▁SLIGHT
+- ▁DIAMOND
+- ▁POSSIBLY
+- ▁AFFAIR
+- ▁ORIGINAL
+- ▁HEARING
+- ▁WORTHY
+- ▁SELL
+- NEY
+- ICK
+- ▁COTTAGE
+- ▁SACRIFICE
+- ▁PROGRESS
+- ▁SHOCK
+- ▁DESIGN
+- ▁SOUGHT
+- ▁PIT
+- ▁SUNDAY
+- ▁OTHERWISE
+- ▁CABIN
+- ▁PRAYER
+- ▁DWELL
+- ▁GAIN
+- ▁BRIDGE
+- ▁PARTICULARLY
+- ▁YIELD
+- ▁TREAT
+- RIGHT
+- ▁OAK
+- ▁ROPE
+- WIN
+- ▁ORDERS
+- ▁SUSPECT
+- ▁EDWARD
+- AB
+- ▁ELEVEN
+- ▁TEETH
+- ▁OCCURRED
+- DDING
+- ▁AMERICA
+- ▁FALLING
+- ▁LION
+- ▁DEPART
+- ▁KEEPING
+- ▁DEMAND
+- ▁PAUSED
+- ▁CEASED
+- INA
+- ▁FUN
+- ▁CHEER
+- ▁PARDON
+- ▁NATIVE
+- LUS
+- LOW
+- ▁DOGS
+- ▁REQUIRED
+- ILITY
+- ▁ELECT
+- ▁ENTERTAIN
+- ITUDE
+- ▁HUGE
+- ▁CARRYING
+- ▁BLU
+- ▁INSIST
+- ▁SATISFACTION
+- ▁HUNT
+- ▁COUNTENANCE
+- ▁UPPER
+- ▁MAIDEN
+- ▁FAILED
+- ▁JAMES
+- ▁FOREIGN
+- ▁GATHER
+- ▁TEST
+- BOARD
+- ▁TERMS
+- ▁SILK
+- ▁BEG
+- ▁BROTHERS
+- ▁PAGE
+- ▁KNEES
+- ▁SHOWN
+- ▁PROFESSOR
+- ▁MIGHTY
+- ▁DEFI
+- ▁CHARM
+- ▁REQUIRE
+- ▁LOG
+- MORE
+- ▁PROOF
+- ▁POSSESSED
+- ▁SOFTLY
+- ▁UNFORTUNATE
+- ▁PRICE
+- ▁SEVERE
+- ▁SINGING
+- ▁STAGE
+- ▁FREEDOM
+- ▁SHOUTED
+- ▁FARTHER
+- ▁MAJESTY
+- ▁PREVIOUS
+- ▁GUIDE
+- ▁MATCH
+- ▁CHEST
+- ▁INTENDED
+- ▁BI
+- ▁EXCITEMENT
+- ▁OFFICERS
+- ▁SUR
+- ▁SHAKE
+- ▁SENTIMENT
+- ▁GENTLY
+- ▁SUCCEEDED
+- ▁MENTION
+- ▁LOCK
+- ▁ACQUAINTANCE
+- ▁IMAGINATION
+- ▁PHYSICAL
+- ▁LEADING
+- ▁SLAVE
+- ▁CART
+- ▁POINTED
+- ▁STEAM
+- ▁SHADE
+- ▁PIPE
+- ▁BASE
+- ▁INVENT
+- ▁ALAS
+- ▁WORKED
+- ▁REGRET
+- ▁BUR
+- ▁FAITHFUL
+- ▁MENTIONED
+- ▁RECORD
+- ▁COMPLAIN
+- ▁SUPERIOR
+- ▁BAY
+- ▁PAL
+- EMENT
+- UE
+- ▁SEVENTY
+- ▁HOTEL
+- ▁SHEEP
+- ▁MEAL
+- ▁ADVICE
+- ▁HIDDEN
+- ▁DEMANDED
+- ▁CONSCIOUS
+- ▁BROW
+- ▁POSSESS
+- ▁FOURTH
+- ▁EVENTS
+- ▁FRI
+- ▁PRAISE
+- ▁ADVANCED
+- ▁RESOLVED
+- ▁STUFF
+- ▁CHEERFUL
+- ▁BIRTH
+- ▁GRIEF
+- ▁AFFORD
+- ▁FAIRY
+- ▁WAKE
+- ▁SIDES
+- ▁SUBSTANCE
+- ▁ARTICLE
+- ▁LEVEL
+- ▁MIST
+- ▁JOINED
+- ▁PRACTICAL
+- ▁CLEARLY
+- ▁TRACE
+- ▁AWAKE
+- ▁OBSERVE
+- ▁BASKET
+- ▁LACK
+- VILLE
+- ▁SPIRITS
+- ▁EXCITED
+- ▁ABANDON
+- ▁SHINING
+- ▁FULLY
+- ▁CALLING
+- ▁CONSIDERABLE
+- ▁SPRANG
+- ▁MILE
+- ▁DOZEN
+- ▁PEA
+- ▁DANGEROUS
+- ▁WIT
+- ▁JEW
+- ▁POUNDS
+- ▁FOX
+- ▁INFORMATION
+- ▁LIES
+- ▁DECK
+- NNY
+- ▁PAUL
+- ▁STARS
+- ▁ANGER
+- ▁SETTLE
+- ▁WILLING
+- ▁ADAM
+- ▁FACES
+- ▁SMITH
+- ▁IMPORTANCE
+- ▁STRAIN
+- WAR
+- ▁SAM
+- ▁FEATHER
+- ▁SERVED
+- ▁AUTHOR
+- ▁PERCEIVED
+- ▁FLAME
+- ▁DIVINE
+- ▁TRAIL
+- ▁ANYBODY
+- ▁SIGH
+- ▁DELICATE
+- KY
+- ▁FOLD
+- ▁HAVEN
+- ▁DESIRED
+- ▁CURIOSITY
+- ▁PRACTICE
+- ▁CONSIDERATION
+- ▁ABSOLUTELY
+- ▁CITIZEN
+- ▁BOTTLE
+- ▁INTERESTED
+- ▁MEAT
+- ▁OCCUPIED
+- ▁CHOOSE
+- ▁THROAT
+- ETTE
+- ▁CANDLE
+- ▁DAWN
+- ▁PROTECT
+- ▁SENTENCE
+- IED
+- ▁ROCKS
+- ▁PORTION
+- ▁APPARENTLY
+- ▁PRESENTED
+- ▁TIGHT
+- ▁ACTUALLY
+- ▁DYING
+- ▁HAM
+- ▁DAILY
+- ▁SUFFERED
+- ▁POLITICAL
+- ▁BODIES
+- ▁MODERN
+- ▁COMPLETELY
+- ▁SOONER
+- TAN
+- ▁PROP
+- ▁ADVANCE
+- ▁REFUSED
+- ▁FARMER
+- ▁POLITE
+- ▁THUNDER
+- ▁BRIEF
+- ▁ELSIE
+- ▁SAILOR
+- ▁SUGGESTED
+- ▁PLATE
+- ▁AID
+- ▁FLESH
+- ▁WEEP
+- ▁BUCK
+- ▁ANTI
+- ▁OCEAN
+- ▁SPEND
+- WELL
+- ▁ODD
+- ▁GOVERNOR
+- ▁ENTRANCE
+- ▁SUSPICION
+- ▁STEPPED
+- ▁RAPIDLY
+- ▁CHECK
+- ▁HIDE
+- ▁FLIGHT
+- ▁CLUB
+- ▁ENTIRE
+- ▁INDIANS
+- ASH
+- ▁CAPITAL
+- ▁MAMMA
+- HAR
+- ▁CORRECT
+- ▁CRACK
+- ▁SENSATION
+- ▁WORST
+- ▁PACE
+- ▁MIDST
+- ▁AUGUST
+- ▁PROPORTION
+- ▁INNOCENT
+- LINESS
+- ▁REGARDED
+- ▁DRIVEN
+- ORD
+- ▁HASTE
+- ▁EDUCATION
+- ▁EMPLOY
+- ▁TRULY
+- ▁INSTRUMENT
+- ▁MAG
+- ▁FRAME
+- ▁FOOLISH
+- ▁TAUGHT
+- ▁HANG
+- ▁ARGUMENT
+- ▁NINETEEN
+- ▁ELDER
+- ▁NAY
+- ▁NEEDED
+- ▁NEIGHBOR
+- ▁INSTRUCT
+- ▁PAPERS
+- ▁REWARD
+- ▁EQUALLY
+- ▁FIELDS
+- ▁DIG
+- HIN
+- ▁CONDITIONS
+- JA
+- ▁SPAR
+- ▁REQUEST
+- ▁WORN
+- ▁REMARKABLE
+- ▁LOAD
+- ▁WORSHIP
+- ▁PARK
+- ▁KI
+- ▁INTERRUPTED
+- ▁SKILL
+- ▁TERM
+- LAC
+- ▁CRITIC
+- ▁DISTRESS
+- ▁BELIEF
+- ▁STERN
+- IGHT
+- ▁TRACK
+- ▁HUNTING
+- ▁JEWEL
+- ▁GRADUALLY
+- ▁GLOW
+- ▁RUSHED
+- ▁MENTAL
+- ▁VISITOR
+- ▁PICKED
+- ▁BEHOLD
+- ▁EXPRESSED
+- ▁RUB
+- ▁SKI
+- ARTAGNAN
+- ▁MOREOVER
+- ▁OPERATION
+- ▁CAREFUL
+- ▁KEEN
+- ▁ASSERT
+- ▁WANDER
+- ▁ENEMIES
+- ▁MYSTERIOUS
+- ▁DEPTH
+- ▁PREFER
+- ▁CROSSED
+- ▁CHARMING
+- ▁DREAD
+- ▁FLOUR
+- ▁ROBIN
+- ▁TRE
+- ▁RELIEF
+- ▁INQUIRED
+- ▁APPLE
+- ▁HENCE
+- ▁WINGS
+- ▁CHOICE
+- ▁JUD
+- OO
+- ▁SPECIES
+- ▁DELIGHTED
+- IUM
+- ▁RAPID
+- ▁APPEAL
+- ▁FAMOUS
+- ▁USEFUL
+- ▁HELEN
+- ▁NEWSPAPER
+- ▁PLENTY
+- ▁BEARING
+- ▁NERVOUS
+- ▁PARA
+- ▁URGE
+- ▁ROAR
+- ▁WOUNDED
+- ▁CHAIN
+- ▁PRODUCE
+- ▁REFLECTION
+- ▁MERCHANT
+- ▁QUARREL
+- ▁GLORY
+- ▁BEGUN
+- ▁BARON
+- CUS
+- ▁QUEER
+- ▁MIX
+- ▁GAZE
+- ▁WHISPER
+- ▁BURIED
+- ▁DIV
+- ▁CARD
+- ▁FREQUENTLY
+- ▁TIP
+- ▁KNEE
+- ▁REGION
+- ▁ROOT
+- ▁LEST
+- ▁JEALOUS
+- CTOR
+- ▁SAVED
+- ▁ASKING
+- ▁TRIP
+- QUA
+- ▁UNION
+- HY
+- ▁COMPANIONS
+- ▁SHIPS
+- ▁HALE
+- ▁APPROACHED
+- ▁HARRY
+- ▁DRUNK
+- ▁ARRIVAL
+- ▁SLEPT
+- ▁FURNISH
+- HEAD
+- ▁PIG
+- ▁ABSENCE
+- ▁PHIL
+- ▁HEAP
+- ▁SHOES
+- ▁CONSCIOUSNESS
+- ▁KINDLY
+- ▁EVIDENT
+- ▁SCAR
+- ▁DETERMIN
+- ▁GRASP
+- ▁STEAL
+- ▁OWE
+- ▁KNIFE
+- ▁PRECIOUS
+- ▁ELEMENT
+- ▁PROCEEDED
+- ▁FEVER
+- ▁LEADER
+- ▁RISK
+- ▁EASE
+- ▁GRIM
+- ▁MOUNT
+- ▁MEANWHILE
+- ▁CENTURY
+- OON
+- ▁JUDGMENT
+- ▁AROSE
+- ▁VISION
+- ▁SPARE
+- ▁EXTREME
+- ▁CONSTANT
+- ▁OBSERVATION
+- ▁THRUST
+- ▁DELAY
+- ▁CENT
+- ▁INCLUD
+- ▁LIFT
+- ▁ADMIRE
+- ▁ISSUE
+- ▁FRIENDSHIP
+- ▁LESSON
+- ▁PRINCIPAL
+- ▁MOURN
+- ▁ACCEPTED
+- ▁BURNING
+- ▁CAPABLE
+- ▁EXTRAORDINARY
+- ▁SANG
+- ▁REMOVED
+- ▁HOPED
+- ▁HORN
+- ▁ALICE
+- ▁MUD
+- ▁APARTMENT
+- ▁FIGHTING
+- ▁BLAME
+- ▁TREMBLING
+- ▁SOMEBODY
+- ▁ANYONE
+- ▁BRIDE
+- ▁READER
+- ▁ROB
+- ▁EVERYWHERE
+- ▁LABOUR
+- ▁RECALL
+- ▁BULL
+- ▁HIT
+- ▁COUNCIL
+- ▁POPULAR
+- ▁CHAP
+- ▁TRIAL
+- ▁DUN
+- ▁WISHES
+- ▁BRILLIANT
+- ▁ASSURED
+- ▁FORGOT
+- ▁CONTINUE
+- ▁ACKNOWLEDG
+- ▁RETREAT
+- ▁INCREASED
+- ▁CONTEMPT
+- ▁GRANDFATHER
+- ▁SYMPATHY
+- ▁GHOST
+- ▁STRETCHED
+- ▁CREATURES
+- ▁CAB
+- ▁HIND
+- ▁PLAYING
+- ▁MISERABLE
+- ▁MEMBERS
+- ▁KINDNESS
+- ▁HIGHEST
+- ▁PRIM
+- ▁KISSED
+- ▁DESERVE
+- ▁HUT
+- ▁BEGGED
+- ▁EIGHTY
+- ▁CLOSELY
+- ▁WONDERED
+- ▁MILITARY
+- ▁REMIND
+- ▁ACCORDINGLY
+- ▁LARGER
+- ▁MAINTAIN
+- ▁ENGINE
+- ▁MOTIVE
+- ▁DESTROY
+- ▁STRIP
+- ▁HANS
+- ▁AHEAD
+- ▁INFINITE
+- ▁PROMPT
+- ▁INFORMED
+- TTLE
+- ▁PEER
+- ▁PRESSED
+- ▁TRAP
+- ▁SOMEWHERE
+- ▁BOUGHT
+- ▁VISIBLE
+- ▁ASHAMED
+- ▁TEAR
+- ▁NEIGHBOUR
+- ▁CONSTITUTION
+- ▁INTELLIGENCE
+- ▁PROFESSION
+- ▁HUNGRY
+- RIDGE
+- ▁SMELL
+- ▁STORIES
+- ▁LISTENING
+- ▁APPROACH
+- ▁STRING
+- ▁EXPLANATION
+- ▁IMMENSE
+- ▁RELIGIOUS
+- ▁THROUGHOUT
+- ▁HOLLOW
+- ▁AWAIT
+- ▁FLYING
+- ▁SCREAM
+- ▁ACTIVE
+- ▁RUM
+- ▁PRODUCT
+- ▁UNHAPPY
+- ▁VAGUE
+- ARIES
+- ▁ELIZABETH
+- ▁STUPID
+- ▁DIGNITY
+- ▁ISABEL
+- GAR
+- ▁BRO
+- ▁PITCH
+- ▁COMRADE
+- ▁STIFF
+- ▁RECKON
+- ▁SOLD
+- ▁SPARK
+- ▁STRO
+- ▁CRYING
+- ▁MAGIC
+- ▁REPEAT
+- PORT
+- ▁MARKED
+- ▁COMFORTABLE
+- ▁PROJECT
+- ▁BECOMING
+- ▁PARENTS
+- ▁SHELTER
+- ▁STOLE
+- ▁HINT
+- ▁NEST
+- ▁TRICK
+- ▁THOROUGHLY
+- ▁HOSPITAL
+- ▁WEAPON
+- ▁ROME
+- ▁STYLE
+- ▁ADMITTED
+- ▁SAFETY
+- FIELD
+- ▁UNDERSTANDING
+- ▁TREMBLE
+- ▁PRINT
+- ▁SLAVES
+- ▁WEARY
+- ▁ARTIST
+- ▁CREDIT
+- BURG
+- ▁CONCLUSION
+- ▁SELDOM
+- ▁UNUSUAL
+- ▁CLOUDS
+- ▁UNABLE
+- ▁GAY
+- ▁HANGING
+- ▁SCR
+- ▁BOWED
+- ▁DAVID
+- ▁VOL
+- ▁PUSHED
+- ▁ESCAPED
+- MOND
+- ▁WARN
+- ▁BETRAY
+- ▁EGGS
+- ▁PLAINLY
+- ▁EXHIBIT
+- ▁DISPLAY
+- ▁MEMBER
+- ▁GRIN
+- ▁PROSPECT
+- ▁BRUSH
+- ▁BID
+- ▁SUCCESSFUL
+- ▁EXTENT
+- ▁PERSUADE
+- ▁MID
+- ▁MOOD
+- ▁ARRANGED
+- ▁UNIVERSAL
+- ▁JIM
+- ▁SIGNAL
+- ▁WHILST
+- ▁PHILIP
+- ▁WOLF
+- RATE
+- ▁EAGERLY
+- ▁BILLY
+- ▁RETURNING
+- ▁CONSCIENCE
+- ▁FORTUNATE
+- ▁FEMALE
+- ▁GLEAM
+- ▁HASTILY
+- ▁PROVIDED
+- ▁OBTAIN
+- ▁INSTINCT
+- ▁CONCERNED
+- ▁CONCERNING
+- ▁SOMEHOW
+- ▁PINK
+- ▁RAGE
+- ▁ACCUSTOMED
+- ▁UNCONSCIOUS
+- ▁ADVISE
+- ▁BRANCHES
+- ▁TINY
+- ▁REFUSE
+- ▁BISHOP
+- ▁SUPPLY
+- ▁PEASANT
+- ▁LAWYER
+- ▁WASTE
+- ▁CONNECTION
+- ▁DEVELOP
+- ▁CORRESPOND
+- ▁PLUM
+- ▁NODDED
+- ▁SLIPPED
+- ▁EU
+- ▁CONSTANTLY
+- CUM
+- MMED
+- ▁FAIRLY
+- HOUSE
+- ▁KIT
+- ▁RANG
+- ▁FEATURES
+- ▁PAUSE
+- ▁PAINFUL
+- ▁JOE
+- ▁WHENCE
+- ▁LAUGHTER
+- ▁COACH
+- ▁CHRISTMAS
+- ▁EATING
+- ▁WHOLLY
+- ▁APART
+- ▁SUPER
+- ▁REVOLUTION
+- ▁LONELY
+- ▁CHEEKS
+- ▁THRONE
+- ▁CREW
+- ▁ATTAIN
+- ▁ESTABLISHED
+- TIME
+- ▁DASH
+- ▁FRIENDLY
+- ▁OPERA
+- ▁EARL
+- ▁EXHAUST
+- ▁CLIFF
+- ▁REVEAL
+- ▁ADOPT
+- ▁CENTRE
+- ▁MERRY
+- ▁SYLVIA
+- ▁IDEAL
+- ▁MISFORTUNE
+- ▁FEAST
+- ▁ARAB
+- ▁NUT
+- ▁FETCH
+- ▁FOUGHT
+- ▁PILE
+- ▁SETTING
+- ▁SOURCE
+- ▁PERSIST
+- ▁MERCY
+- ▁BARK
+- ▁LUC
+- ▁DEEPLY
+- ▁COMPARE
+- ▁ATTITUDE
+- ▁ENDURE
+- ▁DELIGHTFUL
+- ▁BEARD
+- ▁PATIENCE
+- ▁LOCAL
+- ▁UTTERED
+- ▁VICTORY
+- ▁TREATED
+- ▁SEPARATE
+- ▁WAG
+- ▁DRAGG
+- ▁TITLE
+- ▁TROOPS
+- ▁TRIUMPH
+- ▁REAR
+- ▁GAINED
+- ▁SINK
+- ▁DEFEND
+- ▁TIED
+- ▁FLED
+- ▁DARED
+- ▁INCREASE
+- ▁POND
+- ▁CONQUER
+- ▁FOREHEAD
+- ▁FAN
+- ▁ANXIETY
+- ▁ENCOUNTER
+- ▁SEX
+- ▁HALT
+- ▁SANK
+- ▁CHEEK
+- ▁HUMBLE
+- ▁WRITER
+- ▁EMPLOYED
+- ▁DISTINGUISHED
+- ▁RAISE
+- ▁WHIP
+- ▁GIANT
+- ▁RANGE
+- ▁OBTAINED
+- ▁FLAG
+- ▁MAC
+- ▁JUMPED
+- ▁DISCOVERY
+- ▁NATIONAL
+- ▁COMMISSION
+- ▁POSITIVE
+- ▁LOVING
+- ▁EXACT
+- ▁MURMURED
+- ▁GAZED
+- ▁REFER
+- ▁COLLEGE
+- ▁ENCOURAGE
+- ▁NOVEL
+- ▁CLOCK
+- ▁MORTAL
+- ▁ROLLED
+- ▁RAT
+- IZING
+- ▁GUILTY
+- ▁VICTOR
+- WORTH
+- ▁PRA
+- ▁APPROACHING
+- ▁RELATIVE
+- ▁ESTATE
+- ▁UGLY
+- ▁METAL
+- ▁ROBERT
+- ▁TENT
+- ▁ADMIRATION
+- ▁FOURTEEN
+- ▁BARBAR
+- ▁WITCH
+- ELLA
+- ▁CAKE
+- ▁SHONE
+- ▁MANAGED
+- ▁VOLUME
+- ▁GREEK
+- ▁DANCING
+- ▁WRETCHED
+- ▁CONDEMN
+- ▁MAGNIFICENT
+- ▁CONSULT
+- J
+- ▁ORGAN
+- ▁FLEET
+- ▁ARRANGEMENT
+- ▁INCIDENT
+- ▁MISERY
+- ▁ARROW
+- ▁STROKE
+- ▁ASSIST
+- ▁BUILD
+- ▁SUCCEED
+- ▁DESPERATE
+- ▁WIDOW
+- UDE
+- ▁MARKET
+- ▁WISDOM
+- ▁PRECISE
+- ▁CURRENT
+- ▁SPOIL
+- ▁BADE
+- ▁WOODEN
+- ▁RESIST
+- ▁OBVIOUS
+- ▁SENSIBLE
+- FALL
+- ▁ADDRESSED
+- ▁GIL
+- ▁COUNSEL
+- ▁PURCHASE
+- ▁SELECT
+- ▁USELESS
+- ▁STARED
+- ▁ARREST
+- ▁POISON
+- ▁FIN
+- ▁SWALLOW
+- ▁BLOCK
+- ▁SLID
+- ▁NINETY
+- ▁SPORT
+- ▁PROVIDE
+- ▁ANNA
+- ▁LAMB
+- ▁INTERVAL
+- ▁JUMP
+- ▁DESCRIBED
+- ▁STRIKING
+- ▁PROVISION
+- ▁PROPOSED
+- ▁MELANCHOLY
+- ▁WARRIOR
+- ▁SUGGEST
+- ▁DEPARTURE
+- ▁BURDEN
+- ▁LIMB
+- ▁TROUBLED
+- ▁MEADOW
+- ▁SACRED
+- ▁SOLID
+- ▁TRU
+- ▁LUCY
+- ▁RECOVER
+- ▁ENERGY
+- ▁POWDER
+- ▁RESUMED
+- ▁INTENSE
+- ▁BRITISH
+- ▁STRAW
+- ▁AGREEABLE
+- ▁EVERYONE
+- ▁CONCERN
+- ▁VOYAGE
+- ▁SOUTHERN
+- ▁BOSOM
+- ▁UTTERLY
+- ▁FEED
+- ▁ESSENTIAL
+- ▁CONFINE
+- ▁HOUSEHOLD
+- ▁EXTREMELY
+- ▁WONDERING
+- ▁LIST
+- ▁PINE
+- PHA
+- ▁EXPERIMENT
+- ▁JOSEPH
+- ▁MYSTERY
+- ▁RESTORE
+- ▁BLUSH
+- FOLD
+- ▁CHOSEN
+- ▁INTELLECT
+- ▁CURTAIN
+- OLOGY
+- ▁MOUNTED
+- ▁LAP
+- ▁EPI
+- ▁PUNISH
+- ▁WEDDING
+- ▁RECOGNIZED
+- ▁DRIFT
+- ▁PREPARATION
+- ▁RESOLUTION
+- ▁OPPRESS
+- ▁FIX
+- ▁VICTIM
+- OGRAPH
+- ▁SUMMON
+- ▁JULIA
+- ▁FLOOD
+- ▁WAL
+- ULATION
+- ▁SLIGHTLY
+- ▁LODGE
+- ▁WIRE
+- ▁CONFUSION
+- ▁UNEXPECTED
+- ▁CONCEIVE
+- ▁PRIZE
+- ▁JESUS
+- ▁ADDITION
+- ▁RUDE
+- ▁FATAL
+- ▁CARELESS
+- ▁PATCH
+- ▁KO
+- ▁CATHERINE
+- ▁PARLIAMENT
+- ▁PROFOUND
+- ▁ALOUD
+- ▁RELIEVE
+- ▁PUSH
+- ABILITY
+- ▁ACCOMPANIED
+- ▁SOVEREIGN
+- ▁SINGULAR
+- ▁ECHO
+- ▁COMPOSED
+- ▁SHAKING
+- ATORY
+- ▁ASSISTANCE
+- ▁TEACHER
+- ▁HORRIBLE
+- ▁STRICT
+- ▁VERSE
+- ▁PUNISHMENT
+- ▁GOWN
+- ▁MISTAKEN
+- ▁VARI
+- ▁SWEPT
+- ▁GESTURE
+- ▁BUSH
+- ▁STEEL
+- ▁AFFECTED
+- ▁DIRECTED
+- ▁SURROUNDED
+- ▁ABSURD
+- ▁SUGAR
+- ▁SCRAP
+- ▁IMMEDIATE
+- ▁SADDLE
+- ▁TY
+- ▁ARISE
+- ▁SIGHED
+- ▁EXCHANGE
+- ▁IMPATIENT
+- ▁SNAP
+- ▁EMBRACE
+- ▁DISEASE
+- ▁PROFIT
+- ▁RIDING
+- ▁RECOVERED
+- ▁GOVERN
+- ▁STRETCH
+- ▁CONVINCED
+- ▁LEANING
+- ▁DOMESTIC
+- ▁COMPLEX
+- ▁MANIFEST
+- ▁INDULGE
+- ▁GENIUS
+- ▁AGENT
+- ▁VEIL
+- ▁DESCRIPTION
+- ▁INCLINED
+- ▁DECEIVE
+- ▁DARLING
+- ▁REIGN
+- HU
+- ▁ENORMOUS
+- ▁RESTRAIN
+- ▁DUTIES
+- BURY
+- TTERED
+- ▁POLE
+- ▁ENABLE
+- ▁EXCEPTION
+- ▁INTIMATE
+- ▁COUNTESS
+- ▁TRIBE
+- ▁HANDKERCHIEF
+- ▁MIDNIGHT
+- ▁PROBLEM
+- ▁TRAMP
+- ▁OIL
+- CAST
+- ▁CRUSH
+- ▁DISCUSS
+- ▁RAM
+- ▁TROT
+- ▁UNRE
+- ▁WHIRL
+- ▁LOCKED
+- ▁HORIZON
+- ▁OFFICIAL
+- ▁SCHEME
+- ▁DROWN
+- ▁PIERRE
+- ▁PERMITTED
+- ▁CONNECTED
+- ▁ASSURE
+- ▁COCK
+- ▁UTMOST
+- ▁DEVOTED
+- ▁RELI
+- ▁SUFFICIENTLY
+- ▁INTELLECTUAL
+- ▁CARPET
+- ▁OBJECTION
+- ▁AFTERWARD
+- ▁REALITY
+- ▁NEGRO
+- ▁RETAIN
+- ▁ASCEND
+- ▁CEASE
+- ▁KATE
+- ▁MARVEL
+- KO
+- ▁BOND
+- MOST
+- ▁COAL
+- GATE
+- ▁IGNORANT
+- ▁BREAKING
+- ▁TWIN
+- ▁ASTONISHMENT
+- ▁COFFEE
+- ▁JAR
+- ▁CITIES
+- ▁ORIGIN
+- ▁EXECUT
+- ▁FINAL
+- ▁INHABITANTS
+- ▁STABLE
+- ▁CHIN
+- ▁PARTIES
+- ▁PLUNGE
+- ▁GENEROUS
+- ▁DESCRIBE
+- ▁ANNOUNCED
+- ▁MERIT
+- ▁REVERE
+- ▁ERE
+- ACIOUS
+- ZI
+- ▁DISAPPOINT
+- ▁SUGGESTION
+- ▁DOUBTLESS
+- ▁TRUNK
+- ▁STAMP
+- ▁JOB
+- ▁APPOINTED
+- ▁DIVIDED
+- ▁ACQUAINTED
+- CHI
+- ▁ABSOLUTE
+- ▁FEARFUL
+- ▁PRIVILEGE
+- ▁CRAFT
+- ▁STEEP
+- ▁HUNTER
+- ▁FORBID
+- ▁MODEST
+- ▁ENDEAVOUR
+- ▁SWEEP
+- ▁BEHELD
+- ▁ABSORB
+- ▁CONSTRUCT
+- ▁EMPIRE
+- ▁EXPEDITION
+- ▁ERECT
+- ▁OFFEND
+- ▁INTEND
+- ▁PERMIT
+- ▁DESTROYED
+- ▁CONTRACT
+- ▁THIRST
+- ▁WAGON
+- ▁EVA
+- ▁GLOOM
+- ▁ATMOSPHERE
+- ▁RESERVE
+- ▁VOTE
+- ▁GER
+- ▁NONSENSE
+- ▁PREVAIL
+- ▁QUALITY
+- ▁CLASP
+- ▁CONCLUDED
+- ▁RAP
+- ▁KATY
+- ▁ETERNAL
+- ▁MUTTERED
+- ▁NEGLECT
+- ▁SQUIRE
+- ▁CREEP
+- LOCK
+- ▁ELECTRIC
+- ▁HAY
+- ▁EXPENSE
+- ▁SCORN
+- ▁RETIRED
+- ▁STOUT
+- ▁MURMUR
+- ▁SHARPLY
+- ▁DISTRICT
+- ▁LEAF
+- ▁FAILURE
+- WICK
+- ▁JEAN
+- ▁NUMEROUS
+- ▁INFANT
+- ▁REALIZED
+- ▁TRAVELLER
+- ▁HUNGER
+- ▁JUNE
+- ▁MUN
+- ▁RECOMMEND
+- ▁CREP
+- ZZLE
+- ▁RICHARD
+- WORK
+- ▁MONTE
+- ▁PREACH
+- ▁PALM
+- AVI
+- ▁ANYWHERE
+- ▁DISPOSITION
+- ▁MIRROR
+- ▁VENTURE
+- ▁POUND
+- ▁CIGAR
+- ▁INVITED
+- ▁BENCH
+- ▁PROTECTION
+- ▁BENEFIT
+- ▁THOMAS
+- ▁CLERK
+- ▁REPROACH
+- ▁UNIFORM
+- ▁GENERATION
+- ▁SEAL
+- ▁COMPASS
+- ▁WARNING
+- ▁EXTENDED
+- ▁DIFFICULTIES
+- ▁MAYBE
+- ▁GROAN
+- ▁AFFECT
+- ▁COMB
+- ▁EARN
+- ▁WESTERN
+- ▁IDLE
+- ▁SCORE
+- ▁TAP
+- ▁ASTONISHED
+- ▁INTRODUCED
+- ▁LEISURE
+- ▁LIEUTENANT
+- ▁VIOLENCE
+- ▁FIRMLY
+- ▁MONSTER
+- ▁UR
+- ▁PROPERLY
+- ▁TWIST
+- ▁PIRATE
+- ▁ROBBER
+- ▁BATTER
+- ▁WEPT
+- ▁LEANED
+- ▁FOG
+- ▁ORNAMENT
+- ▁ANDREW
+- ▁BUSHES
+- ▁REPUBLIC
+- ▁CONFIDENT
+- ▁LEAN
+- ▁DART
+- ▁STOOP
+- ▁CURL
+- ▁COUNTER
+- ▁NORTHERN
+- ▁PEARL
+- ▁NEAREST
+- ▁FRANCIS
+- ▁WANDERING
+- ▁FREQUENT
+- ▁STARTLED
+- ▁STATEMENT
+- ▁OCCUR
+- ▁BLOOM
+- ▁NERVE
+- ▁INSPECT
+- ▁INDUCE
+- ▁FLATTER
+- ▁DATE
+- ▁AMBITION
+- ▁SLOPE
+- ▁MALE
+- ▁MADAM
+- ▁MONK
+- ▁RENT
+- ▁CONFIRM
+- ▁INVESTIGAT
+- ▁RABBIT
+- ▁REGIMENT
+- ▁SUBMIT
+- ▁SPELL
+- ▁FURIOUS
+- ▁RAIL
+- ▁BESTOW
+- ▁RALPH
+- ▁SCATTERED
+- ▁COMPELLED
+- ▁THREAD
+- ▁CHILL
+- ▁DENY
+- ▁PRONOUNC
+- ▁MANKIND
+- ▁CATTLE
+- ▁EXECUTION
+- ▁REBEL
+- ▁SUPREME
+- ▁VALUABLE
+- ▁LIKEWISE
+- ▁CONVEY
+- ▁TIDE
+- ▁GLOOMY
+- ▁COIN
+- ▁ACTUAL
+- ▁TAX
+- ▁PROVINCE
+- ▁GRATEFUL
+- ▁SPIRITUAL
+- ▁VANISHED
+- ▁DIANA
+- ▁HAUNT
+- ▁DRAGON
+- ▁CRAWL
+- ▁CHINA
+- ▁GRATITUDE
+- ▁NEAT
+- ▁FINISH
+- ▁INTENT
+- ▁FRIGHT
+- ▁EMBARRASS
+- ▁THIRTEEN
+- ▁RUTH
+- ▁SLIGHTEST
+- ▁DEVELOPMENT
+- ▁INTERVIEW
+- ▁SPECTACLE
+- ▁BROOK
+- VIE
+- ▁WEAKNESS
+- ▁AUDIENCE
+- ▁CONSEQUENTLY
+- ▁ABROAD
+- ▁ASPECT
+- ▁PAINTED
+- ▁RELEASE
+- ▁INSULT
+- ▁SOOTH
+- ▁DISAPPOINTMENT
+- ▁EMERG
+- ▁BRIG
+- ▁ESTEEM
+- ▁INVITATION
+- ▁PASSENGER
+- ▁PUBLISH
+- ▁PIANO
+- ▁IRISH
+- ▁DESK
+- ▁BEATEN
+- ▁FIFTH
+- ▁IMPULSE
+- ▁SWEAR
+- ▁EATEN
+- ▁PURPLE
+- ▁COMMITTED
+- ▁COUNTRIES
+- ▁PERCEIVE
+- ISON
+- ▁CELEBRAT
+- ▁GRANDMOTHER
+- ▁SHUDDER
+- ▁SUNSHINE
+- ▁SPANISH
+- ▁HITHERTO
+- ▁MARILLA
+- ▁SNAKE
+- ▁MOCK
+- ▁INTERFERE
+- ▁WALTER
+- ▁AMID
+- ▁MARBLE
+- ▁MISSION
+- TERIOR
+- ▁DRIVING
+- ▁FURNITURE
+- ▁STEADY
+- ▁CIRCUMSTANCE
+- ▁INTERPRET
+- ▁ENCHANT
+- ▁ERROR
+- ▁CONVICTION
+- ▁HELPLESS
+- ▁MEDICINE
+- ▁QUALITIES
+- ▁ITALIAN
+- ▁HASTENED
+- ▁OCCASIONALLY
+- ▁PURSUED
+- ▁HESITATED
+- ▁INDEPENDENT
+- ▁OLIVER
+- ▁LINGER
+- UX
+- ▁EXAMINED
+- ▁REPENT
+- ▁PHYSICIAN
+- ▁CHASE
+- ▁BELOVED
+- ▁ATTACHED
+- ▁FLORENCE
+- ▁HONEY
+- ▁MOUSE
+- ▁CRIES
+- ▁BAKE
+- ▁POEM
+- ▁DESTRUCTION
+- ▁FULFIL
+- ▁MESSENGER
+- ▁TRISTRAM
+- ▁FANCIED
+- ▁EXCESS
+- ▁CURSE
+- ▁CHU
+- ▁QUANTITY
+- ▁THORNTON
+- ▁CREATED
+- ▁CONTINUALLY
+- ▁LIGHTNING
+- ▁BORNE
+- ▁TOTAL
+- ▁DISPOSED
+- ▁RIFLE
+- ▁POLLY
+- ▁GOAT
+- ▁BACKWARD
+- ▁VIRGINIA
+- ▁KICK
+- ▁PERIL
+- ▁QUO
+- ▁GLORIOUS
+- ▁MULTITUDE
+- ▁LEATHER
+- ▁ABSENT
+- ▁DEMON
+- ▁DEBT
+- ▁TORTURE
+- ▁ACCORD
+- ▁MATE
+- ▁CATHOLIC
+- ▁PILL
+- ▁LIBRARY
+- ▁PURSUIT
+- ▁SHIRT
+- ▁DEAREST
+- ▁COLLAR
+- ▁BEACH
+- ▁ROBE
+- ▁DECLARE
+- ▁BRANCH
+- ▁TEMPT
+- ▁STEADILY
+- ▁DISGUST
+- ▁SILLY
+- ▁ARRIVE
+- ▁DRANK
+- ▁LEVI
+- ▁COMMUNICAT
+- ▁RACHEL
+- ▁WASHINGTON
+- ▁RESIGN
+- ▁MEANTIME
+- ▁LACE
+- ▁ENGAGEMENT
+- ▁QUIVER
+- ▁SEPARATED
+- ▁DISCUSSION
+- ▁VENTURED
+- ▁SURROUNDING
+- ▁POLISH
+- ▁NAIL
+- ▁SWELL
+- ▁JOKE
+- ▁LINCOLN
+- ▁STUDENT
+- ▁GLITTER
+- ▁RUSSIAN
+- ▁READILY
+- ▁CHRIS
+- ▁POVERTY
+- ▁DISGRACE
+- ▁CHEESE
+- ▁HEAVILY
+- ▁SCALE
+- ▁STAFF
+- ▁ENTREAT
+- ▁FAREWELL
+- ▁LUNCH
+- ▁PEEP
+- ▁MULE
+- ▁SOMEONE
+- ▁DISAPPEAR
+- ▁DECISION
+- ▁PISTOL
+- ▁PUN
+- ▁SPUR
+- ▁ASSUMED
+- ▁EXTEND
+- ▁ENTHUSIASM
+- ▁DEFINITE
+- ▁UNDERTAKE
+- ▁COMMITTEE
+- ▁SIMON
+- ▁FENCE
+- ▁APPLIED
+- ▁RELATED
+- ▁VICE
+- ▁UNPLEASANT
+- ▁PROBABLE
+- ▁PROCURE
+- ▁FROWN
+- ▁CLOAK
+- ▁HUMANITY
+- ▁FAMILIES
+- ▁PHILOSOPHER
+- ▁DWARF
+- ▁OVERCOME
+- ▁DEFEAT
+- ▁FASTENED
+- ▁MARSH
+- ▁CLASSES
+- ▁TOMB
+- ▁GRACIOUS
+- ▁REMOTE
+- ▁CELL
+- ▁SHRIEK
+- ▁RESCUE
+- ▁POOL
+- ▁ORGANIZ
+- ▁CHOSE
+- ▁CUTTING
+- ▁COWARD
+- ▁BORDER
+- ▁DIRTY
+- ▁MONKEY
+- ▁HOOK
+- ▁CHUCK
+- ▁EMILY
+- ▁JEST
+- ▁PLAC
+- ▁WEIGH
+- ▁ASSOCIATE
+- ▁GLIMPSE
+- ▁STUCK
+- ▁BOLT
+- ▁MURDERER
+- ▁PONY
+- ▁DISTINGUISH
+- ▁INSTITUTION
+- ▁CUNNING
+- ▁COMPLIMENT
+- ▁APPETITE
+- ▁REPUTATION
+- ▁FEEBLE
+- ▁KIN
+- ▁SERIES
+- ▁GRACEFUL
+- ▁PLATFORM
+- ▁BREEZE
+- ▁PHRASE
+- ▁CLAY
+- MONT
+- ▁RATTL
+- ▁OPPOSITION
+- ▁LANE
+- ▁BOAST
+- ▁GROWTH
+- ▁INCLINATION
+- ▁BEHAVE
+- ▁SUSAN
+- ▁DISTINCTION
+- ▁DISLIKE
+- ▁NICHOLAS
+- ▁SATISFY
+- ▁DRAMA
+- ▁ELBOW
+- ▁GAZING
+- ▁CONSUM
+- ▁SPIN
+- ▁OATH
+- ▁CHANNEL
+- ▁CHARACTERISTIC
+- ▁SPEAR
+- ▁SLAIN
+- ▁SAUCE
+- ▁FROG
+- ▁CONCEPTION
+- ▁TIMID
+- ▁ZEAL
+- ▁APPARENT
+- SHIRE
+- ▁CENTER
+- ▁VARIETY
+- ▁DUSK
+- ▁APT
+- ▁COLUMN
+- ▁REVENGE
+- ▁RIVAL
+- ▁IMITAT
+- ▁PASSIONATE
+- ▁SELFISH
+- ▁NORMAN
+- ▁REPAIR
+- ▁THRILL
+- ▁TREATMENT
+- ▁ROSA
+- ▁MARTIN
+- ▁INDIFFERENT
+- ▁THITHER
+- ▁GALLANT
+- ▁PEPPER
+- ▁RECOLLECT
+- ▁VINE
+- ▁SCARCE
+- ▁SHIELD
+- ▁MINGLED
+- CLOSE
+- ▁HARSH
+- ▁BRICK
+- ▁HUMOR
+- ▁MISCHIEF
+- ▁TREMENDOUS
+- ▁FUNCTION
+- ▁SMART
+- ▁SULTAN
+- ▁DISMISS
+- ▁THREATENED
+- ▁CHEAP
+- ▁FLOCK
+- ▁ENDEAVOR
+- ▁WHISK
+- ▁ITALY
+- ▁WAIST
+- ▁FLUTTER
+- ▁SMOKING
+- ▁MONARCH
+- ▁AFRICA
+- ▁ACCUSE
+- ▁HERBERT
+- ▁REFRESH
+- ▁REJOICE
+- ▁PILLOW
+- ▁EXPECTATION
+- ▁POETRY
+- ▁HOPELESS
+- ▁PERISH
+- ▁PHILOSOPHY
+- ▁WHISTLE
+- ▁BERNARD
+- ▁LAMENT
+- ▁IMPROVE
+- ▁SUP
+- ▁PERPLEX
+- ▁FOUNTAIN
+- ▁LEAGUE
+- ▁DESPISE
+- ▁IGNORANCE
+- ▁REFERENCE
+- ▁DUCK
+- ▁GROVE
+- ▁PURSE
+- ▁PARTNER
+- ▁PROPHET
+- ▁SHIVER
+- ▁NEIGHBOURHOOD
+- ▁REPRESENTATIVE
+- SAIL
+- ▁WIP
+- ▁ACQUIRED
+- ▁CHIMNEY
+- ▁DOCTRINE
+- ▁MAXIM
+- ▁ANGLE
+- ▁MAJORITY
+- ▁AUTUMN
+- ▁CONFUSED
+- ▁CRISTO
+- ▁ACHIEVE
+- ▁DISGUISE
+- ▁REDUCED
+- ▁EARLIER
+- ▁THEATRE
+- ▁DECIDE
+- MINATED
+- OLOGICAL
+- ▁OCCUPATION
+- ▁VIGOROUS
+- ▁CONTINENT
+- ▁DECLINE
+- ▁COMMUNITY
+- ▁MOTIONLESS
+- ▁HATRED
+- ▁COMMUNICATION
+- ▁BOWL
+- ▁COMMENT
+- ▁APPROVE
+- ▁CEREMONY
+- ▁CRIMINAL
+- ▁SCIENTIFIC
+- ▁DUCHESS
+- ▁VIVID
+- ▁SHIFT
+- ▁AVAIL
+- ▁DAMP
+- ▁JOHNSON
+- ▁SLENDER
+- ▁CONTRAST
+- ▁AMUSEMENT
+- ▁PLOT
+- ▁LYN
+- ▁ASSOCIATION
+- ▁SNATCH
+- ▁UNCERTAIN
+- ▁PRESSURE
+- ▁PERCH
+- ▁APPLY
+- ▁PLANET
+- ▁NOTWITHSTANDING
+- ▁SWUNG
+- ▁STIRRED
+- ▁ATTENDANT
+- ▁ENJOYMENT
+- ▁WORRY
+- ▁ALBERT
+- ▁NAKED
+- ▁TALENT
+- ▁MARIAN
+- ▁REFORM
+- ▁DELIBERATE
+- ▁INTELLIGENT
+- ▁SENSITIVE
+- ▁YONDER
+- ▁PUPIL
+- ▁FRIGHTFUL
+- ▁DOUBTFUL
+- ▁STANDARD
+- ▁MAGISTRATE
+- ▁SHEPHERD
+- ▁STOMACH
+- ▁DEPOSIT
+- ▁RENEW
+- ▁HEDGE
+- ▁FRANCS
+- ▁POSSIBILITY
+- ▁RESEMBLE
+- ▁FATIGUE
+- ▁PORTRAIT
+- ▁FAVORITE
+- ▁CREAM
+- ▁BURG
+- ▁SECRETARY
+- ▁DIVERS
+- ▁ACTIVITY
+- ▁SPECULAT
+- ▁HUMOUR
+- ▁FITTED
+- ▁EXTERNAL
+- ▁CETERA
+- ▁WRAPPED
+- ▁WHIT
+- ▁FRED
+- ▁EXAMINATION
+- ▁LODGING
+- ▁OWING
+- ▁JAW
+- ▁CROW
+- ▁BALANCE
+- ▁PUFF
+- ▁TENDERNESS
+- ▁PORTHOS
+- ▁ANCHOR
+- ▁INTERRUPT
+- ▁NECESSARILY
+- ▁PERPETUAL
+- ▁AGONY
+- ▁POPE
+- ▁SCHOLAR
+- ▁SCOTLAND
+- ▁SUPPRESS
+- ▁WRATH
+- ▁WRECK
+- ▁EXCEED
+- ▁PERFECTION
+- ▁INDIA
+- ▁TRADITION
+- ▁SECTION
+- ▁EASTERN
+- ▁DOORWAY
+- ▁WIVES
+- ▁CONVENTION
+- ▁ANNOUNC
+- ▁EGYPT
+- ▁CONTRADICT
+- ▁SCRATCH
+- ▁CENTRAL
+- ▁GLOVE
+- ▁WAX
+- ▁PREPARE
+- ▁ACCOMPANY
+- ▁INCREASING
+- ▁LIBERAL
+- ▁RAISING
+- ▁ORANGE
+- ▁SHOE
+- ▁ATTRIBUTE
+- ▁LITERATURE
+- ▁PUZZLED
+- ▁WITHDRAW
+- ▁WHITHER
+- ▁HAWK
+- ▁MOONLIGHT
+- ▁EXAMINE
+- ▁HAPPILY
+- ▁PRECEDE
+- ▁DETECTIVE
+- ▁INCHES
+- ▁SOLITARY
+- ▁DUTCH
+- ▁NAPOLEON
+- ▁UNEASY
+- ▁CARDINAL
+- ▁BLEW
+- ▁FOWL
+- ▁DECORAT
+- ▁CHILDHOOD
+- ▁TORMENT
+- ▁LOSING
+- ▁PERMISSION
+- ▁BLANK
+- ▁UPSTAIRS
+- ▁CAPACITY
+- ▁TRIFLE
+- ▁FOLLY
+- ▁RECOGNIZE
+- ▁REMOVE
+- ▁VENGEANCE
+- ▁ENTERPRISE
+- ▁BEDROOM
+- ▁ANYHOW
+- ▁INQUIRY
+- ▁ASHES
+- ▁DRAG
+- ▁HUSH
+- ▁AWKWARD
+- ▁SATURDAY
+- ▁GENUINE
+- ▁SURVIV
+- ▁SKIRT
+- ▁AFFECTIONATE
+- ▁TANG
+- ▁MUTUAL
+- ▁DISPUTE
+- ▁EAGLE
+- ▁INCOME
+- ▁BIND
+- ▁FAME
+- ▁IMPROVEMENT
+- ROVING
+- ▁DIFFER
+- ▁AWOKE
+- ▁SLEEVE
+- ▁SOLITUDE
+- ▁FAVOURITE
+- JI
+- ▁DETECT
+- ▁COMPREHEND
+- ▁PREPARING
+- ▁SERPENT
+- ▁SUMMIT
+- ▁KNOT
+- ▁KNIT
+- ▁COPY
+- ▁STOPPING
+- ▁FADED
+- ▁HIDEOUS
+- ▁JULIE
+- STEAD
+- ▁SHINE
+- ▁CONFLICT
+- ▁PROPOSITION
+- ▁REFUGE
+- ▁GALLERY
+- ▁BUNDLE
+- ▁AXE
+- ▁SLAVERY
+- ▁MASK
+- ▁ALYOSHA
+- ▁LADDER
+- ▁DEPARTMENT
+- ▁DISCHARGE
+- ▁DEPRESS
+- ▁GALLOP
+- ▁SCARLET
+- ▁KITTY
+- ▁RECEIVING
+- ▁SURRENDER
+- ▁SUSTAIN
+- ▁TWILIGHT
+- ▁CONGRESS
+- ▁IRELAND
+- ▁FUNNY
+- ▁LEND
+- ▁CONSTITUTE
+- ▁FUNERAL
+- ▁CRYSTAL
+- ▁SPAIN
+- ▁EXCEEDINGLY
+- ▁DAMN
+- ▁COMMUN
+- ▁CIVILIZATION
+- ▁PREJUDICE
+- ▁PORCH
+- ▁ASSISTANT
+- ▁INDUSTRY
+- ▁TUMBLE
+- ▁DEFENCE
+- ▁HITHER
+- ▁SMOT
+- ▁COLONI
+- ▁AMAZEMENT
+- ▁MARGUERITE
+- ▁MIRACLE
+- ▁INHERIT
+- ▁BEGGAR
+- ▁ENVELOPE
+- ▁INDIGNATION
+- ▁NATASHA
+- ▁PROPOSAL
+- ▁FRAGMENT
+- ▁ROUSED
+- ▁ROAST
+- ENCIES
+- ▁COMMENCED
+- ▁RESOURCE
+- ▁POPULATION
+- ▁QUOTH
+- ▁PURSUE
+- ▁EDUCAT
+- ▁AFFLICT
+- ▁CONTACT
+- ▁CRIMSON
+- ▁DIVISION
+- ▁DISORDER
+- ▁COPPER
+- ▁SOLICIT
+- ▁MODERATE
+- ▁DRUM
+- ▁SWIM
+- ▁SALUTE
+- ▁ASSUME
+- ▁MUSCLE
+- ▁OVERWHELM
+- ▁SHAKESPEARE
+- ▁STRUGGLING
+- ▁TRANQUIL
+- ▁CHICKEN
+- ▁TREAD
+- ▁CLAW
+- ▁BIBLE
+- ▁RIDGE
+- ▁THREAT
+- ▁VELVET
+- ▁EXPOSED
+- ▁IDIOT
+- ▁BARREL
+- ▁PENNY
+- ▁TEMPTATION
+- ▁DANGLARS
+- ▁CENTURIES
+- ▁DISTRIBUT
+- ▁REJECT
+- ▁RETORTED
+- ▁CONCENTRAT
+- ▁CORDIAL
+- ▁MOTOR
+- ▁CANNON
+- KEEP
+- ▁WRETCH
+- ▁ASSURANCE
+- ▁THIEF
+- ▁SURVEY
+- ▁VITAL
+- ▁RAILWAY
+- ▁JACKSON
+- ▁CRASH
+- ▁GROWL
+- ▁COMBAT
+- ▁RECOLLECTION
+- ▁SECURITY
+- ▁JACOB
+- ▁CLUTCH
+- ▁BLANKET
+- ▁NANCY
+- ▁CELLAR
+- ▁CONVENIENT
+- ▁INDIGNANT
+- ▁COARSE
+- ▁WORM
+- ▁SCREEN
+- ▁TRANSPORT
+- ▁BULLET
+- ▁APPRECIATE
+- ▁DEVOTION
+- ▁INVISIBLE
+- ▁DRIED
+- ▁MIXTURE
+- ▁CANDID
+- ▁PERFORMANCE
+- ▁RIPE
+- ▁EXQUISITE
+- ▁BARGAIN
+- ▁TOBACCO
+- ▁LOYAL
+- ▁MOULD
+- ▁ATTENTIVE
+- ▁DOROTHY
+- ▁BRUTE
+- ▁ESTABLISHMENT
+- ▁ABILITY
+- ▁INHABIT
+- ▁OBSCURE
+- ▁BORROW
+- ▁ESSENCE
+- ▁DISMAY
+- ▁FLEE
+- ▁BLADE
+- ▁PLUCK
+- ▁COFFIN
+- ▁SUNSET
+- ▁STEPHEN
+- ▁ECONOMIC
+- ▁HOLIDAY
+- ▁MECHANICAL
+- ▁COTTON
+- ▁AWAKENED
+- ▁SEIZE
+- ▁RIDICULOUS
+- ▁SANCHO
+- ▁HESITATION
+- ▁CORPSE
+- ▁SAVING
+- HOLD
+- FOOT
+- ▁ELDEST
+- ▁DESPITE
+- ▁EDITH
+- ▁CHERISH
+- ▁RESISTANCE
+- ▁WILSON
+- ▁ARGUE
+- ▁INQUIRE
+- ▁APPREHENSION
+- ▁AVENUE
+- ▁DRAKE
+- ▁PROPOSE
+- HURST
+- ▁INFERIOR
+- ▁STAIRCASE
+- ▁WHEREFORE
+- ▁CARLYLE
+- ▁COUCH
+- ▁ROUTE
+- ▁POLITICS
+- ▁TOMORROW
+- ▁THRONG
+- ▁NAUGHT
+- ▁SUNLIGHT
+- ▁INDIFFERENCE
+- ▁OBEDIENCE
+- ▁RECEPTION
+- ▁VEGETABLE
+- ▁IMPERFECT
+- ▁RESIDENCE
+- ▁TURKEY
+- ▁VIOLET
+- ▁SARAH
+- ▁ALTAR
+- ▁GRIEVE
+- ▁JERK
+- ▁ENSU
+- ▁MAGICIAN
+- ▁BLOSSOM
+- ▁LANTERN
+- ▁RESOLUTE
+- ▁THOUGHTFULLY
+- ▁FORTNIGHT
+- ▁TRUMPET
+- ▁VALJEAN
+- ▁UNWILLING
+- ▁LECTURE
+- ▁WHEREUPON
+- ▁HOLLAND
+- ▁CHANGING
+- ▁CREEK
+- ▁SLICE
+- ▁NORMAL
+- ▁ANNIE
+- ▁ACCENT
+- ▁FREDERICK
+- ▁DISAGREEABLE
+- ▁RUBBED
+- ▁DUMB
+- ▁ESTABLISH
+- ▁IMPORT
+- ▁AFFIRM
+- ▁MATTHEW
+- ▁BRISK
+- ▁CONVERT
+- ▁BENDING
+- ▁IVAN
+- ▁MADEMOISELLE
+- ▁MICHAEL
+- ▁EASIER
+- ▁JONES
+- ▁FACING
+- ▁EXCELLENCY
+- ▁LITERARY
+- ▁GOSSIP
+- ▁DEVOUR
+- ▁STAGGER
+- ▁PENCIL
+- ▁AVERAGE
+- ▁HAMMER
+- ▁TRIUMPHANT
+- ▁PREFERRED
+- ▁APPLICATION
+- ▁OCCUPY
+- ▁AUTHORITIES
+- BURN
+- ▁ASCERTAIN
+- ▁CORRIDOR
+- ▁DELICIOUS
+- ▁PRACTISE
+- ▁UNIVERSE
+- ▁SHILLING
+- ▁CONTEST
+- ▁ASHORE
+- ▁COMMIT
+- ▁ADMINISTRATION
+- ▁STUDIED
+- ▁RIGID
+- ▁ADORN
+- ▁ELSEWHERE
+- ▁INNOCENCE
+- ▁JOURNAL
+- ▁LANDSCAPE
+- ▁TELEGRAPH
+- ▁ANGRILY
+- ▁CAMPAIGN
+- ▁UNJUST
+- ▁CHALLENGE
+- ▁TORRENT
+- ▁RELATE
+- ▁ASSEMBLED
+- ▁IMPRESSED
+- ▁CANOE
+- ▁CONCLUD
+- ▁QUIXOTE
+- ▁SATISFACTORY
+- ▁NIECE
+- ▁DEAF
+- ▁RAFT
+- ▁JIMMY
+- ▁GLID
+- ▁REGULAT
+- ▁CHATTER
+- ▁GLACIER
+- ▁ENVY
+- ▁STATUE
+- ▁BOSTON
+- ▁RICHMOND
+- ▁DENIED
+- ▁FANNY
+- ▁SOLOMON
+- ▁VULGAR
+- ▁STALK
+- ▁REPLACE
+- ▁SPOON
+- ▁BASIN
+- ▁FEATURE
+- ▁CONVICT
+- ▁ARCHITECT
+- ▁ADMIRAL
+- ▁RIBBON
+- ▁PERMANENT
+- ▁APRIL
+- ▁JOLLY
+- ▁NEIGHBORHOOD
+- ▁IMPART
+- BOROUGH
+- CAMP
+- ▁HORRID
+- ▁IMMORTAL
+- ▁PRUDENCE
+- ▁SPANIARD
+- ▁SUPPOSING
+- ▁TELEPHONE
+- ▁TEMPERATURE
+- ▁PENETRATE
+- ▁OYSTER
+- ▁APPOINTMENT
+- ▁EGYPTIAN
+- ▁DWELT
+- ▁NEPHEW
+- ▁RAILROAD
+- ▁SEPTEMBER
+- ▁DEVICE
+- ▁WHEAT
+- ▁GILBERT
+- ▁ELEGANT
+- ▁ADVERTISE
+- ▁RATIONAL
+- ▁TURTLE
+- ▁BROOD
+- ▁ASSEMBLY
+- ▁CULTIVATE
+- ▁EDITOR
+- ▁SPECIMEN
+- ▁UNDOUBTEDLY
+- ▁WHALE
+- ▁DROPPING
+- ▁BALLOON
+- ▁MEDICAL
+- COMB
+- ▁COMPOSITION
+- ▁FOOTSTEPS
+- ▁LAUNCELOT
+- ▁DISCOURSE
+- ▁ERRAND
+- ▁CONVERSE
+- ▁ADVANCING
+- ▁DOWNSTAIRS
+- ▁TUMULT
+- ▁CORRUPT
+- ▁SUFFICE
+- ▁ANGUISH
+- ▁SHAGGY
+- ▁RETIRE
+- ▁TIMBER
+- ▁BLAZE
+- ▁ABSTRACT
+- ▁EMBROIDER
+- ▁PHOTOGRAPH
+- ▁PROSPERITY
+- ▁TERRIBLY
+- ▁TERRITORY
+- ▁THRESHOLD
+- ▁PAVEMENT
+- ▁INJURED
+- ▁LIMP
+- ▁AGITATION
+- ▁RASCAL
+- ▁PRESUME
+- ▁OBSERVING
+- ▁OBSTACLE
+- ▁SIMPLICITY
+- ▁SLUMBER
+- ▁SUPPLIED
+- ▁COMBINATION
+- ▁DRAIN
+- ▁WILDERNESS
+- ▁BELIEVING
+- ▁VILLAIN
+- ▁RECKLESS
+- ▁INJURY
+- ▁CLAPP
+- ▁FRIDAY
+- ▁HERCULES
+- ▁KENNEDY
+- ▁SYMPTOM
+- ▁SLEDGE
+- ▁CEILING
+- ▁LEMON
+- ▁PLAGUE
+- ▁MONDAY
+- ▁CANVAS
+- ▁IMPATIENCE
+- ▁UNCOMFORTABLE
+- ▁ACCESS
+- ▁FROZEN
+- ▁SENATOR
+- ▁FRANZ
+- ▁SWIMMING
+- ▁BARRIER
+- ▁ADJUST
+- ▁COMPARISON
+- ▁PROCLAIM
+- ▁WRINKL
+- ▁OVERLOOK
+- ▁MITYA
+- ▁GUILT
+- ▁PERCEPTION
+- ▁PRECAUTION
+- ▁SPECTATOR
+- ▁SURPRISING
+- ▁DISTRACT
+- ▁DISDAIN
+- ▁BONNET
+- ▁MAGNET
+- ▁PROFESS
+- ▁CONFOUND
+- ▁NARRATIVE
+- ▁STRUCTURE
+- ▁SKETCH
+- ▁ULTIMATE
+- ▁GLOBE
+- ▁INSECT
+- FICIENCY
+- ▁ORCHARD
+- ▁AMIABLE
+- ▁DESCENT
+- ▁INDEPENDENCE
+- ▁MANUFACTURE
+- ▁SPRINKLE
+- ▁NIGHTINGALE
+- ▁CUSHION
+- ▁EMINENT
+- ▁SCOTT
+- ▁ARRAY
+- ▁COSETTE
+- ▁WAVING
+- ▁EXTRACT
+- ▁IRREGULAR
+- ▁PERSECUT
+- ▁DERIVED
+- ▁WITHDREW
+- ▁CAUTION
+- ▁SUSPICIOUS
+- ▁MEMORIES
+- ▁NOWHERE
+- ▁SUBTLE
+- ▁THOROUGH
+- Q
+- ▁APPROPRIATE
+- ▁SLAUGHTER
+- ▁YOURSELVES
+- ▁THUMB
+- ▁TWAS
+- ▁ABODE
+- ▁BIDDING
+- ▁CONSPICUOUS
+- ▁REBECCA
+- ▁SERGEANT
+- ▁APRON
+- ▁ANTICIPATE
+- ▁DISCIPLINE
+- ▁GLANCING
+- ▁PILGRIM
+- ▁SULLEN
+- ▁CONTRIBUTE
+- ▁PRAIRIE
+- ▁CARVED
+- ▁COMMERCE
+- ▁EXCLAMATION
+- ▁MUSCULAR
+- ▁NOVEMBER
+- ▁PHENOMENA
+- ▁SYMBOL
+- ▁UMBRELLA
+- ▁DIMINISH
+- ▁PARLOUR
+- ▁THREATENING
+- ▁STUMP
+- ▁EXTENSIVE
+- ▁PLEASING
+- ▁REMEMBRANCE
+- ▁COMBINED
+- ▁SHERIFF
+- ▁SHAFT
+- ▁LAURA
+- ▁INTERCOURSE
+- ▁STRICKEN
+- ▁SUPPLIES
+- ▁LANDLORD
+- ▁SHRINK
+- ▁PRICK
+- ▁CAESAR
+- ▁DRUG
+- ▁BEWILDERED
+- ▁NAUTILUS
+- ▁BRUTAL
+- ▁COMMERCIAL
+- ▁MAGGIE
+- ▁SPHERE
+- ▁VIRGIN
+- ▁BRETHREN
+- ▁DESTINY
+- ▁POLICY
+- ▁TERRIFIED
+- ▁HOUSEKEEPER
+- ▁CRAZY
+- ▁ARDENT
+- ▁DISCERN
+- ▁WRAP
+- ▁MARQUIS
+- ▁RUSSIA
+- MOUTH
+- ▁BRITAIN
+- ▁HARBOUR
+- ▁CONCERT
+- ▁DONKEY
+- ▁DAMAGE
+- ▁SLIM
+- ABOUT
+- ▁LUXURY
+- ▁MONSTROUS
+- ▁TENDENCY
+- ▁PARADISE
+- ▁CULTURE
+- ▁JULIUS
+- ▁RAOUL
+- ▁REMEDY
+- ▁DECAY
+- ▁SCOLD
+- ▁SPLIT
+- ▁ASSAULT
+- ▁DECEMBER
+- ▁MOSCOW
+- ▁EXPLORE
+- ▁TROUSERS
+- ▁WRIST
+- PIECE
+- ▁MUSKET
+- ▁VALENTINE
+- ▁TYRANT
+- ▁ABRAHAM
+- ▁MEDIUM
+- ▁ARTIFICIAL
+- ▁FACULTY
+- ▁OBLIGATION
+- ▁RESEMBLANCE
+- ▁INQUIRIES
+- ▁DETAIN
+- ▁SWARM
+- ▁PLEDGE
+- ▁ADMIRABLE
+- ▁DEFECT
+- ▁SUPERINTEND
+- ▁PATRIOT
+- ▁CLUNG
+- ▁DISMAL
+- ▁RECIT
+- ▁IGNOR
+- ▁AMELIA
+- ▁JUSTIFY
+- ▁ELEPHANT
+- ▁ESTIMATE
+- ▁KNELT
+- ▁SERVING
+- ▁WHIM
+- ▁SHRILL
+- ▁STUDIO
+- ▁TEXT
+- ▁ALEXANDER
+- ▁WROUGHT
+- ▁ABUNDANT
+- ▁SITUATED
+- ▁REGAIN
+- ▁FIERY
+- ▁SNEER
+- ▁SWEAT
+- ▁GLARE
+- ▁NIGH
+- ▁ESCORT
+- ▁INEVITABLE
+- ▁PSMITH
+- ▁RELUCTANT
+- ▁PRECEDING
+- ▁RESORT
+- ▁OUTRAGE
+- ▁AMBASSADOR
+- ▁CONSOLATION
+- ▁RECOGNITION
+- ▁REMORSE
+- ▁BEHALF
+- ▁FORMIDABLE
+- ▁GRAVITY
+- ▁DIVIDE
+- ▁CONFRONT
+- ▁GIGANTIC
+- ▁OCTOBER
+- ▁FLANK
+- ▁SLEW
+- ▁CLARA
+- ▁FILM
+- ▁BULK
+- ▁POMP
+- ▁ELEANOR
+- ▁EMPHASIS
+- ▁JAPANESE
+- ▁CAVALRY
+- ▁EXCLUSIVE
+- ▁PERFUME
+- ▁BRONZE
+- ▁FEDERAL
+- ▁LIQUID
+- ▁RUBBING
+- ▁OVEN
+- DOLPH
+- ▁CONVULS
+- ▁DEPRIVED
+- ▁RESPONSIBILITY
+- ▁SIGNIFICANT
+- ▁WAISTCOAT
+- ▁CLUSTER
+- ▁MARTHA
+- ▁REVERSE
+- ▁ATTORNEY
+- ▁DROOP
+- ▁SKILFUL
+- ▁HABITUAL
+- ▁PUMP
+- ▁INTERVEN
+- ▁OWL
+- ▁CONJECTURE
+- ▁FANTASTIC
+- ▁RESPONSIBLE
+- ▁DESTINED
+- ▁DOCUMENT
+- ▁THEREUPON
+- ▁GODDESS
+- ▁PACIFIC
+- ▁WARRANT
+- ▁COSTUME
+- ▁BRIDLE
+- ▁CALIFORNIA
+- ▁DEMOCRATIC
+- ▁EUSTACE
+- ▁SQUIRREL
+- ▁UNCOMMON
+- ▁MARVELLOUS
+- ▁PLOUGH
+- ▁TRAGEDY
+- ▁VAULT
+- ▁HESITATE
+- ▁REFRAIN
+- ▁ADMIRING
+- ▁CORPORAL
+- ▁ENTITLED
+- ▁SHREWD
+- ▁SQUEEZ
+- ▁ACCURATE
+- ▁TEMPEST
+- ▁MONUMENT
+- ▁SIEGE
+- ▁CHINESE
+- ▁RAVEN
+- ▁LOUNG
+- ▁ASSASSIN
+- ▁INFLICT
+- ▁AGITATED
+- ▁DESIRABLE
+- ▁EARLIEST
+- ▁LAUNCH
+- ▁PILOT
+- ▁PULSE
+- ▁MUTE
+- LEIGH
+- ▁LIQUOR
+- ▁SCARECROW
+- ▁SKULL
+- ▁DESOLATE
+- ▁SUBLIME
+- ▁SERENE
+- ▁RECESS
+- ▁WAKING
+- ▁CHARLOTTE
+- ▁CIRCULAR
+- ▁INJUSTICE
+- ▁PINOCCHIO
+- ▁PRISCILLA
+- ▁THYSELF
+- ▁OCCURRENCE
+- ▁CASUAL
+- ▁FRANTIC
+- ▁LEGEND
+- ▁FERTIL
+- ▁BACKGROUND
+- ▁DELICACY
+- ▁ESTRALLA
+- ▁MANUSCRIPT
+- ▁RESPONSE
+- ▁UNIVERSITY
+- ▁WOLVES
+- ▁SCANDAL
+- ▁STUMBLE
+- ▁HOARSE
+- ▁BODILY
+- ▁CONVENT
+- ▁EXAMINING
+- ▁INCAPABLE
+- ▁PERCEIVING
+- ▁PHILADELPHIA
+- ▁SUBSEQUENT
+- ▁THIEVES
+- ▁ACCUMULAT
+- ▁DAMSEL
+- ▁SCOTCH
+- ▁UNDERNEATH
+- ▁NOBILITY
+- ▁SMASH
+- ▁REVOLT
+- ▁ENGAGE
+- ▁CATHEDRAL
+- ▁CHAMPION
+- ▁DESPATCH
+- ▁ETERNITY
+- ▁JANUARY
+- ▁PLEADED
+- ▁PROBABILITY
+- ▁JIMMIE
+- ▁PARALLEL
+- ▁FISHERMAN
+- ▁JERRY
+- ▁SWORE
+- ▁DRAUGHT
+- ▁OPPONENT
+- ▁PRIMITIVE
+- ▁SIGNIFICANCE
+- ▁SUBSTANTIAL
+- ▁AMAZED
+- ▁DUNBAR
+- ▁COMMEND
+- ▁CONTEMPLATE
+- ▁TESTIMONY
+- ▁IMPERIAL
+- ▁ADAPT
+- ▁JUICE
+- ▁CALAMIT
+- CULAR
+- ▁CHATEAU
+- ▁PHOENIX
+- ▁PRUDENT
+- ▁SOLUTION
+- ▁VILLEFORT
+- ▁REACTION
+- ▁RELAX
+- ▁YU
+- ▁PROHIBIT
+- ▁DISTRUST
+- ▁PLUNDER
+- ▁WELFARE
+- ▁NAVIGAT
+- ▁PARLOR
+- ▁LAZY
+- ▁DETACH
+- OMETER
+- ▁PRIV
+- ▁DISCOURAGE
+- ▁OBSTINATE
+- ▁REJOICING
+- ▁SERMON
+- ▁VEHICLE
+- ▁FANCIES
+- ▁ENLIGHTEN
+- ▁ACUTE
+- ▁ILLUSION
+- ▁ANTHEA
+- ▁MARTIAN
+- ▁EXCITE
+- ▁GENEROSITY
+- OLOGIST
+- ▁AMAZING
+- ▁UNWORTHY
+- ▁INTERNAL
+- ▁INCENSE
+- ▁VIBRAT
+- ▁ADHERE
+- ROACH
+- ▁FEBRUARY
+- ▁MEXICAN
+- ▁POTATOES
+- ▁INCESSANT
+- ▁INTERPOSED
+- ▁PARCEL
+- ▁VEXED
+- ▁PROMOTE
+- MIDST
+- ▁ARISTOCRAT
+- ▁CYRIL
+- ▁EMBARK
+- ▁ABUNDANCE
+- ▁LITERALLY
+- ▁SURGEON
+- ▁TERRACE
+- ▁ATLANTIC
+- ▁MARTYR
+- ▁SPECK
+- ▁SENATE
+- ▁LOAF
+- ▁ADMINISTER
+- ▁APPREHEND
+- ▁SUBDUED
+- ▁TEMPORARY
+- ▁DOMINION
+- ▁ELABORATE
+- ▁DIGNIFIED
+- ▁ELIZA
+- ▁SPLASH
+- ▁CONSEIL
+- ▁DEXTER
+- ▁UNSEEN
+- ▁TRAGIC
+- VOCATION
+- ▁GRATIFY
+- ▁BACHELOR
+- ▁DEFENSE
+- ▁EXCURSION
+- ▁FACULTIES
+- ▁PROPRIETOR
+- ▁SYMPATHETIC
+- ▁UNNECESSARY
+- ▁RADIANT
+- ▁VACANT
+- ▁OUNCE
+- ▁SCREW
+- ▁PHENOMENON
+- ▁PROMINENT
+- ▁WORRIED
+- ▁STUDIES
+- ▁CLIMATE
+- ▁KEITH
+- ▁ARAMIS
+- ▁BLISS
+- ▁CONTINUAL
+- ▁SURPASS
+- ▁HEBREW
+- ▁IDENTITY
+- ▁PROVOKE
+- ▁TEMPERAMENT
+- ▁CHARIOT
+- ▁HARBOR
+- ▁NINTH
+- ▁PRIOR
+- ▁DESIROUS
+- ▁JERUSALEM
+- ▁UNDERTAKING
+- ▁EDISON
+- ▁MIRTH
+- ▁SCOUT
+- ▁APPARATUS
+- ▁ILLUSTRATION
+- ▁INTELLIGIBLE
+- ▁INVARIABLY
+- ▁PIERCED
+- ▁REVIEW
+- ▁FLICKER
+- ▁HAZARD
+- ▁REVELATION
+- ▁DIXON
+- ▁EXCITING
+- ▁GOSPEL
+- ▁CONSTANCE
+- ▁OVERTAKE
+- ▁GUINEA
+- ▁ALADDIN
+- ▁CHICAGO
+- ▁TULLIVER
+- ▁HAMILTON
+- ▁GARRISON
+- ▁DISCIPLE
+- ▁INTENSITY
+- ▁TRAITOR
+- ▁CHANCELLOR
+- ▁PROVERB
+- ▁DAGGER
+- ▁FORESEE
+- ▁CONFIDE
+- ▁GLIMMER
+- ▁CHAUVELIN
+- ▁ILLUSTRATE
+- ▁VOLUNTEER
+- ▁JUNGLE
+- ▁STREAK
+- ▁SUNRISE
+- ▁DISSOLV
+- ▁QUEST
+- ▁AWHILE
+- ▁FELICITY
+- ▁LEGISLATURE
+- ▁LEONORA
+- ▁MAGAZINE
+- ▁PITIFUL
+- ▁COLONY
+- ▁SHAWL
+- ▁ARRIVING
+- ▁FUNDAMENTAL
+- ▁CARPENTER
+- ▁OVERFLOW
+- ▁EXPAND
+- ▁HARVEST
+- ▁FEMININE
+- ▁INNUMERABLE
+- ▁SCRAMBLE
+- ▁TWENTIETH
+- ▁TRIFLING
+- ▁GHASTL
+- ▁CONQUEST
+- ▁DANIEL
+- ▁FACILIT
+- ▁FORSAKE
+- ▁BEHAVIOUR
+- ▁GORGEOUS
+- ▁PRODUCING
+- ▁HAPPIER
+- ▁PROMISING
+- ▁RAINBOW
+- ▁INSTINCTIVELY
+- ▁DECREE
+- ▁EYEBROWS
+- ▁IRRESISTIBLE
+- ▁PHARAOH
+- ▁SCROOGE
+- ▁UNNATURAL
+- ▁CRUMBS
+- ▁REFINED
+- ▁DREARY
+- ▁TRENCH
+- ▁CONVINCE
+- ▁FRINGE
+- ▁EXTREMITY
+- ▁INTIMACY
+- ▁SCOUNDREL
+- ▁SUFFRAGE
+- ▁UNEASINESS
+- ▁BARRICADE
+- ▁CIRCULAT
+- ▁SAMUEL
+- ▁BRUCE
+- ▁DARCY
+- <sos/eos>
+init: xavier_uniform
+input_size: null
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: null
+    zero_infinity: true
+    brctc_risk_strategy: exp
+    brctc_group_strategy: end
+    brctc_risk_factor: 0.0
+joint_net_conf: null
+use_preprocessor: true
+use_lang_prompt: false
+use_nlp_prompt: false
+token_type: bpe
+bpemodel: data/en_token_list/bpe_unigram5000/bpe.model
+non_linguistic_symbols: null
+cleaner: null
+g2p: null
+speech_volume_normalize: null
+rir_scp: null
+rir_apply_prob: 1.0
+noise_scp: null
+noise_apply_prob: 1.0
+noise_db_range: '13_15'
+short_noise_thres: 0.5
+aux_ctc_tasks: []
+frontend: default
+frontend_conf:
+    fs: 16k
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
+normalize: global_mvn
+normalize_conf:
+    stats_file: exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz
+model: espnet
+model_conf:
+    ctc_weight: 0.3
+    lsm_weight: 0.1
+    length_normalized_loss: false
+preencoder: null
+preencoder_conf: {}
+encoder: q_transformer
+encoder_conf:
+    output_size: 512
+    attention_heads: 8
+    attention_layer_type: HierDecayv2
+    linear_units: 2048
+    num_blocks: 18
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.1
+    input_layer: conv2d6
+    normalize_before: true
+postencoder: null
+postencoder_conf: {}
+decoder: transformer
+decoder_conf:
+    attention_heads: 8
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.1
+    src_attention_dropout_rate: 0.1
+preprocessor: default
+preprocessor_conf: {}
+required:
+- output_dir
+- token_list
+version: '202402'
+distributed: true
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/acc.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/acc.png
new file mode 100644
index 0000000000000000000000000000000000000000..3b9fae7ba74c09956513c6f0589fe3673a03429b
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/acc.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/backward_time.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/backward_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..7f0d47e5545a713210bf03421b04f83704171089
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/backward_time.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer.png
new file mode 100644
index 0000000000000000000000000000000000000000..07af221c2a37646f2f4da8304bb2ff536ac54ea8
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer_ctc.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer_ctc.png
new file mode 100644
index 0000000000000000000000000000000000000000..fd896dcc82bfa2bc1b6d4d5401ced9981916451d
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/cer_ctc.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/clip.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/clip.png
new file mode 100644
index 0000000000000000000000000000000000000000..490d2c5acece3d73a6d832551b5bd1897a0fb90c
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/clip.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/forward_time.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/forward_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..bccd3965cd9aed3372e88e2a7949371b2d7b719d
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/forward_time.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/gpu_max_cached_mem_GB.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/gpu_max_cached_mem_GB.png
new file mode 100644
index 0000000000000000000000000000000000000000..2bfd80c684be4204ab5e686f54cf8e22f01b5c26
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/gpu_max_cached_mem_GB.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/grad_norm.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/grad_norm.png
new file mode 100644
index 0000000000000000000000000000000000000000..419735bae7ec81a963647ea036f06b334017b6b8
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/grad_norm.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/iter_time.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/iter_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..98949acaea78cbc5763a58eaa7efa14b75b90d18
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/iter_time.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss.png
new file mode 100644
index 0000000000000000000000000000000000000000..02f25a0658913e1a34ee5c5335790d4c9b067fef
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_att.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_att.png
new file mode 100644
index 0000000000000000000000000000000000000000..134ddd9e4ac62d12848d28a30a59a7d42af840db
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_att.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_ctc.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_ctc.png
new file mode 100644
index 0000000000000000000000000000000000000000..144ba6672e67287596164dc18bc1c5508e4f3541
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_ctc.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_scale.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_scale.png
new file mode 100644
index 0000000000000000000000000000000000000000..5c34a39d539ba692d9382e4670b8bfccc1dfd0cf
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/loss_scale.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim0_lr0.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim0_lr0.png
new file mode 100644
index 0000000000000000000000000000000000000000..368cfbad77df1650eda4f0cbceb0a1512d8452d7
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim0_lr0.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim_step_time.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim_step_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..0ece80ff035c63c34e1ad51f00b87382c0644daf
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/optim_step_time.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/train_time.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/train_time.png
new file mode 100644
index 0000000000000000000000000000000000000000..1874ee97e420cfb2a36a06df0a6458d9513fba23
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/train_time.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/wer.png b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/wer.png
new file mode 100644
index 0000000000000000000000000000000000000000..7ecb8e671855c2ee9e5361c43577fcaf152902f0
Binary files /dev/null and b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/images/wer.png differ
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/latest.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/latest.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2ff0ee5449b9b8ee13277d3889f14fa2c370a5a3
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/latest.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dd0daaefe2355e6b0151b39c8f53b350fc4f531973a9cff2cb5dfb6fffeaddf
+size 398012507
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/run.sh b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/run.sh
new file mode 100644
index 0000000000000000000000000000000000000000..cb7f39c2e11650edae95c93c7669c8bf3ca82eb2
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/run.sh
@@ -0,0 +1 @@
+./asr.sh --skip_data_prep true --stage 11 --lang en --ngpu 4 --use_lm false --nbpe 5000 --max_wav_duration 30 --asr_tag transformer_HierDecayv2_woBias_XavierInit_Qv7 --speed_perturb_factors '0.9 1.0 1.1' --asr_config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --lm_config conf/tuning/train_lm_transformer2.yaml --inference_config conf/decode_asr.yaml --train_set train_960 --valid_set dev --test_sets 'test_clean test_other dev_clean dev_other' --lm_train_text 'data/train_960/text data/local/other_text/text' --bpe_train_text data/train_960/text --stage 11 --stage 11 --stage 11 "$@"; exit $?
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1720237326.bmi2.3524061.0 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1720237326.bmi2.3524061.0
new file mode 100644
index 0000000000000000000000000000000000000000..2ffc90ea2949374c7dcf021a374ff65402673d92
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1720237326.bmi2.3524061.0
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e389810df809c7f2dae4fa896397bb161c71f7d0a80d0f562b62cc88f0291f0
+size 140505218
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721133711.bmi2.1781101.0 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721133711.bmi2.1781101.0
new file mode 100644
index 0000000000000000000000000000000000000000..bf90b68eb1c8b0c40fc5f54341a909192aebc53d
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721133711.bmi2.1781101.0
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c7227f05b7abc16967fc1a41727ea41d048a5d8cae4eec3ab31c9431854db80
+size 723
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721134148.bmi2.1856035.0 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721134148.bmi2.1856035.0
new file mode 100644
index 0000000000000000000000000000000000000000..bd286a093741223da3961bf6c973e1de2801c5da
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/train/events.out.tfevents.1721134148.bmi2.1856035.0
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f12dae579bb931487ce29814644f11cbc964734e8f3cebf67052bc1c3dd8557
+size 352256376
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1720237326.bmi2.3524061.1 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1720237326.bmi2.3524061.1
new file mode 100644
index 0000000000000000000000000000000000000000..c899b3095a138a2e191f347978de6bff848f18f9
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1720237326.bmi2.3524061.1
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7df5e3f820a7014cd8b7729120f0335c00b06ff2ca558dd210df3b42c229c556
+size 27601
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721133711.bmi2.1781101.1 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721133711.bmi2.1781101.1
new file mode 100644
index 0000000000000000000000000000000000000000..e2b3810daee605e446813b4a2573d5e46b7a9195
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721133711.bmi2.1781101.1
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d01cc65e5c3d9a8578da8dc9cca430c31d30ab82661c01f2986e23e0db71f6b1
+size 88
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721134148.bmi2.1856035.1 b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721134148.bmi2.1856035.1
new file mode 100644
index 0000000000000000000000000000000000000000..6a7da44587ffbd215ccdb003785309ba79efbe42
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/tensorboard/valid/events.out.tfevents.1721134148.bmi2.1856035.1
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f4a8940d85c376e7ea7afcb132a358fc632019cb8d9ae4f3dc960e92d1fe152
+size 10267
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.1.log b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.1.log
new file mode 100644
index 0000000000000000000000000000000000000000..9698f293ed18a9cd49ec4e53b4f5839d296ba6ba
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.1.log
@@ -0,0 +1,1456 @@
+# python3 -m espnet2.bin.asr_train --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True 
+# Started at Tue Jul 16 20:41:07 HKT 2024
+#
+/home/zysong/espnet/tools/miniconda/envs/espnet/bin/python3 /home/zysong/espnet/espnet2/bin/asr_train.py --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True
+[bmi2:0/4] 2024-07-16 20:41:28,938 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[bmi2:0/4] 2024-07-16 20:41:28,939 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes.
+[bmi2:0/4] 2024-07-16 20:41:28,955 (asr:523) INFO: Vocabulary size: 5000
+[bmi2:0/4] 2024-07-16 20:41:30,590 (initialize:88) INFO: Initialize encoder.embed.conv.0.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,590 (initialize:88) INFO: Initialize encoder.embed.conv.2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,590 (initialize:88) INFO: Initialize encoder.embed.out.0.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,591 (initialize:88) INFO: Initialize encoder.encoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.6.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.6.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.6.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.7.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.7.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.7.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.8.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.8.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.8.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.9.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.9.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.9.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.10.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,592 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.10.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.10.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.11.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.11.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.11.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.12.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.12.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.12.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.13.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.13.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.13.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.14.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.14.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.14.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.15.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.15.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,593 (initialize:88) INFO: Initialize encoder.encoders.15.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.16.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.16.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.16.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.17.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.17.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.encoders.17.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize encoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.output_layer.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.0.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,594 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.1.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.2.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,595 (initialize:88) INFO: Initialize decoder.decoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.3.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.4.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize decoder.decoders.5.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:30,596 (initialize:88) INFO: Initialize ctc.ctc_lo.bias to zeros
+[bmi2:0/4] 2024-07-16 20:41:32,559 (abs_task:1320) INFO: pytorch.version=1.13.1+cu117, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[bmi2:0/4] 2024-07-16 20:41:32,566 (abs_task:1321) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): Q_TransformerEncoder(
+    (embed): Conv2dSubsampling6(
+      (conv): Sequential(
+        (0): Conv2d(1, 512, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(512, 512, kernel_size=(5, 5), stride=(3, 3))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=6144, out_features=512, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (1): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (2): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (3): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (4): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (5): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (6): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (7): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (8): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (9): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (10): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (11): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (12): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (13): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (14): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (15): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (16): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (17): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(5000, 512)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=512, out_features=5000, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=512, out_features=5000, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 99.34 M
+    Number of trainable parameters: 99.34 M (100.0%)
+    Size: 397.35 MB
+    Type: torch.float32
+[bmi2:0/4] 2024-07-16 20:41:32,567 (abs_task:1324) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    capturable: False
+    differentiable: False
+    eps: 1e-08
+    foreach: None
+    fused: False
+    initial_lr: 0.002
+    lr: 8e-08
+    maximize: False
+    weight_decay: 0
+)
+[bmi2:0/4] 2024-07-16 20:41:32,567 (abs_task:1325) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[bmi2:0/4] 2024-07-16 20:41:32,580 (abs_task:1334) INFO: Saving the configuration in exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml
+[bmi2:0/4] 2024-07-16 20:41:34,569 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:41:46,187 (abs_task:1714) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/train_960_sp/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/train_960_sp/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f02b1eaeac0>)
+[bmi2:0/4] 2024-07-16 20:41:46,187 (abs_task:1715) INFO: [train] Batch sampler: NumElementsBatchSampler(N-batch=8267, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-16 20:41:46,189 (abs_task:1716) INFO: [train] mini-batch sizes summary: N-batch=8267, mean=102.1, min=41, max=683
+[bmi2:0/4] 2024-07-16 20:41:46,271 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:41:46,311 (abs_task:1714) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f02b1c480d0>)
+[bmi2:0/4] 2024-07-16 20:41:46,312 (abs_task:1715) INFO: [valid] Batch sampler: NumElementsBatchSampler(N-batch=34, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-16 20:41:46,312 (abs_task:1716) INFO: [valid] mini-batch sizes summary: N-batch=34, mean=163.3, min=7, max=445
+[bmi2:0/4] 2024-07-16 20:41:46,322 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:41:46,328 (abs_task:1714) INFO: [plot_att] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f02b1c481c0>)
+[bmi2:0/4] 2024-07-16 20:41:46,328 (abs_task:1715) INFO: [plot_att] Batch sampler: UnsortedBatchSampler(N-batch=5551, batch_size=1, key_file=exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape, 
+[bmi2:0/4] 2024-07-16 20:41:46,329 (abs_task:1716) INFO: [plot_att] mini-batch sizes summary: N-batch=3, mean=1.0, min=1, max=1
+[bmi2:0/4] 2024-07-16 20:41:47,887 (trainer:174) INFO: The training was resumed using exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/checkpoint.pth
+bmi2:1781101:1781101 [0] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1781101:1781101 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1781101:1781101 [0] NCCL INFO cudaDriverVersion 12040
+NCCL version 2.14.3+cuda11.7
+bmi2:1781104:1781104 [1] NCCL INFO cudaDriverVersion 12040
+bmi2:1781104:1781104 [1] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1781104:1781104 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1781104:1786269 [1] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1781104:1786269 [1] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1781104:1786269 [1] NCCL INFO Using network Socket
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) abmi2:1781101:1785895 [0] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1781101:1785895 [0] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1781101:1785895 [0] NCCL INFO Using network Socket
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) andbmi2:1781111:1781111 [2] NCCL INFO cudaDriverVersion 12040
+bmi2:1781111:1781111 [2] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1781111:1781111 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1781111:1785897 [2] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1781111:1785897 [2] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1781111:1785897 [2] NCCL INFO Using network Socket
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) abmi2:1781114:1781114 [3] NCCL INFO cudaDriverVersion 12040
+bmi2:1781114:1781114 [3] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1781114:1781114 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1781114:1785896 [3] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1781114:1785896 [3] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1781114:1785896 [3] NCCL INFO Using network Socket
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) a dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff,00000000,ffffffff
+bmi2:1781101:1785895 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+bmi2:1781101:1785895 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+bmi2:1781101:1785895 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+bmi2:1781101:1785895 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+bmi2:1781101:1785895 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO Channel 00 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO Channel 01 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO Channel 02 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+nd dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781114:1785896 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff,00000000,ffffffff
+bmi2:1781114:1785896 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 00 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 01 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 02 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1781114:1785896 [3] NCCLnd dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff,00000000,ffffffff
+bmi2:1781111:1785897 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 00 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 01 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 02 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFnd dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781104:1786269 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff,00000000,ffffffff
+bmi2:1781104:1786269 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 00 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 01 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 02 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INF[bmi2:0/4] 2024-07-16 20:41:51,900 (trainer:311) INFO: 74/100epoch started
+[bmi2:0/4] 2024-07-16 20:42:08,674 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[bmi2:0/4] 2024-07-16 20:42:10,842 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 20:42:37,414 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 20:47:01,251 (trainer:779) INFO: 74epoch:train:1-413batch: iter_time=0.002, forward_time=0.183, loss_ctc=66.995, loss_att=36.113, acc=0.828, loss=45.378, backward_time=0.334, grad_norm=366.153, clip=100.000, loss_scale=1.425e+33, optim_step_time=0.039, optim0_lr0=5.757e-04, train_time=1.499
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/runpy.py", line 194, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/runpy.py", line 87, in _run_code
+    exec(code, run_globals)
+  File "/home/zysong/espnet/espnet2/bin/asr_train.py", line 23, in <module>
+    main()
+  File "/home/zysong/espnet/espnet2/bin/asr_train.py", line 19, in main
+    ASRTask.main(cmd=cmd)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1219, in main
+    while not ProcessContext(processes, error_queues).join():
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 109, in join
+    ready = multiprocessing.connection.wait(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/connection.py", line 931, in wait
+    ready = selector.select(timeout)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/selectors.py", line 415, in select
+    fd_event_list = self._selector.poll(timeout)
+KeyboardInterrupt
+Process SpawnProcess-3:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 677, in train_one_epoch
+    scaler.scale(loss).backward()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/_tensor.py", line 488, in backward
+    torch.autograd.backward(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/autograd/__init__.py", line 197, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+KeyboardInterrupt
+O Channel 03 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO Connected all rings
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 00 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 01 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 02 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781111:1785897 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1781111:1785897 [2] NCCL INFO Channel 03 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1781111:1785897 [2] NCCL INFO Connected all trees
+bmi2:1781111:1785897 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1781111:1785897 [2] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1781111:1785897 [2] NCCL INFO comm 0xaff2450 rank 2 nranks 4 cudaDev 2 busId 56000 - Init COMPLETE
+Process SpawnProcess-1:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 677, in train_one_epoch
+    scaler.scale(loss).backward()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/_tensor.py", line 488, in backward
+    torch.autograd.backward(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/autograd/__init__.py", line 197, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+KeyboardInterrupt
+bmi2:1781101:1785895 [0] NCCL INFO Channel 03 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1781101:1785895 [0] NCCL INFO Connected all rings
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781101:1785895 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1781101:1785895 [0] NCCL INFO Connected all trees
+bmi2:1781101:1785895 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1781101:1785895 [0] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1781101:1785895 [0] NCCL INFO comm 0x976a890 rank 0 nranks 4 cudaDev 0 busId 4f000 - Init COMPLETE
+Process SpawnProcess-4:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 677, in train_one_epoch
+    scaler.scale(loss).backward()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/_tensor.py", line 488, in backward
+    torch.autograd.backward(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/autograd/__init__.py", line 197, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+KeyboardInterrupt
+ INFO Channel 03 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO Connected all rings
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 00 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 01 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 02 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781114:1785896 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1781114:1785896 [3] NCCL INFO Channel 03 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1781114:1785896 [3] NCCL INFO Connected all trees
+bmi2:1781114:1785896 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1781114:1785896 [3] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1781114:1785896 [3] NCCL INFO comm 0x8887bb0 rank 3 nranks 4 cudaDev 3 busId 57000 - Init COMPLETE
+Process SpawnProcess-2:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 677, in train_one_epoch
+    scaler.scale(loss).backward()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/_tensor.py", line 488, in backward
+    torch.autograd.backward(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/autograd/__init__.py", line 197, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+KeyboardInterrupt
+O Channel 03 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO Connected all rings
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 00 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 01 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 02 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1781104:1786269 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1781104:1786269 [1] NCCL INFO Channel 03 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1781104:1786269 [1] NCCL INFO Connected all trees
+bmi2:1781104:1786269 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1781104:1786269 [1] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1781104:1786269 [1] NCCL INFO comm 0x8f0ae70 rank 1 nranks 4 cudaDev 1 busId 52000 - Init COMPLETE
+bmi2:1781111:1786288 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+bmi2:1781111:1781111 [2] NCCL INFO comm 0xaff2450 rank 2 nranks 4 cudaDev 2 busId 56000 - Abort COMPLETE
+bmi2:1781101:1786287 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+bmi2:1781101:1781101 [0] NCCL INFO comm 0x976a890 rank 0 nranks 4 cudaDev 0 busId 4f000 - Abort COMPLETE
+bmi2:1781114:1786289 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+bmi2:1781114:1781114 [3] NCCL INFO comm 0x8887bb0 rank 3 nranks 4 cudaDev 3 busId 57000 - Abort COMPLETE
+bmi2:1781104:1786286 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+bmi2:1781104:1781104 [1] NCCL INFO comm 0x8f0ae70 rank 1 nranks 4 cudaDev 1 busId 52000 - Abort COMPLETE
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.2.log b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.2.log
new file mode 100644
index 0000000000000000000000000000000000000000..e5f0baf2dca22868a05b2344e00e646558d2c7bf
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.2.log
@@ -0,0 +1,3242 @@
+# python3 -m espnet2.bin.asr_train --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True 
+# Started at Sat Jul  6 11:40:55 HKT 2024
+#
+/home/zysong/espnet/tools/miniconda/envs/espnet/bin/python3 /home/zysong/espnet/espnet2/bin/asr_train.py --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True
+[bmi2:0/4] 2024-07-06 11:41:53,317 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[bmi2:0/4] 2024-07-06 11:41:53,318 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes.
+[bmi2:0/4] 2024-07-06 11:41:53,333 (asr:523) INFO: Vocabulary size: 5000
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.embed.conv.0.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.embed.conv.2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.embed.out.0.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.encoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,903 (initialize:88) INFO: Initialize encoder.encoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,904 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.6.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.6.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.6.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.7.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.7.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.7.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.8.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.8.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.8.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.9.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.9.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.9.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.10.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,905 (initialize:88) INFO: Initialize encoder.encoders.10.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.10.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.11.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.11.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.11.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.12.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.12.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.12.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.13.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.13.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.13.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.14.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.14.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.14.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.15.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.15.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,906 (initialize:88) INFO: Initialize encoder.encoders.15.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.16.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.16.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.16.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.17.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.17.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.encoders.17.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize encoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.output_layer.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.0.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,907 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.1.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.2.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,908 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.3.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.4.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,909 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,910 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,910 (initialize:88) INFO: Initialize decoder.decoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,910 (initialize:88) INFO: Initialize decoder.decoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,910 (initialize:88) INFO: Initialize decoder.decoders.5.norm3.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:54,910 (initialize:88) INFO: Initialize ctc.ctc_lo.bias to zeros
+[bmi2:0/4] 2024-07-06 11:41:56,243 (abs_task:1320) INFO: pytorch.version=1.13.1+cu117, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[bmi2:0/4] 2024-07-06 11:41:56,249 (abs_task:1321) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): Q_TransformerEncoder(
+    (embed): Conv2dSubsampling6(
+      (conv): Sequential(
+        (0): Conv2d(1, 512, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(512, 512, kernel_size=(5, 5), stride=(3, 3))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=6144, out_features=512, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (1): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (2): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (3): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (4): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (5): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (6): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (7): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (8): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (9): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (10): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (11): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (12): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (13): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (14): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (15): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (16): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (17): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(5000, 512)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=512, out_features=5000, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=512, out_features=5000, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 99.34 M
+    Number of trainable parameters: 99.34 M (100.0%)
+    Size: 397.35 MB
+    Type: torch.float32
+[bmi2:0/4] 2024-07-06 11:41:56,249 (abs_task:1324) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    capturable: False
+    differentiable: False
+    eps: 1e-08
+    foreach: None
+    fused: False
+    initial_lr: 0.002
+    lr: 8e-08
+    maximize: False
+    weight_decay: 0
+)
+[bmi2:0/4] 2024-07-06 11:41:56,249 (abs_task:1325) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[bmi2:0/4] 2024-07-06 11:41:56,250 (abs_task:1334) INFO: Saving the configuration in exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml
+[bmi2:0/4] 2024-07-06 11:41:57,823 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-06 11:42:04,734 (abs_task:1714) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/train_960_sp/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/train_960_sp/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7fd48e7f05b0>)
+[bmi2:0/4] 2024-07-06 11:42:04,735 (abs_task:1715) INFO: [train] Batch sampler: NumElementsBatchSampler(N-batch=8267, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-06 11:42:04,736 (abs_task:1716) INFO: [train] mini-batch sizes summary: N-batch=8267, mean=102.1, min=41, max=683
+[bmi2:0/4] 2024-07-06 11:42:04,779 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-06 11:42:04,812 (abs_task:1714) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7fd48eb2e580>)
+[bmi2:0/4] 2024-07-06 11:42:04,812 (abs_task:1715) INFO: [valid] Batch sampler: NumElementsBatchSampler(N-batch=34, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-06 11:42:04,812 (abs_task:1716) INFO: [valid] mini-batch sizes summary: N-batch=34, mean=163.3, min=7, max=445
+[bmi2:0/4] 2024-07-06 11:42:04,821 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-06 11:42:04,827 (abs_task:1714) INFO: [plot_att] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7fd48eb2e2b0>)
+[bmi2:0/4] 2024-07-06 11:42:04,827 (abs_task:1715) INFO: [plot_att] Batch sampler: UnsortedBatchSampler(N-batch=5551, batch_size=1, key_file=exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape, 
+[bmi2:0/4] 2024-07-06 11:42:04,827 (abs_task:1716) INFO: [plot_att] mini-batch sizes summary: N-batch=3, mean=1.0, min=1, max=1
+bmi2:3524061:3524061 [0] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:3524061:3524061 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:3524061:3524061 [0] NCCL INFO cudaDriverVersion 12040
+NCCL version 2.14.3+cuda11.7
+bmi2:3524072:3524072 [3] NCCL INFO cudaDriverVersion 12040
+bmi2:3524072:3524072 [3] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:3524072:3524072 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:3524072:3530900 [3] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:3524072:3530900 [3] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0> [4]tailscale0:fe80::863c:41d5:90ef:ed81%tailscale0<0>
+bmi2:3524072:3530900 [3] NCCL INFO Using network Socket
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2Pbmi2:3524064:3524064 [1] NCCL INFO cudaDriverVersion 12040
+bmi2:3524064:3524064 [1] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:3524064:3524064 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:3524064:3530922 [1] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:3524064:3530922 [1] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0> [4]tailscale0:fe80::863c:41d5:90ef:ed81%tailscale0<0>
+bmi2:3524064:3530922 [1] NCCL INFO Using network Socket
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2Pbmi2:3524071:3524071 [2] NCCL INFO cudaDriverVersion 12040
+bmi2:3524071:3524071 [2] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:3524071:3524071 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:3524071:3530901 [2] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:3524071:3530901 [2] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0> [4]tailscale0:fe80::863c:41d5:90ef:ed81%tailscale0<0>
+bmi2:3524071:3530901 [2] NCCL INFO Using network Socket
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2Pbmi2:3524061:3530899 [0] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:3524061:3530899 [0] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0> [4]tailscale0:fe80::863c:41d5:90ef:ed81%tailscale0<0>
+bmi2:3524061:3530899 [0] NCCL INFO Using network Socket
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P b between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524064:3530922 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff,00000000,ffffffff
+bmi2:3524064:3530922 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 00 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 01 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 02 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524etween dev 1(=52000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff,00000000,ffffffff
+bmi2:3524061:3530899 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+bmi2:3524061:3530899 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+bmi2:3524061:3530899 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+bmi2:3524061:3530899 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+bmi2:3524061:3530899 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO Channel 00 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO Channel 01 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO Channel 02 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4 between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524072:3530900 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff,00000000,ffffffff
+bmi2:3524072:3530900 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 00 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 01 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 02 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2: between dev 0(=4f000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff,00000000,ffffffff
+bmi2:3524071:3530901 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 00 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 01 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 02 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524[bmi2:0/4] 2024-07-06 11:42:06,409 (trainer:311) INFO: 1/100epoch started
+[bmi2:0/4] 2024-07-06 11:42:31,273 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[bmi2:0/4] 2024-07-06 11:46:36,724 (trainer:779) INFO: 1epoch:train:1-413batch: iter_time=0.003, forward_time=0.161, loss_ctc=658.407, loss_att=297.081, acc=0.022, loss=405.479, backward_time=0.253, grad_norm=582.291, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=8.360e-06, train_time=1.310
+[bmi2:0/4] 2024-07-06 11:50:42,543 (trainer:779) INFO: 1epoch:train:414-826batch: iter_time=0.004, forward_time=0.156, loss_ctc=278.871, loss_att=237.321, acc=0.056, loss=249.786, backward_time=0.255, grad_norm=30.432, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=2.488e-05, train_time=1.190
+[bmi2:0/4] 2024-07-06 11:54:46,321 (trainer:779) INFO: 1epoch:train:827-1239batch: iter_time=8.925e-04, forward_time=0.155, loss_ctc=280.212, loss_att=229.676, acc=0.080, loss=244.837, backward_time=0.255, grad_norm=22.870, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=4.140e-05, train_time=1.181
+[bmi2:0/4] 2024-07-06 11:58:52,465 (trainer:779) INFO: 1epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.156, loss_ctc=273.968, loss_att=215.814, acc=0.098, loss=233.260, backward_time=0.256, grad_norm=21.760, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=5.792e-05, train_time=1.191
+[bmi2:0/4] 2024-07-06 12:02:58,214 (trainer:779) INFO: 1epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.158, loss_ctc=275.927, loss_att=213.819, acc=0.108, loss=232.451, backward_time=0.256, grad_norm=23.912, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.030, optim0_lr0=7.444e-05, train_time=1.191
+[bmi2:0/4] 2024-07-06 12:07:05,390 (trainer:779) INFO: 1epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.158, loss_ctc=263.549, loss_att=210.477, acc=0.114, loss=226.399, backward_time=0.256, grad_norm=28.277, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=9.096e-05, train_time=1.196
+[bmi2:0/4] 2024-07-06 12:11:11,053 (trainer:779) INFO: 1epoch:train:2479-2891batch: iter_time=7.455e-04, forward_time=0.156, loss_ctc=244.853, loss_att=206.796, acc=0.121, loss=218.213, backward_time=0.257, grad_norm=30.372, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=1.075e-04, train_time=1.190
+[bmi2:0/4] 2024-07-06 12:15:17,609 (trainer:779) INFO: 1epoch:train:2892-3304batch: iter_time=0.004, forward_time=0.156, loss_ctc=230.017, loss_att=204.451, acc=0.127, loss=212.121, backward_time=0.256, grad_norm=31.964, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=1.240e-04, train_time=1.193
+[bmi2:0/4] 2024-07-06 12:19:23,435 (trainer:779) INFO: 1epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.156, loss_ctc=216.515, loss_att=199.798, acc=0.132, loss=204.813, backward_time=0.257, grad_norm=35.193, clip=100.000, loss_scale=6.554e+04, optim_step_time=0.029, optim0_lr0=1.405e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 12:23:28,915 (trainer:779) INFO: 1epoch:train:3718-4130batch: iter_time=0.004, forward_time=0.156, loss_ctc=199.709, loss_att=189.908, acc=0.136, loss=192.848, backward_time=0.255, grad_norm=39.211, clip=100.000, loss_scale=8.611e+04, optim_step_time=0.029, optim0_lr0=1.570e-04, train_time=1.188
+[bmi2:0/4] 2024-07-06 12:27:35,286 (trainer:779) INFO: 1epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.157, loss_ctc=199.327, loss_att=194.060, acc=0.139, loss=195.640, backward_time=0.257, grad_norm=42.631, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=1.736e-04, train_time=1.194
+[bmi2:0/4] 2024-07-06 12:31:41,866 (trainer:779) INFO: 1epoch:train:4544-4956batch: iter_time=0.007, forward_time=0.155, loss_ctc=182.858, loss_att=181.486, acc=0.146, loss=181.897, backward_time=0.255, grad_norm=44.159, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=1.901e-04, train_time=1.193
+[bmi2:0/4] 2024-07-06 12:35:44,235 (trainer:779) INFO: 1epoch:train:4957-5369batch: iter_time=1.903e-04, forward_time=0.154, loss_ctc=182.138, loss_att=184.078, acc=0.153, loss=183.496, backward_time=0.254, grad_norm=47.819, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=2.066e-04, train_time=1.174
+[bmi2:0/4] 2024-07-06 12:39:50,251 (trainer:779) INFO: 1epoch:train:5370-5782batch: iter_time=0.001, forward_time=0.157, loss_ctc=192.487, loss_att=199.507, acc=0.161, loss=197.401, backward_time=0.258, grad_norm=55.407, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.030, optim0_lr0=2.231e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 12:43:56,211 (trainer:779) INFO: 1epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.157, loss_ctc=176.247, loss_att=186.157, acc=0.173, loss=183.184, backward_time=0.256, grad_norm=59.166, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=2.396e-04, train_time=1.192
+[bmi2:0/4] 2024-07-06 12:48:04,195 (trainer:779) INFO: 1epoch:train:6196-6608batch: iter_time=0.005, forward_time=0.157, loss_ctc=169.721, loss_att=182.727, acc=0.185, loss=178.825, backward_time=0.258, grad_norm=63.594, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=2.562e-04, train_time=1.200
+[bmi2:0/4] 2024-07-06 12:52:09,701 (trainer:779) INFO: 1epoch:train:6609-7021batch: iter_time=5.995e-04, forward_time=0.157, loss_ctc=171.420, loss_att=186.509, acc=0.196, loss=181.983, backward_time=0.256, grad_norm=74.305, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=2.727e-04, train_time=1.189
+[bmi2:0/4] 2024-07-06 12:56:13,301 (trainer:779) INFO: 1epoch:train:7022-7434batch: iter_time=0.003, forward_time=0.155, loss_ctc=155.415, loss_att=168.885, acc=0.217, loss=164.844, backward_time=0.255, grad_norm=84.708, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=2.892e-04, train_time=1.179
+[bmi2:0/4] 2024-07-06 13:00:19,177 (trainer:779) INFO: 1epoch:train:7435-7847batch: iter_time=0.006, forward_time=0.155, loss_ctc=152.312, loss_att=164.904, acc=0.234, loss=161.126, backward_time=0.256, grad_norm=99.091, clip=100.000, loss_scale=1.311e+05, optim_step_time=0.029, optim0_lr0=3.057e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 13:04:22,054 (trainer:779) INFO: 1epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.153, loss_ctc=149.949, loss_att=159.622, acc=0.257, loss=156.720, backward_time=0.254, grad_norm=111.075, clip=100.000, loss_scale=2.134e+05, optim_step_time=0.028, optim0_lr0=3.222e-04, train_time=1.175
+[bmi2:0/4] 2024-07-06 13:05:13,198 (font_manager:1021) INFO: Failed to extract font properties from /usr/share/fonts/truetype/noto/NotoColorEmoji.ttf: In FT2Font: Can not load face (unknown file format; error code 0x2)
+[bmi2:0/4] 2024-07-06 13:05:13,992 (font_manager:1547) INFO: generated new fontManager
+[bmi2:0/4] 2024-07-06 13:06:04,820 (trainer:365) INFO: 1epoch results: [train] iter_time=0.003, forward_time=0.156, loss_ctc=232.297, loss_att=200.416, acc=0.143, loss=209.980, backward_time=0.256, grad_norm=76.381, clip=100.000, loss_scale=1.036e+05, optim_step_time=0.029, optim0_lr0=1.654e-04, train_time=1.195, time=1 hour, 22 minutes and 20.62 seconds, total_count=8267, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=71.025, cer_ctc=0.497, loss_att=84.263, acc=0.325, cer=0.534, wer=0.999, loss=80.292, time=41.53 seconds, total_count=34, gpu_max_cached_mem_GB=22.482, [att_plot] time=56.27 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 13:06:08,742 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 13:06:08,742 (trainer:299) INFO: 2/100epoch started. Estimated time to finish: 5 days, 18 hours and 39 minutes
+[bmi2:0/4] 2024-07-06 13:10:44,717 (trainer:779) INFO: 2epoch:train:1-413batch: iter_time=0.002, forward_time=0.153, loss_ctc=149.269, loss_att=156.014, acc=0.279, loss=153.990, backward_time=0.252, grad_norm=124.242, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=3.390e-04, train_time=1.337
+[bmi2:0/4] 2024-07-06 13:14:49,892 (trainer:779) INFO: 2epoch:train:414-826batch: iter_time=0.005, forward_time=0.156, loss_ctc=141.931, loss_att=144.342, acc=0.311, loss=143.619, backward_time=0.254, grad_norm=144.819, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=3.555e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 13:18:54,008 (trainer:779) INFO: 2epoch:train:827-1239batch: iter_time=0.002, forward_time=0.156, loss_ctc=133.539, loss_att=131.589, acc=0.346, loss=132.174, backward_time=0.254, grad_norm=159.508, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=3.720e-04, train_time=1.183
+[bmi2:0/4] 2024-07-06 13:22:57,421 (trainer:779) INFO: 2epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.155, loss_ctc=131.189, loss_att=124.549, acc=0.388, loss=126.541, backward_time=0.253, grad_norm=179.873, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=3.886e-04, train_time=1.178
+[bmi2:0/4] 2024-07-06 13:27:00,968 (trainer:779) INFO: 2epoch:train:1653-2065batch: iter_time=3.886e-04, forward_time=0.156, loss_ctc=127.761, loss_att=114.801, acc=0.437, loss=118.689, backward_time=0.254, grad_norm=178.515, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=4.051e-04, train_time=1.180
+[bmi2:0/4] 2024-07-06 13:31:05,036 (trainer:779) INFO: 2epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.155, loss_ctc=118.942, loss_att=101.239, acc=0.482, loss=106.550, backward_time=0.255, grad_norm=185.188, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=4.216e-04, train_time=1.181
+[bmi2:0/4] 2024-07-06 13:35:09,445 (trainer:779) INFO: 2epoch:train:2479-2891batch: iter_time=0.004, forward_time=0.156, loss_ctc=110.607, loss_att=90.400, acc=0.519, loss=96.462, backward_time=0.254, grad_norm=180.579, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=4.381e-04, train_time=1.184
+[bmi2:0/4] 2024-07-06 13:39:15,040 (trainer:779) INFO: 2epoch:train:2892-3304batch: iter_time=0.005, forward_time=0.156, loss_ctc=104.671, loss_att=82.077, acc=0.555, loss=88.855, backward_time=0.254, grad_norm=192.893, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.029, optim0_lr0=4.546e-04, train_time=1.189
+[bmi2:0/4] 2024-07-06 13:43:17,192 (trainer:779) INFO: 2epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.154, loss_ctc=100.524, loss_att=75.018, acc=0.596, loss=82.670, backward_time=0.252, grad_norm=182.370, clip=100.000, loss_scale=2.621e+05, optim_step_time=0.028, optim0_lr0=4.712e-04, train_time=1.173
+[bmi2:0/4] 2024-07-06 13:47:21,720 (trainer:779) INFO: 2epoch:train:3718-4130batch: iter_time=1.901e-04, forward_time=0.156, loss_ctc=99.729, loss_att=72.394, acc=0.627, loss=80.594, backward_time=0.255, grad_norm=168.477, clip=100.000, loss_scale=5.129e+05, optim_step_time=0.029, optim0_lr0=4.877e-04, train_time=1.183
+[bmi2:0/4] 2024-07-06 13:51:26,813 (trainer:779) INFO: 2epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.156, loss_ctc=94.836, loss_att=67.305, acc=0.644, loss=75.564, backward_time=0.255, grad_norm=183.770, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.042e-04, train_time=1.187
+[bmi2:0/4] 2024-07-06 13:55:31,371 (trainer:779) INFO: 2epoch:train:4544-4956batch: iter_time=0.001, forward_time=0.156, loss_ctc=91.249, loss_att=63.133, acc=0.666, loss=71.568, backward_time=0.254, grad_norm=173.414, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.207e-04, train_time=1.183
+[bmi2:0/4] 2024-07-06 13:59:37,289 (trainer:779) INFO: 2epoch:train:4957-5369batch: iter_time=5.232e-04, forward_time=0.158, loss_ctc=89.347, loss_att=60.611, acc=0.689, loss=69.231, backward_time=0.256, grad_norm=172.203, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.372e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 14:03:43,392 (trainer:779) INFO: 2epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.157, loss_ctc=80.441, loss_att=53.552, acc=0.695, loss=61.619, backward_time=0.255, grad_norm=165.063, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.538e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 14:07:48,855 (trainer:779) INFO: 2epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.156, loss_ctc=79.827, loss_att=52.547, acc=0.707, loss=60.731, backward_time=0.255, grad_norm=167.243, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.703e-04, train_time=1.189
+[bmi2:0/4] 2024-07-06 14:11:54,121 (trainer:779) INFO: 2epoch:train:6196-6608batch: iter_time=1.841e-04, forward_time=0.157, loss_ctc=81.312, loss_att=52.952, acc=0.723, loss=61.460, backward_time=0.255, grad_norm=163.220, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=5.868e-04, train_time=1.187
+[bmi2:0/4] 2024-07-06 14:15:59,869 (trainer:779) INFO: 2epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.157, loss_ctc=75.534, loss_att=48.557, acc=0.727, loss=56.650, backward_time=0.255, grad_norm=167.325, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=6.033e-04, train_time=1.190
+[bmi2:0/4] 2024-07-06 14:20:03,854 (trainer:779) INFO: 2epoch:train:7022-7434batch: iter_time=2.599e-04, forward_time=0.155, loss_ctc=73.690, loss_att=47.375, acc=0.734, loss=55.270, backward_time=0.253, grad_norm=173.186, clip=100.000, loss_scale=5.243e+05, optim_step_time=0.029, optim0_lr0=6.198e-04, train_time=1.181
+[bmi2:0/4] 2024-07-06 14:24:09,544 (trainer:779) INFO: 2epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.158, loss_ctc=74.673, loss_att=46.590, acc=0.751, loss=55.015, backward_time=0.255, grad_norm=166.994, clip=100.000, loss_scale=6.668e+05, optim_step_time=0.029, optim0_lr0=6.364e-04, train_time=1.190
+[bmi2:0/4] 2024-07-06 14:28:14,559 (trainer:779) INFO: 2epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.156, loss_ctc=71.129, loss_att=44.365, acc=0.753, loss=52.394, backward_time=0.255, grad_norm=167.263, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=6.529e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 14:30:03,589 (trainer:365) INFO: 2epoch results: [train] iter_time=0.002, forward_time=0.156, loss_ctc=101.478, loss_att=81.458, acc=0.581, loss=87.464, backward_time=0.254, grad_norm=169.807, clip=100.000, loss_scale=4.396e+05, optim_step_time=0.029, optim0_lr0=4.961e-04, train_time=1.193, time=1 hour, 22 minutes and 10.97 seconds, total_count=16534, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=31.342, cer_ctc=0.180, loss_att=19.826, acc=0.831, cer=0.118, wer=0.865, loss=23.281, time=44.36 seconds, total_count=68, gpu_max_cached_mem_GB=22.482, [att_plot] time=59.51 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 14:30:07,938 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 14:30:07,939 (trainer:299) INFO: 3/100epoch started. Estimated time to finish: 5 days, 17 hours and 13 minutes
+[bmi2:0/4] 2024-07-06 14:34:25,262 (trainer:779) INFO: 3epoch:train:1-413batch: iter_time=0.004, forward_time=0.158, loss_ctc=69.721, loss_att=42.643, acc=0.767, loss=50.766, backward_time=0.256, grad_norm=155.217, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.030, optim0_lr0=6.696e-04, train_time=1.247
+[bmi2:0/4] 2024-07-06 14:38:29,734 (trainer:779) INFO: 3epoch:train:414-826batch: iter_time=0.001, forward_time=0.156, loss_ctc=68.795, loss_att=41.761, acc=0.772, loss=49.871, backward_time=0.254, grad_norm=166.664, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=6.862e-04, train_time=1.183
+[bmi2:0/4] 2024-07-06 14:42:34,260 (trainer:779) INFO: 3epoch:train:827-1239batch: iter_time=0.002, forward_time=0.156, loss_ctc=66.575, loss_att=40.423, acc=0.773, loss=48.269, backward_time=0.254, grad_norm=164.808, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.030, optim0_lr0=7.027e-04, train_time=1.185
+[bmi2:0/4] 2024-07-06 14:46:42,172 (trainer:779) INFO: 3epoch:train:1240-1652batch: iter_time=0.006, forward_time=0.158, loss_ctc=64.301, loss_att=38.605, acc=0.776, loss=46.314, backward_time=0.255, grad_norm=166.969, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=7.192e-04, train_time=1.200
+[bmi2:0/4] 2024-07-06 14:50:47,645 (trainer:779) INFO: 3epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.156, loss_ctc=64.322, loss_att=38.418, acc=0.784, loss=46.189, backward_time=0.255, grad_norm=162.857, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=7.357e-04, train_time=1.189
+[bmi2:0/4] 2024-07-06 14:54:51,412 (trainer:779) INFO: 3epoch:train:2066-2478batch: iter_time=8.568e-04, forward_time=0.156, loss_ctc=63.236, loss_att=37.445, acc=0.789, loss=45.183, backward_time=0.254, grad_norm=149.951, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=7.522e-04, train_time=1.180
+[bmi2:0/4] 2024-07-06 14:58:57,113 (trainer:779) INFO: 3epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.157, loss_ctc=62.908, loss_att=37.339, acc=0.793, loss=45.010, backward_time=0.256, grad_norm=165.018, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=7.688e-04, train_time=1.190
+[bmi2:0/4] 2024-07-06 15:03:02,222 (trainer:779) INFO: 3epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.157, loss_ctc=60.245, loss_att=35.228, acc=0.796, loss=42.733, backward_time=0.255, grad_norm=153.076, clip=100.000, loss_scale=1.049e+06, optim_step_time=0.029, optim0_lr0=7.853e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 15:07:07,637 (trainer:779) INFO: 3epoch:train:3305-3717batch: iter_time=0.003, forward_time=0.156, loss_ctc=60.346, loss_att=35.096, acc=0.799, loss=42.671, backward_time=0.254, grad_norm=159.564, clip=100.000, loss_scale=1.680e+06, optim_step_time=0.029, optim0_lr0=8.018e-04, train_time=1.189
+[bmi2:0/4] 2024-07-06 15:11:13,877 (trainer:779) INFO: 3epoch:train:3718-4130batch: iter_time=0.004, forward_time=0.157, loss_ctc=59.283, loss_att=34.439, acc=0.802, loss=41.892, backward_time=0.255, grad_norm=155.214, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.030, optim0_lr0=8.183e-04, train_time=1.192
+[bmi2:0/4] 2024-07-06 15:15:18,731 (trainer:779) INFO: 3epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.157, loss_ctc=56.685, loss_att=32.738, acc=0.804, loss=39.922, backward_time=0.254, grad_norm=166.817, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=8.348e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 15:19:23,444 (trainer:779) INFO: 3epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.156, loss_ctc=57.837, loss_att=33.151, acc=0.813, loss=40.557, backward_time=0.255, grad_norm=147.692, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=8.514e-04, train_time=1.184
+[bmi2:0/4] 2024-07-06 15:23:28,361 (trainer:779) INFO: 3epoch:train:4957-5369batch: iter_time=1.904e-04, forward_time=0.156, loss_ctc=57.460, loss_att=33.970, acc=0.810, loss=41.017, backward_time=0.256, grad_norm=163.960, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=8.679e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 15:27:35,271 (trainer:779) INFO: 3epoch:train:5370-5782batch: iter_time=0.006, forward_time=0.157, loss_ctc=55.115, loss_att=31.486, acc=0.811, loss=38.575, backward_time=0.255, grad_norm=159.036, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=8.844e-04, train_time=1.195
+[bmi2:0/4] 2024-07-06 15:31:40,177 (trainer:779) INFO: 3epoch:train:5783-6195batch: iter_time=0.004, forward_time=0.155, loss_ctc=54.443, loss_att=30.911, acc=0.818, loss=37.970, backward_time=0.253, grad_norm=163.678, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=9.009e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 15:35:46,231 (trainer:779) INFO: 3epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.157, loss_ctc=55.292, loss_att=31.284, acc=0.824, loss=38.487, backward_time=0.255, grad_norm=159.487, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=9.174e-04, train_time=1.191
+[bmi2:0/4] 2024-07-06 15:39:52,542 (trainer:779) INFO: 3epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.157, loss_ctc=54.138, loss_att=30.504, acc=0.825, loss=37.594, backward_time=0.255, grad_norm=163.544, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.030, optim0_lr0=9.340e-04, train_time=1.193
+[bmi2:0/4] 2024-07-06 15:43:58,073 (trainer:779) INFO: 3epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.156, loss_ctc=52.803, loss_att=29.698, acc=0.825, loss=36.630, backward_time=0.255, grad_norm=161.883, clip=100.000, loss_scale=2.097e+06, optim_step_time=0.029, optim0_lr0=9.505e-04, train_time=1.188
+[bmi2:0/4] 2024-07-06 15:48:02,948 (trainer:779) INFO: 3epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.156, loss_ctc=52.041, loss_att=29.359, acc=0.825, loss=36.164, backward_time=0.254, grad_norm=169.079, clip=100.000, loss_scale=4.021e+06, optim_step_time=0.029, optim0_lr0=9.670e-04, train_time=1.186
+[bmi2:0/4] 2024-07-06 15:52:07,299 (trainer:779) INFO: 3epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.155, loss_ctc=51.641, loss_att=28.731, acc=0.828, loss=35.604, backward_time=0.254, grad_norm=156.422, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=9.835e-04, train_time=1.183
+[bmi2:0/4] 2024-07-06 15:53:54,666 (trainer:365) INFO: 3epoch results: [train] iter_time=0.003, forward_time=0.156, loss_ctc=59.334, loss_att=34.643, acc=0.802, loss=42.050, backward_time=0.255, grad_norm=160.507, clip=100.000, loss_scale=1.860e+06, optim_step_time=0.029, optim0_lr0=8.267e-04, train_time=1.191, time=1 hour, 22 minutes and 4.31 seconds, total_count=24801, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=22.126, cer_ctc=0.124, loss_att=13.139, acc=0.873, cer=0.079, wer=0.770, loss=15.835, time=44.05 seconds, total_count=102, gpu_max_cached_mem_GB=22.482, [att_plot] time=58.36 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 15:53:59,132 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 15:53:59,132 (trainer:299) INFO: 4/100epoch started. Estimated time to finish: 5 days, 15 hours and 44 minutes
+[bmi2:0/4] 2024-07-06 15:58:47,352 (trainer:779) INFO: 4epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=51.684, loss_att=28.838, acc=0.837, loss=35.692, backward_time=0.255, grad_norm=165.422, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.397
+[bmi2:0/4] 2024-07-06 16:02:51,641 (trainer:779) INFO: 4epoch:train:414-826batch: iter_time=5.057e-04, forward_time=0.157, loss_ctc=51.154, loss_att=28.163, acc=0.835, loss=35.060, backward_time=0.254, grad_norm=154.652, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-06 16:06:57,900 (trainer:779) INFO: 4epoch:train:827-1239batch: iter_time=0.006, forward_time=0.156, loss_ctc=48.690, loss_att=26.880, acc=0.833, loss=33.423, backward_time=0.255, grad_norm=161.468, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.193
+[bmi2:0/4] 2024-07-06 16:11:03,638 (trainer:779) INFO: 4epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.156, loss_ctc=49.326, loss_att=27.753, acc=0.834, loss=34.225, backward_time=0.255, grad_norm=159.481, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-06 16:15:07,908 (trainer:779) INFO: 4epoch:train:1653-2065batch: iter_time=0.003, forward_time=0.155, loss_ctc=48.690, loss_att=26.725, acc=0.838, loss=33.314, backward_time=0.254, grad_norm=168.923, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-06 16:19:12,726 (trainer:779) INFO: 4epoch:train:2066-2478batch: iter_time=0.003, forward_time=0.156, loss_ctc=48.135, loss_att=26.213, acc=0.842, loss=32.789, backward_time=0.254, grad_norm=152.052, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-06 16:23:18,688 (trainer:779) INFO: 4epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.157, loss_ctc=47.913, loss_att=25.937, acc=0.844, loss=32.530, backward_time=0.255, grad_norm=168.717, clip=100.000, loss_scale=4.194e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.192
+[bmi2:0/4] 2024-07-06 16:27:24,664 (trainer:779) INFO: 4epoch:train:2892-3304batch: iter_time=0.003, forward_time=0.157, loss_ctc=48.459, loss_att=26.332, acc=0.847, loss=32.970, backward_time=0.255, grad_norm=160.838, clip=100.000, loss_scale=5.228e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-06 16:31:31,193 (trainer:779) INFO: 4epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.157, loss_ctc=47.529, loss_att=25.832, acc=0.849, loss=32.341, backward_time=0.256, grad_norm=155.866, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.194
+[bmi2:0/4] 2024-07-06 16:35:36,263 (trainer:779) INFO: 4epoch:train:3718-4130batch: iter_time=0.003, forward_time=0.156, loss_ctc=45.506, loss_att=24.808, acc=0.844, loss=31.018, backward_time=0.254, grad_norm=164.067, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-06 16:39:40,377 (trainer:779) INFO: 4epoch:train:4131-4543batch: iter_time=7.592e-04, forward_time=0.156, loss_ctc=47.339, loss_att=25.588, acc=0.854, loss=32.113, backward_time=0.254, grad_norm=161.318, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-06 16:43:47,148 (trainer:779) INFO: 4epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.157, loss_ctc=46.720, loss_att=25.253, acc=0.852, loss=31.693, backward_time=0.256, grad_norm=174.110, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.194
+[bmi2:0/4] 2024-07-06 16:47:53,745 (trainer:779) INFO: 4epoch:train:4957-5369batch: iter_time=0.004, forward_time=0.157, loss_ctc=45.916, loss_att=24.682, acc=0.853, loss=31.052, backward_time=0.255, grad_norm=164.364, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.194
+[bmi2:0/4] 2024-07-06 16:51:57,614 (trainer:779) INFO: 4epoch:train:5370-5782batch: iter_time=1.852e-04, forward_time=0.156, loss_ctc=46.337, loss_att=24.991, acc=0.857, loss=31.395, backward_time=0.254, grad_norm=173.031, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-06 16:56:04,257 (trainer:779) INFO: 4epoch:train:5783-6195batch: iter_time=0.004, forward_time=0.156, loss_ctc=44.028, loss_att=23.685, acc=0.851, loss=29.788, backward_time=0.256, grad_norm=169.016, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.195
+[bmi2:0/4] 2024-07-06 17:00:08,008 (trainer:779) INFO: 4epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.155, loss_ctc=43.859, loss_att=23.567, acc=0.857, loss=29.655, backward_time=0.254, grad_norm=163.685, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-06 17:04:13,314 (trainer:779) INFO: 4epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.157, loss_ctc=44.013, loss_att=23.464, acc=0.859, loss=29.629, backward_time=0.255, grad_norm=168.126, clip=100.000, loss_scale=8.389e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-06 17:08:18,492 (trainer:779) INFO: 4epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.156, loss_ctc=44.940, loss_att=24.140, acc=0.862, loss=30.380, backward_time=0.255, grad_norm=161.037, clip=100.000, loss_scale=1.309e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-06 17:12:24,117 (trainer:779) INFO: 4epoch:train:7435-7847batch: iter_time=0.004, forward_time=0.156, loss_ctc=42.836, loss_att=23.289, acc=0.852, loss=29.153, backward_time=0.254, grad_norm=188.526, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-06 17:16:29,528 (trainer:779) INFO: 4epoch:train:7848-8260batch: iter_time=7.469e-04, forward_time=0.158, loss_ctc=43.827, loss_att=23.357, acc=0.864, loss=29.498, backward_time=0.254, grad_norm=166.953, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-06 17:18:18,021 (trainer:365) INFO: 4epoch results: [train] iter_time=0.002, forward_time=0.156, loss_ctc=46.836, loss_att=25.472, acc=0.848, loss=31.881, backward_time=0.255, grad_norm=165.101, clip=100.000, loss_scale=7.844e+06, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.198, time=1 hour, 22 minutes and 35.41 seconds, total_count=33068, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=18.644, cer_ctc=0.105, loss_att=10.671, acc=0.900, cer=0.068, wer=0.724, loss=13.063, time=43.5 seconds, total_count=136, gpu_max_cached_mem_GB=22.482, [att_plot] time=59.97 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 17:18:22,919 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 17:18:22,919 (trainer:299) INFO: 5/100epoch started. Estimated time to finish: 5 days, 14 hours and 30 minutes
+[bmi2:0/4] 2024-07-06 17:23:13,351 (trainer:779) INFO: 5epoch:train:1-413batch: iter_time=0.004, forward_time=0.156, loss_ctc=41.591, loss_att=22.087, acc=0.862, loss=27.938, backward_time=0.256, grad_norm=165.594, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.407
+[bmi2:0/4] 2024-07-06 17:27:18,220 (trainer:779) INFO: 5epoch:train:414-826batch: iter_time=0.002, forward_time=0.155, loss_ctc=42.873, loss_att=22.676, acc=0.869, loss=28.735, backward_time=0.254, grad_norm=178.409, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-06 17:31:23,097 (trainer:779) INFO: 5epoch:train:827-1239batch: iter_time=0.003, forward_time=0.156, loss_ctc=41.674, loss_att=22.084, acc=0.865, loss=27.961, backward_time=0.254, grad_norm=168.012, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-06 17:35:27,584 (trainer:779) INFO: 5epoch:train:1240-1652batch: iter_time=7.926e-04, forward_time=0.157, loss_ctc=42.460, loss_att=22.559, acc=0.867, loss=28.530, backward_time=0.254, grad_norm=170.583, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-06 17:39:32,100 (trainer:779) INFO: 5epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.156, loss_ctc=41.874, loss_att=22.322, acc=0.869, loss=28.188, backward_time=0.254, grad_norm=200.715, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-06 17:43:38,739 (trainer:779) INFO: 5epoch:train:2066-2478batch: iter_time=0.004, forward_time=0.155, loss_ctc=40.488, loss_att=21.415, acc=0.868, loss=27.137, backward_time=0.256, grad_norm=177.689, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.194
+[bmi2:0/4] 2024-07-06 17:47:43,492 (trainer:779) INFO: 5epoch:train:2479-2891batch: iter_time=0.005, forward_time=0.156, loss_ctc=39.614, loss_att=21.339, acc=0.861, loss=26.821, backward_time=0.254, grad_norm=177.812, clip=100.000, loss_scale=1.678e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-06 17:51:48,200 (trainer:779) INFO: 5epoch:train:2892-3304batch: iter_time=9.185e-04, forward_time=0.157, loss_ctc=40.228, loss_att=21.261, acc=0.868, loss=26.951, backward_time=0.255, grad_norm=167.973, clip=100.000, loss_scale=3.169e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-06 17:55:51,897 (trainer:779) INFO: 5epoch:train:3305-3717batch: iter_time=8.280e-04, forward_time=0.156, loss_ctc=40.919, loss_att=21.553, acc=0.874, loss=27.363, backward_time=0.254, grad_norm=167.050, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-06 17:59:58,948 (trainer:779) INFO: 5epoch:train:3718-4130batch: iter_time=0.005, forward_time=0.158, loss_ctc=40.021, loss_att=21.005, acc=0.870, loss=26.710, backward_time=0.255, grad_norm=183.386, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.196
+[bmi2:0/4] 2024-07-06 18:04:06,910 (trainer:779) INFO: 5epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.160, loss_ctc=39.561, loss_att=20.814, acc=0.870, loss=26.438, backward_time=0.256, grad_norm=174.221, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.031, optim0_lr0=0.001, train_time=1.201
+[bmi2:0/4] 2024-07-06 18:08:12,205 (trainer:779) INFO: 5epoch:train:4544-4956batch: iter_time=4.633e-04, forward_time=0.157, loss_ctc=39.472, loss_att=20.778, acc=0.872, loss=26.386, backward_time=0.255, grad_norm=172.080, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 18:12:18,248 (trainer:779) INFO: 5epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.158, loss_ctc=39.167, loss_att=20.693, acc=0.872, loss=26.235, backward_time=0.255, grad_norm=178.894, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 18:16:23,272 (trainer:779) INFO: 5epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.155, loss_ctc=39.771, loss_att=20.892, acc=0.874, loss=26.555, backward_time=0.254, grad_norm=176.613, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.186
+[bmi2:0/4] 2024-07-06 18:20:26,901 (trainer:779) INFO: 5epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.156, loss_ctc=39.489, loss_att=20.729, acc=0.876, loss=26.357, backward_time=0.254, grad_norm=182.771, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.180
+[bmi2:0/4] 2024-07-06 18:24:32,116 (trainer:779) INFO: 5epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.156, loss_ctc=38.546, loss_att=20.246, acc=0.874, loss=25.736, backward_time=0.255, grad_norm=184.634, clip=100.000, loss_scale=3.355e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 18:28:37,341 (trainer:779) INFO: 5epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.155, loss_ctc=39.280, loss_att=20.567, acc=0.880, loss=26.180, backward_time=0.256, grad_norm=179.434, clip=100.000, loss_scale=4.040e+07, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 18:32:42,375 (trainer:779) INFO: 5epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.156, loss_ctc=38.612, loss_att=20.265, acc=0.877, loss=25.769, backward_time=0.254, grad_norm=191.877, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.186
+[bmi2:0/4] 2024-07-06 18:36:48,116 (trainer:779) INFO: 5epoch:train:7435-7847batch: iter_time=5.366e-04, forward_time=0.157, loss_ctc=39.294, loss_att=20.674, acc=0.880, loss=26.260, backward_time=0.256, grad_norm=180.298, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 18:40:52,606 (trainer:779) INFO: 5epoch:train:7848-8260batch: iter_time=0.001, forward_time=0.156, loss_ctc=37.669, loss_att=19.699, acc=0.880, loss=25.090, backward_time=0.254, grad_norm=184.306, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.183
+[bmi2:0/4] 2024-07-06 18:42:34,123 (trainer:365) INFO: 5epoch results: [train] iter_time=0.002, forward_time=0.156, loss_ctc=40.125, loss_att=21.181, acc=0.871, loss=26.864, backward_time=0.255, grad_norm=178.088, clip=100.000, loss_scale=3.299e+07, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.198, time=1 hour, 22 minutes and 34.58 seconds, total_count=41335, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=16.167, cer_ctc=0.089, loss_att=9.564, acc=0.911, cer=0.059, wer=0.674, loss=11.545, time=41.35 seconds, total_count=170, gpu_max_cached_mem_GB=22.482, [att_plot] time=55.27 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 18:42:38,381 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 18:42:38,381 (trainer:299) INFO: 6/100epoch started. Estimated time to finish: 5 days, 13 hours and 10 minutes
+[bmi2:0/4] 2024-07-06 18:47:27,554 (trainer:779) INFO: 6epoch:train:1-413batch: iter_time=0.007, forward_time=0.155, loss_ctc=37.001, loss_att=19.283, acc=0.878, loss=24.598, backward_time=0.255, grad_norm=189.479, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.401
+[bmi2:0/4] 2024-07-06 18:51:33,166 (trainer:779) INFO: 6epoch:train:414-826batch: iter_time=0.002, forward_time=0.157, loss_ctc=37.606, loss_att=19.556, acc=0.881, loss=24.971, backward_time=0.255, grad_norm=187.595, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 18:55:39,432 (trainer:779) INFO: 6epoch:train:827-1239batch: iter_time=0.007, forward_time=0.157, loss_ctc=36.713, loss_att=19.258, acc=0.878, loss=24.495, backward_time=0.255, grad_norm=202.211, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-06 18:59:45,180 (trainer:779) INFO: 6epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.158, loss_ctc=37.379, loss_att=19.668, acc=0.880, loss=24.981, backward_time=0.255, grad_norm=192.929, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-06 19:03:50,116 (trainer:779) INFO: 6epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.158, loss_ctc=36.677, loss_att=19.089, acc=0.882, loss=24.365, backward_time=0.254, grad_norm=182.605, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 19:07:55,342 (trainer:779) INFO: 6epoch:train:2066-2478batch: iter_time=0.003, forward_time=0.156, loss_ctc=36.277, loss_att=18.889, acc=0.877, loss=24.106, backward_time=0.255, grad_norm=198.184, clip=100.000, loss_scale=6.711e+07, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 19:11:59,436 (trainer:779) INFO: 6epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.156, loss_ctc=36.959, loss_att=19.146, acc=0.886, loss=24.490, backward_time=0.254, grad_norm=188.322, clip=100.000, loss_scale=1.029e+08, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.182
+[bmi2:0/4] 2024-07-06 19:16:04,136 (trainer:779) INFO: 6epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.155, loss_ctc=36.380, loss_att=18.882, acc=0.884, loss=24.131, backward_time=0.255, grad_norm=187.121, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.184
+[bmi2:0/4] 2024-07-06 19:20:08,264 (trainer:779) INFO: 6epoch:train:3305-3717batch: iter_time=1.956e-04, forward_time=0.156, loss_ctc=36.925, loss_att=19.173, acc=0.886, loss=24.499, backward_time=0.255, grad_norm=189.162, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.182
+[bmi2:0/4] 2024-07-06 19:24:13,638 (trainer:779) INFO: 6epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.158, loss_ctc=36.770, loss_att=19.092, acc=0.886, loss=24.396, backward_time=0.255, grad_norm=186.119, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 19:28:20,159 (trainer:779) INFO: 6epoch:train:4131-4543batch: iter_time=0.005, forward_time=0.157, loss_ctc=36.456, loss_att=19.053, acc=0.884, loss=24.274, backward_time=0.256, grad_norm=213.808, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-06 19:32:24,923 (trainer:779) INFO: 6epoch:train:4544-4956batch: iter_time=0.004, forward_time=0.155, loss_ctc=36.121, loss_att=18.897, acc=0.882, loss=24.064, backward_time=0.254, grad_norm=210.003, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-06 19:36:29,457 (trainer:779) INFO: 6epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.157, loss_ctc=35.918, loss_att=18.638, acc=0.884, loss=23.822, backward_time=0.254, grad_norm=186.853, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-06 19:40:35,077 (trainer:779) INFO: 6epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.157, loss_ctc=35.916, loss_att=18.557, acc=0.885, loss=23.765, backward_time=0.254, grad_norm=189.931, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 19:44:41,712 (trainer:779) INFO: 6epoch:train:5783-6195batch: iter_time=0.005, forward_time=0.158, loss_ctc=35.581, loss_att=18.535, acc=0.881, loss=23.649, backward_time=0.254, grad_norm=198.802, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-06 19:48:46,942 (trainer:779) INFO: 6epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.156, loss_ctc=35.985, loss_att=18.808, acc=0.885, loss=23.961, backward_time=0.255, grad_norm=217.909, clip=100.000, loss_scale=1.342e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 19:52:52,885 (trainer:779) INFO: 6epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.158, loss_ctc=36.241, loss_att=18.688, acc=0.891, loss=23.954, backward_time=0.255, grad_norm=201.253, clip=100.000, loss_scale=2.482e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-06 19:56:55,855 (trainer:779) INFO: 6epoch:train:7022-7434batch: iter_time=2.890e-04, forward_time=0.155, loss_ctc=36.084, loss_att=18.614, acc=0.890, loss=23.855, backward_time=0.253, grad_norm=205.204, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.176
+[bmi2:0/4] 2024-07-06 20:00:59,972 (trainer:779) INFO: 6epoch:train:7435-7847batch: iter_time=3.262e-04, forward_time=0.157, loss_ctc=35.843, loss_att=18.530, acc=0.888, loss=23.724, backward_time=0.255, grad_norm=218.511, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.182
+[bmi2:0/4] 2024-07-06 20:05:06,713 (trainer:779) INFO: 6epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.157, loss_ctc=35.580, loss_att=18.312, acc=0.891, loss=23.492, backward_time=0.257, grad_norm=222.537, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-06 20:06:50,242 (trainer:365) INFO: 6epoch results: [train] iter_time=0.003, forward_time=0.156, loss_ctc=36.420, loss_att=18.934, acc=0.884, loss=24.179, backward_time=0.255, grad_norm=198.400, clip=100.000, loss_scale=1.385e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.198, time=1 hour, 22 minutes and 33.16 seconds, total_count=49602, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=15.426, cer_ctc=0.087, loss_att=9.699, acc=0.905, cer=0.058, wer=0.667, loss=11.417, time=41.97 seconds, total_count=204, gpu_max_cached_mem_GB=22.482, [att_plot] time=56.74 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 20:06:54,250 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-06 20:06:54,250 (trainer:299) INFO: 7/100epoch started. Estimated time to finish: 5 days, 11 hours and 48 minutes
+[bmi2:0/4] 2024-07-06 20:11:41,845 (trainer:779) INFO: 7epoch:train:1-413batch: iter_time=0.005, forward_time=0.155, loss_ctc=36.276, loss_att=19.464, acc=0.887, loss=24.507, backward_time=0.255, grad_norm=297.179, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.394
+[bmi2:0/4] 2024-07-06 20:15:47,619 (trainer:779) INFO: 7epoch:train:414-826batch: iter_time=0.002, forward_time=0.157, loss_ctc=34.818, loss_att=17.856, acc=0.889, loss=22.944, backward_time=0.255, grad_norm=196.512, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 20:19:53,692 (trainer:779) INFO: 7epoch:train:827-1239batch: iter_time=0.003, forward_time=0.157, loss_ctc=34.676, loss_att=18.556, acc=0.884, loss=23.392, backward_time=0.256, grad_norm=238.507, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 20:23:59,292 (trainer:779) INFO: 7epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.156, loss_ctc=34.181, loss_att=17.519, acc=0.892, loss=22.517, backward_time=0.255, grad_norm=195.227, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 20:28:03,465 (trainer:779) INFO: 7epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.156, loss_ctc=33.784, loss_att=17.264, acc=0.891, loss=22.220, backward_time=0.254, grad_norm=177.666, clip=100.000, loss_scale=2.684e+08, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.183
+[bmi2:0/4] 2024-07-06 20:32:07,945 (trainer:779) INFO: 7epoch:train:2066-2478batch: iter_time=0.003, forward_time=0.156, loss_ctc=33.778, loss_att=17.247, acc=0.893, loss=22.206, backward_time=0.254, grad_norm=181.304, clip=100.000, loss_scale=3.164e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.183
+[bmi2:0/4] 2024-07-06 20:36:12,892 (trainer:779) INFO: 7epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.156, loss_ctc=33.347, loss_att=17.029, acc=0.895, loss=21.924, backward_time=0.255, grad_norm=183.287, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 20:40:17,068 (trainer:779) INFO: 7epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.156, loss_ctc=33.329, loss_att=17.053, acc=0.895, loss=21.936, backward_time=0.254, grad_norm=179.227, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.182
+[bmi2:0/4] 2024-07-06 20:44:22,925 (trainer:779) INFO: 7epoch:train:3305-3717batch: iter_time=0.004, forward_time=0.156, loss_ctc=32.890, loss_att=16.846, acc=0.893, loss=21.660, backward_time=0.255, grad_norm=194.152, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-06 20:48:28,118 (trainer:779) INFO: 7epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.156, loss_ctc=33.888, loss_att=17.665, acc=0.894, loss=22.532, backward_time=0.255, grad_norm=210.264, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.187
+[bmi2:0/4] 2024-07-06 20:52:35,336 (trainer:779) INFO: 7epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.156, loss_ctc=32.325, loss_att=16.495, acc=0.894, loss=21.244, backward_time=0.257, grad_norm=181.977, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.198
+[bmi2:0/4] 2024-07-06 20:56:40,389 (trainer:779) INFO: 7epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.156, loss_ctc=32.846, loss_att=16.717, acc=0.898, loss=21.555, backward_time=0.254, grad_norm=183.854, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.186
+[bmi2:0/4] 2024-07-06 21:00:47,425 (trainer:779) INFO: 7epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.157, loss_ctc=32.316, loss_att=16.526, acc=0.895, loss=21.263, backward_time=0.256, grad_norm=228.663, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.197
+[bmi2:0/4] 2024-07-06 21:04:52,007 (trainer:779) INFO: 7epoch:train:5370-5782batch: iter_time=0.001, forward_time=0.155, loss_ctc=31.916, loss_att=16.197, acc=0.899, loss=20.913, backward_time=0.254, grad_norm=170.712, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.184
+[bmi2:0/4] 2024-07-06 21:08:56,456 (trainer:779) INFO: 7epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.156, loss_ctc=32.466, loss_att=16.537, acc=0.903, loss=21.316, backward_time=0.254, grad_norm=190.165, clip=100.000, loss_scale=5.369e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.184
+[bmi2:0/4] 2024-07-06 21:13:03,678 (trainer:779) INFO: 7epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.158, loss_ctc=32.176, loss_att=16.727, acc=0.902, loss=21.362, backward_time=0.256, grad_norm=232.982, clip=100.000, loss_scale=8.014e+08, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.196
+[bmi2:0/4] 2024-07-06 21:17:10,696 (trainer:779) INFO: 7epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.158, loss_ctc=31.513, loss_att=16.029, acc=0.900, loss=20.674, backward_time=0.256, grad_norm=199.412, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.197
+[bmi2:0/4] 2024-07-06 21:21:16,902 (trainer:779) INFO: 7epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.157, loss_ctc=31.235, loss_att=15.831, acc=0.903, loss=20.452, backward_time=0.256, grad_norm=173.818, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 21:25:22,232 (trainer:779) INFO: 7epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.157, loss_ctc=30.976, loss_att=15.666, acc=0.903, loss=20.259, backward_time=0.256, grad_norm=167.919, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 21:29:27,355 (trainer:779) INFO: 7epoch:train:7848-8260batch: iter_time=7.936e-04, forward_time=0.156, loss_ctc=31.213, loss_att=15.792, acc=0.901, loss=20.418, backward_time=0.255, grad_norm=175.486, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.186
+[bmi2:0/4] 2024-07-06 21:31:11,726 (trainer:365) INFO: 7epoch results: [train] iter_time=0.002, forward_time=0.156, loss_ctc=32.994, loss_att=16.948, acc=0.896, loss=21.761, backward_time=0.255, grad_norm=197.906, clip=100.000, loss_scale=5.797e+08, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.199, time=1 hour, 22 minutes and 38.3 seconds, total_count=57869, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=13.631, cer_ctc=0.075, loss_att=8.075, acc=0.922, cer=0.050, wer=0.621, loss=9.742, time=43.65 seconds, total_count=238, gpu_max_cached_mem_GB=22.482, [att_plot] time=55.53 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 21:31:15,823 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 21:31:15,823 (trainer:299) INFO: 8/100epoch started. Estimated time to finish: 5 days, 10 hours and 27 minutes
+[bmi2:0/4] 2024-07-06 21:36:06,537 (trainer:779) INFO: 8epoch:train:1-413batch: iter_time=0.004, forward_time=0.155, loss_ctc=30.485, loss_att=15.389, acc=0.906, loss=19.918, backward_time=0.256, grad_norm=190.766, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.409
+[bmi2:0/4] 2024-07-06 21:40:12,445 (trainer:779) INFO: 8epoch:train:414-826batch: iter_time=1.898e-04, forward_time=0.157, loss_ctc=30.727, loss_att=15.497, acc=0.907, loss=20.066, backward_time=0.256, grad_norm=207.713, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 21:44:17,679 (trainer:779) INFO: 8epoch:train:827-1239batch: iter_time=0.001, forward_time=0.157, loss_ctc=30.253, loss_att=15.200, acc=0.907, loss=19.716, backward_time=0.255, grad_norm=165.946, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 21:48:23,883 (trainer:779) INFO: 8epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.157, loss_ctc=29.932, loss_att=15.026, acc=0.907, loss=19.498, backward_time=0.256, grad_norm=172.799, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 21:52:29,673 (trainer:779) INFO: 8epoch:train:1653-2065batch: iter_time=9.812e-04, forward_time=0.157, loss_ctc=30.878, loss_att=16.063, acc=0.906, loss=20.507, backward_time=0.255, grad_norm=202.414, clip=100.000, loss_scale=1.074e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 21:56:36,014 (trainer:779) INFO: 8epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.157, loss_ctc=29.733, loss_att=15.017, acc=0.907, loss=19.432, backward_time=0.255, grad_norm=178.893, clip=100.000, loss_scale=1.956e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 22:00:41,536 (trainer:779) INFO: 8epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.157, loss_ctc=29.441, loss_att=14.767, acc=0.909, loss=19.169, backward_time=0.255, grad_norm=186.772, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 22:04:47,434 (trainer:779) INFO: 8epoch:train:2892-3304batch: iter_time=0.003, forward_time=0.156, loss_ctc=29.060, loss_att=14.569, acc=0.905, loss=18.916, backward_time=0.255, grad_norm=165.448, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 22:08:50,819 (trainer:779) INFO: 8epoch:train:3305-3717batch: iter_time=4.480e-04, forward_time=0.155, loss_ctc=29.176, loss_att=14.604, acc=0.908, loss=18.976, backward_time=0.254, grad_norm=171.855, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.179
+[bmi2:0/4] 2024-07-06 22:12:56,850 (trainer:779) INFO: 8epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.156, loss_ctc=29.724, loss_att=14.929, acc=0.909, loss=19.367, backward_time=0.255, grad_norm=200.949, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-06 22:17:01,481 (trainer:779) INFO: 8epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.156, loss_ctc=29.024, loss_att=14.543, acc=0.910, loss=18.887, backward_time=0.255, grad_norm=161.396, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-06 22:21:05,936 (trainer:779) INFO: 8epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.156, loss_ctc=28.774, loss_att=14.472, acc=0.908, loss=18.763, backward_time=0.254, grad_norm=172.837, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.183
+[bmi2:0/4] 2024-07-06 22:25:11,463 (trainer:779) INFO: 8epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.156, loss_ctc=28.606, loss_att=14.280, acc=0.911, loss=18.578, backward_time=0.256, grad_norm=166.102, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 22:29:17,183 (trainer:779) INFO: 8epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.157, loss_ctc=28.690, loss_att=14.593, acc=0.907, loss=18.822, backward_time=0.255, grad_norm=184.434, clip=100.000, loss_scale=2.147e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-06 22:33:22,321 (trainer:779) INFO: 8epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.157, loss_ctc=29.019, loss_att=14.442, acc=0.914, loss=18.815, backward_time=0.255, grad_norm=155.239, clip=100.000, loss_scale=2.439e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 22:37:28,698 (trainer:779) INFO: 8epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.157, loss_ctc=28.545, loss_att=14.278, acc=0.912, loss=18.558, backward_time=0.256, grad_norm=165.693, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-06 22:41:34,378 (trainer:779) INFO: 8epoch:train:6609-7021batch: iter_time=0.004, forward_time=0.155, loss_ctc=27.910, loss_att=13.916, acc=0.910, loss=18.114, backward_time=0.253, grad_norm=159.087, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 22:45:39,895 (trainer:779) INFO: 8epoch:train:7022-7434batch: iter_time=0.003, forward_time=0.156, loss_ctc=27.706, loss_att=13.870, acc=0.910, loss=18.021, backward_time=0.255, grad_norm=155.408, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 22:49:45,447 (trainer:779) INFO: 8epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.156, loss_ctc=27.521, loss_att=13.781, acc=0.911, loss=17.903, backward_time=0.254, grad_norm=148.691, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.190
+[bmi2:0/4] 2024-07-06 22:53:50,855 (trainer:779) INFO: 8epoch:train:7848-8260batch: iter_time=0.001, forward_time=0.157, loss_ctc=28.121, loss_att=14.041, acc=0.916, loss=18.265, backward_time=0.255, grad_norm=161.806, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 22:55:40,209 (trainer:365) INFO: 8epoch results: [train] iter_time=0.002, forward_time=0.156, loss_ctc=29.158, loss_att=14.659, acc=0.909, loss=19.009, backward_time=0.255, grad_norm=173.703, clip=100.000, loss_scale=2.423e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.200, time=1 hour, 22 minutes and 40.27 seconds, total_count=66136, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=12.778, cer_ctc=0.069, loss_att=7.552, acc=0.928, cer=0.046, wer=0.598, loss=9.120, time=44.07 seconds, total_count=272, gpu_max_cached_mem_GB=22.482, [att_plot] time=1 minute and 0.05 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-06 22:55:45,177 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-06 22:55:45,178 (trainer:299) INFO: 9/100epoch started. Estimated time to finish: 5 days, 9 hours and 6 minutes
+[bmi2:0/4] 2024-07-06 23:00:34,653 (trainer:779) INFO: 9epoch:train:1-413batch: iter_time=0.004, forward_time=0.155, loss_ctc=27.061, loss_att=13.384, acc=0.914, loss=17.487, backward_time=0.254, grad_norm=160.983, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.403
+[bmi2:0/4] 2024-07-06 23:04:39,604 (trainer:779) INFO: 9epoch:train:414-826batch: iter_time=0.002, forward_time=0.157, loss_ctc=27.376, loss_att=13.451, acc=0.915, loss=17.628, backward_time=0.254, grad_norm=152.161, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-06 23:08:44,982 (trainer:779) INFO: 9epoch:train:827-1239batch: iter_time=0.004, forward_time=0.156, loss_ctc=27.602, loss_att=13.645, acc=0.919, loss=17.832, backward_time=0.254, grad_norm=153.168, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-06 23:12:51,862 (trainer:779) INFO: 9epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.155, loss_ctc=27.000, loss_att=13.321, acc=0.916, loss=17.425, backward_time=0.257, grad_norm=160.134, clip=100.000, loss_scale=4.295e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-06 23:16:59,315 (trainer:779) INFO: 9epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.158, loss_ctc=26.347, loss_att=13.000, acc=0.916, loss=17.004, backward_time=0.256, grad_norm=154.353, clip=100.000, loss_scale=6.297e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.199
+[bmi2:0/4] 2024-07-06 23:21:06,019 (trainer:779) INFO: 9epoch:train:2066-2478batch: iter_time=0.003, forward_time=0.156, loss_ctc=26.716, loss_att=13.196, acc=0.914, loss=17.252, backward_time=0.255, grad_norm=155.894, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-06 23:25:12,340 (trainer:779) INFO: 9epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.158, loss_ctc=27.034, loss_att=13.283, acc=0.918, loss=17.409, backward_time=0.256, grad_norm=158.834, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-06 23:29:20,176 (trainer:779) INFO: 9epoch:train:2892-3304batch: iter_time=0.003, forward_time=0.157, loss_ctc=26.651, loss_att=13.172, acc=0.916, loss=17.215, backward_time=0.256, grad_norm=154.767, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.199
+[bmi2:0/4] 2024-07-06 23:33:26,528 (trainer:779) INFO: 9epoch:train:3305-3717batch: iter_time=5.014e-04, forward_time=0.159, loss_ctc=26.767, loss_att=13.267, acc=0.917, loss=17.317, backward_time=0.256, grad_norm=168.190, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-06 23:37:34,245 (trainer:779) INFO: 9epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.158, loss_ctc=26.625, loss_att=13.150, acc=0.918, loss=17.193, backward_time=0.256, grad_norm=150.584, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.199
+[bmi2:0/4] 2024-07-06 23:41:40,822 (trainer:779) INFO: 9epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.158, loss_ctc=26.639, loss_att=13.184, acc=0.917, loss=17.221, backward_time=0.255, grad_norm=159.481, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-06 23:45:47,650 (trainer:779) INFO: 9epoch:train:4544-4956batch: iter_time=7.318e-04, forward_time=0.158, loss_ctc=27.190, loss_att=13.447, acc=0.922, loss=17.570, backward_time=0.255, grad_norm=156.031, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-06 23:49:55,365 (trainer:779) INFO: 9epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.160, loss_ctc=26.532, loss_att=13.019, acc=0.918, loss=17.073, backward_time=0.256, grad_norm=151.477, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.200
+[bmi2:0/4] 2024-07-06 23:54:02,048 (trainer:779) INFO: 9epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.159, loss_ctc=26.293, loss_att=12.928, acc=0.920, loss=16.937, backward_time=0.255, grad_norm=149.776, clip=100.000, loss_scale=8.590e+09, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-06 23:58:05,741 (trainer:779) INFO: 9epoch:train:5783-6195batch: iter_time=0.003, forward_time=0.155, loss_ctc=26.757, loss_att=13.169, acc=0.923, loss=17.245, backward_time=0.253, grad_norm=148.494, clip=100.000, loss_scale=1.530e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.181
+[bmi2:0/4] 2024-07-07 00:02:10,654 (trainer:779) INFO: 9epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.157, loss_ctc=26.073, loss_att=12.812, acc=0.918, loss=16.790, backward_time=0.255, grad_norm=144.685, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-07 00:06:17,063 (trainer:779) INFO: 9epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.158, loss_ctc=26.037, loss_att=12.775, acc=0.920, loss=16.753, backward_time=0.255, grad_norm=148.120, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-07 00:10:20,826 (trainer:779) INFO: 9epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.156, loss_ctc=25.832, loss_att=12.720, acc=0.919, loss=16.654, backward_time=0.254, grad_norm=149.717, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.180
+[bmi2:0/4] 2024-07-07 00:14:26,044 (trainer:779) INFO: 9epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.158, loss_ctc=26.092, loss_att=12.819, acc=0.924, loss=16.801, backward_time=0.255, grad_norm=173.638, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-07 00:18:30,468 (trainer:779) INFO: 9epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.156, loss_ctc=25.837, loss_att=12.763, acc=0.924, loss=16.685, backward_time=0.254, grad_norm=159.063, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.183
+[bmi2:0/4] 2024-07-07 00:20:20,490 (trainer:365) INFO: 9epoch results: [train] iter_time=0.002, forward_time=0.157, loss_ctc=26.624, loss_att=13.126, acc=0.918, loss=17.175, backward_time=0.255, grad_norm=155.516, clip=100.000, loss_scale=1.011e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.202, time=1 hour, 22 minutes and 50.47 seconds, total_count=74403, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=12.058, cer_ctc=0.065, loss_att=6.867, acc=0.933, cer=0.042, wer=0.582, loss=8.424, time=48.22 seconds, total_count=306, gpu_max_cached_mem_GB=22.482, [att_plot] time=56.62 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 00:20:24,935 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 00:20:24,936 (trainer:299) INFO: 10/100epoch started. Estimated time to finish: 5 days, 7 hours and 47 minutes
+[bmi2:0/4] 2024-07-07 00:25:12,899 (trainer:779) INFO: 10epoch:train:1-413batch: iter_time=0.002, forward_time=0.157, loss_ctc=25.227, loss_att=12.241, acc=0.924, loss=16.137, backward_time=0.253, grad_norm=140.035, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.395
+[bmi2:0/4] 2024-07-07 00:29:19,066 (trainer:779) INFO: 10epoch:train:414-826batch: iter_time=0.002, forward_time=0.158, loss_ctc=25.770, loss_att=12.552, acc=0.925, loss=16.517, backward_time=0.256, grad_norm=154.501, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-07 00:33:26,156 (trainer:779) INFO: 10epoch:train:827-1239batch: iter_time=0.002, forward_time=0.159, loss_ctc=25.079, loss_att=12.226, acc=0.924, loss=16.082, backward_time=0.257, grad_norm=140.448, clip=100.000, loss_scale=1.718e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.197
+[bmi2:0/4] 2024-07-07 00:37:32,855 (trainer:779) INFO: 10epoch:train:1240-1652batch: iter_time=9.459e-04, forward_time=0.159, loss_ctc=25.862, loss_att=12.535, acc=0.928, loss=16.533, backward_time=0.256, grad_norm=142.474, clip=100.000, loss_scale=1.909e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-07 00:41:39,745 (trainer:779) INFO: 10epoch:train:1653-2065batch: iter_time=0.003, forward_time=0.158, loss_ctc=24.684, loss_att=12.020, acc=0.922, loss=15.819, backward_time=0.255, grad_norm=146.715, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.196
+[bmi2:0/4] 2024-07-07 00:45:44,531 (trainer:779) INFO: 10epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.155, loss_ctc=24.457, loss_att=11.930, acc=0.923, loss=15.688, backward_time=0.253, grad_norm=137.834, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-07 00:49:55,889 (trainer:779) INFO: 10epoch:train:2479-2891batch: iter_time=0.005, forward_time=0.160, loss_ctc=24.080, loss_att=11.756, acc=0.919, loss=15.453, backward_time=0.258, grad_norm=138.070, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.218
+[bmi2:0/4] 2024-07-07 00:54:04,631 (trainer:779) INFO: 10epoch:train:2892-3304batch: iter_time=2.371e-04, forward_time=0.159, loss_ctc=24.978, loss_att=12.234, acc=0.927, loss=16.057, backward_time=0.258, grad_norm=151.665, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.204
+[bmi2:0/4] 2024-07-07 00:58:11,121 (trainer:779) INFO: 10epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.159, loss_ctc=24.758, loss_att=12.051, acc=0.927, loss=15.863, backward_time=0.255, grad_norm=143.904, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-07 01:02:17,961 (trainer:779) INFO: 10epoch:train:3718-4130batch: iter_time=0.003, forward_time=0.158, loss_ctc=24.966, loss_att=12.120, acc=0.926, loss=15.974, backward_time=0.255, grad_norm=143.843, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.194
+[bmi2:0/4] 2024-07-07 01:06:24,694 (trainer:779) INFO: 10epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.158, loss_ctc=25.150, loss_att=12.170, acc=0.924, loss=16.064, backward_time=0.255, grad_norm=150.245, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-07 01:10:32,453 (trainer:779) INFO: 10epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.161, loss_ctc=24.652, loss_att=11.969, acc=0.924, loss=15.774, backward_time=0.256, grad_norm=148.938, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.199
+[bmi2:0/4] 2024-07-07 01:14:37,651 (trainer:779) INFO: 10epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.157, loss_ctc=24.201, loss_att=11.732, acc=0.926, loss=15.473, backward_time=0.255, grad_norm=137.943, clip=100.000, loss_scale=3.436e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-07 01:18:44,583 (trainer:779) INFO: 10epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.159, loss_ctc=24.494, loss_att=11.875, acc=0.925, loss=15.661, backward_time=0.255, grad_norm=139.357, clip=100.000, loss_scale=4.897e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-07 01:22:56,492 (trainer:779) INFO: 10epoch:train:5783-6195batch: iter_time=6.099e-04, forward_time=0.162, loss_ctc=24.295, loss_att=11.753, acc=0.927, loss=15.516, backward_time=0.256, grad_norm=146.646, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.220
+[bmi2:0/4] 2024-07-07 01:27:02,128 (trainer:779) INFO: 10epoch:train:6196-6608batch: iter_time=0.003, forward_time=0.157, loss_ctc=23.783, loss_att=11.612, acc=0.922, loss=15.263, backward_time=0.255, grad_norm=142.290, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-07 01:31:08,170 (trainer:779) INFO: 10epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.159, loss_ctc=24.555, loss_att=11.945, acc=0.927, loss=15.728, backward_time=0.254, grad_norm=151.614, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.033, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-07 01:35:14,733 (trainer:779) INFO: 10epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.157, loss_ctc=24.368, loss_att=11.834, acc=0.927, loss=15.594, backward_time=0.254, grad_norm=129.799, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.032, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-07 01:39:20,909 (trainer:779) INFO: 10epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.158, loss_ctc=24.750, loss_att=12.165, acc=0.928, loss=15.941, backward_time=0.255, grad_norm=174.555, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.032, optim0_lr0=0.002, train_time=1.192
+[bmi2:0/4] 2024-07-07 01:43:26,368 (trainer:779) INFO: 10epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.158, loss_ctc=24.595, loss_att=11.959, acc=0.928, loss=15.750, backward_time=0.255, grad_norm=147.459, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.032, optim0_lr0=0.002, train_time=1.188
+[bmi2:0/4] 2024-07-07 01:45:09,930 (trainer:365) INFO: 10epoch results: [train] iter_time=0.002, forward_time=0.158, loss_ctc=24.728, loss_att=12.030, acc=0.925, loss=15.839, backward_time=0.255, grad_norm=145.381, clip=100.000, loss_scale=4.208e+10, optim_step_time=0.031, optim0_lr0=0.002, train_time=1.206, time=1 hour, 23 minutes and 6.48 seconds, total_count=82670, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=11.566, cer_ctc=0.061, loss_att=6.592, acc=0.936, cer=0.041, wer=0.565, loss=8.084, time=45.91 seconds, total_count=340, gpu_max_cached_mem_GB=22.482, [att_plot] time=52.6 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 01:45:14,154 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 01:45:14,155 (trainer:299) INFO: 11/100epoch started. Estimated time to finish: 5 days, 6 hours and 28 minutes
+[bmi2:0/4] 2024-07-07 01:49:29,489 (trainer:779) INFO: 11epoch:train:1-413batch: iter_time=0.003, forward_time=0.157, loss_ctc=23.175, loss_att=11.151, acc=0.929, loss=14.758, backward_time=0.254, grad_norm=130.221, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.032, optim0_lr0=0.002, train_time=1.237
+[bmi2:0/4] 2024-07-07 01:53:36,044 (trainer:779) INFO: 11epoch:train:414-826batch: iter_time=0.004, forward_time=0.158, loss_ctc=23.268, loss_att=11.165, acc=0.930, loss=14.796, backward_time=0.255, grad_norm=132.351, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.032, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-07 01:57:42,278 (trainer:779) INFO: 11epoch:train:827-1239batch: iter_time=0.002, forward_time=0.159, loss_ctc=23.763, loss_att=11.519, acc=0.931, loss=15.192, backward_time=0.255, grad_norm=151.212, clip=100.000, loss_scale=6.872e+10, optim_step_time=0.030, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-07 02:01:46,919 (trainer:779) INFO: 11epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.156, loss_ctc=23.421, loss_att=11.268, acc=0.930, loss=14.914, backward_time=0.254, grad_norm=132.959, clip=100.000, loss_scale=1.205e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.184
+[bmi2:0/4] 2024-07-07 02:05:53,481 (trainer:779) INFO: 11epoch:train:1653-2065batch: iter_time=0.006, forward_time=0.158, loss_ctc=23.261, loss_att=11.221, acc=0.929, loss=14.833, backward_time=0.255, grad_norm=138.764, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.195
+[bmi2:0/4] 2024-07-07 02:09:57,602 (trainer:779) INFO: 11epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.156, loss_ctc=23.555, loss_att=11.343, acc=0.933, loss=15.006, backward_time=0.253, grad_norm=129.643, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.181
+[bmi2:0/4] 2024-07-07 02:14:03,000 (trainer:779) INFO: 11epoch:train:2479-2891batch: iter_time=0.003, forward_time=0.157, loss_ctc=23.077, loss_att=11.257, acc=0.931, loss=14.803, backward_time=0.254, grad_norm=147.503, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-07 02:18:09,149 (trainer:779) INFO: 11epoch:train:2892-3304batch: iter_time=0.004, forward_time=0.157, loss_ctc=22.396, loss_att=10.847, acc=0.927, loss=14.312, backward_time=0.254, grad_norm=131.411, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-07 02:22:14,913 (trainer:779) INFO: 11epoch:train:3305-3717batch: iter_time=0.006, forward_time=0.158, loss_ctc=22.755, loss_att=11.055, acc=0.930, loss=14.565, backward_time=0.254, grad_norm=124.999, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191
+[bmi2:0/4] 2024-07-07 02:26:21,466 (trainer:779) INFO: 11epoch:train:3718-4130batch: iter_time=0.007, forward_time=0.156, loss_ctc=22.556, loss_att=10.911, acc=0.928, loss=14.405, backward_time=0.253, grad_norm=135.999, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.193
+[bmi2:0/4] 2024-07-07 02:30:26,059 (trainer:779) INFO: 11epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.157, loss_ctc=23.327, loss_att=11.198, acc=0.931, loss=14.837, backward_time=0.254, grad_norm=142.633, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.185
+[bmi2:0/4] 2024-07-07 02:34:30,592 (trainer:779) INFO: 11epoch:train:4544-4956batch: iter_time=0.003, forward_time=0.156, loss_ctc=23.428, loss_att=11.333, acc=0.930, loss=14.961, backward_time=0.253, grad_norm=165.290, clip=100.000, loss_scale=1.374e+11, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.184
+[bmi2:0/4] 2024-07-07 02:38:34,523 (trainer:779) INFO: 11epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.156, loss_ctc=23.005, loss_att=11.050, acc=0.931, loss=14.636, backward_time=0.253, grad_norm=127.651, clip=100.000, loss_scale=1.468e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.182
+[bmi2:0/4] 2024-07-07 02:42:40,310 (trainer:779) INFO: 11epoch:train:5370-5782batch: iter_time=0.001, forward_time=0.159, loss_ctc=22.843, loss_att=10.984, acc=0.933, loss=14.541, backward_time=0.254, grad_norm=132.846, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.189
+[bmi2:0/4] 2024-07-07 02:46:43,835 (trainer:779) INFO: 11epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.156, loss_ctc=22.798, loss_att=10.999, acc=0.933, loss=14.539, backward_time=0.253, grad_norm=129.824, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.028, optim0_lr0=0.002, train_time=1.180
+[bmi2:0/4] 2024-07-07 02:50:48,698 (trainer:779) INFO: 11epoch:train:6196-6608batch: iter_time=0.003, forward_time=0.157, loss_ctc=23.012, loss_att=11.043, acc=0.934, loss=14.634, backward_time=0.254, grad_norm=133.199, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 02:54:55,459 (trainer:779) INFO: 11epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.158, loss_ctc=23.065, loss_att=11.060, acc=0.935, loss=14.662, backward_time=0.256, grad_norm=134.783, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.195
+[bmi2:0/4] 2024-07-07 02:59:00,759 (trainer:779) INFO: 11epoch:train:7022-7434batch: iter_time=0.004, forward_time=0.158, loss_ctc=22.225, loss_att=10.695, acc=0.930, loss=14.154, backward_time=0.253, grad_norm=133.481, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 03:03:08,884 (trainer:779) INFO: 11epoch:train:7435-7847batch: iter_time=0.009, forward_time=0.158, loss_ctc=22.220, loss_att=10.716, acc=0.929, loss=14.167, backward_time=0.254, grad_norm=133.010, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.202
+[bmi2:0/4] 2024-07-07 03:07:13,342 (trainer:779) INFO: 11epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.157, loss_ctc=22.981, loss_att=11.100, acc=0.934, loss=14.665, backward_time=0.253, grad_norm=136.325, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 03:08:57,947 (trainer:365) INFO: 11epoch results: [train] iter_time=0.004, forward_time=0.157, loss_ctc=23.002, loss_att=11.094, acc=0.931, loss=14.666, backward_time=0.254, grad_norm=136.203, clip=100.000, loss_scale=1.749e+11, optim_step_time=0.029, optim0_lr0=0.002, train_time=1.191, time=1 hour, 22 minutes and 4.23 seconds, total_count=90937, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=11.029, cer_ctc=0.058, loss_att=6.298, acc=0.938, cer=0.040, wer=0.558, loss=7.717, time=46.02 seconds, total_count=374, gpu_max_cached_mem_GB=22.482, [att_plot] time=53.54 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 03:09:01,751 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 03:09:01,752 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/1epoch.pth
+[bmi2:0/4] 2024-07-07 03:09:01,752 (trainer:299) INFO: 12/100epoch started. Estimated time to finish: 5 days, 4 hours and 59 minutes
+[bmi2:0/4] 2024-07-07 03:13:15,762 (trainer:779) INFO: 12epoch:train:1-413batch: iter_time=0.003, forward_time=0.154, loss_ctc=22.048, loss_att=10.497, acc=0.939, loss=13.962, backward_time=0.253, grad_norm=130.928, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.230
+[bmi2:0/4] 2024-07-07 03:17:20,500 (trainer:779) INFO: 12epoch:train:414-826batch: iter_time=0.003, forward_time=0.156, loss_ctc=21.988, loss_att=10.521, acc=0.932, loss=13.961, backward_time=0.254, grad_norm=132.939, clip=100.000, loss_scale=2.749e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 03:21:24,932 (trainer:779) INFO: 12epoch:train:827-1239batch: iter_time=0.002, forward_time=0.156, loss_ctc=21.801, loss_att=10.429, acc=0.931, loss=13.840, backward_time=0.253, grad_norm=129.589, clip=100.000, loss_scale=3.843e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 03:25:29,637 (trainer:779) INFO: 12epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.157, loss_ctc=22.158, loss_att=10.656, acc=0.936, loss=14.106, backward_time=0.254, grad_norm=136.449, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 03:29:34,569 (trainer:779) INFO: 12epoch:train:1653-2065batch: iter_time=0.003, forward_time=0.157, loss_ctc=21.810, loss_att=10.421, acc=0.933, loss=13.838, backward_time=0.254, grad_norm=125.601, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 03:33:41,664 (trainer:779) INFO: 12epoch:train:2066-2478batch: iter_time=0.008, forward_time=0.157, loss_ctc=21.339, loss_att=10.245, acc=0.930, loss=13.573, backward_time=0.254, grad_norm=130.360, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.196
+[bmi2:0/4] 2024-07-07 03:37:47,113 (trainer:779) INFO: 12epoch:train:2479-2891batch: iter_time=0.003, forward_time=0.157, loss_ctc=21.960, loss_att=10.497, acc=0.934, loss=13.936, backward_time=0.254, grad_norm=127.693, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 03:41:54,435 (trainer:779) INFO: 12epoch:train:2892-3304batch: iter_time=0.006, forward_time=0.156, loss_ctc=21.577, loss_att=10.302, acc=0.932, loss=13.685, backward_time=0.255, grad_norm=127.534, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.197
+[bmi2:0/4] 2024-07-07 03:45:57,296 (trainer:779) INFO: 12epoch:train:3305-3717batch: iter_time=6.138e-04, forward_time=0.155, loss_ctc=21.784, loss_att=10.411, acc=0.937, loss=13.823, backward_time=0.253, grad_norm=120.354, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 03:50:02,448 (trainer:779) INFO: 12epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.157, loss_ctc=21.789, loss_att=10.450, acc=0.938, loss=13.852, backward_time=0.255, grad_norm=126.600, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 03:54:06,725 (trainer:779) INFO: 12epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.157, loss_ctc=21.803, loss_att=10.434, acc=0.937, loss=13.845, backward_time=0.254, grad_norm=123.759, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 03:58:11,266 (trainer:779) INFO: 12epoch:train:4544-4956batch: iter_time=0.004, forward_time=0.155, loss_ctc=21.366, loss_att=10.224, acc=0.934, loss=13.567, backward_time=0.253, grad_norm=126.142, clip=100.000, loss_scale=5.498e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 04:02:16,536 (trainer:779) INFO: 12epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.158, loss_ctc=21.620, loss_att=10.296, acc=0.939, loss=13.693, backward_time=0.254, grad_norm=129.818, clip=100.000, loss_scale=9.421e+11, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 04:06:22,103 (trainer:779) INFO: 12epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.157, loss_ctc=21.651, loss_att=10.314, acc=0.936, loss=13.715, backward_time=0.254, grad_norm=126.768, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 04:10:27,571 (trainer:779) INFO: 12epoch:train:5783-6195batch: iter_time=0.003, forward_time=0.157, loss_ctc=21.810, loss_att=10.449, acc=0.937, loss=13.858, backward_time=0.254, grad_norm=127.110, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 04:14:32,895 (trainer:779) INFO: 12epoch:train:6196-6608batch: iter_time=4.637e-04, forward_time=0.158, loss_ctc=21.844, loss_att=10.472, acc=0.940, loss=13.883, backward_time=0.255, grad_norm=126.132, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 04:18:38,907 (trainer:779) INFO: 12epoch:train:6609-7021batch: iter_time=0.006, forward_time=0.156, loss_ctc=21.293, loss_att=10.300, acc=0.931, loss=13.598, backward_time=0.253, grad_norm=127.836, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.192
+[bmi2:0/4] 2024-07-07 04:22:43,535 (trainer:779) INFO: 12epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.157, loss_ctc=21.260, loss_att=10.152, acc=0.939, loss=13.485, backward_time=0.254, grad_norm=125.136, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 04:26:49,715 (trainer:779) INFO: 12epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.157, loss_ctc=21.545, loss_att=10.304, acc=0.939, loss=13.677, backward_time=0.255, grad_norm=120.706, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.193
+[bmi2:0/4] 2024-07-07 04:30:55,646 (trainer:779) INFO: 12epoch:train:7848-8260batch: iter_time=0.004, forward_time=0.157, loss_ctc=21.082, loss_att=10.067, acc=0.936, loss=13.372, backward_time=0.254, grad_norm=127.065, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 04:32:24,433 (trainer:365) INFO: 12epoch results: [train] iter_time=0.003, forward_time=0.157, loss_ctc=21.670, loss_att=10.370, acc=0.935, loss=13.760, backward_time=0.254, grad_norm=127.460, clip=100.000, loss_scale=7.263e+11, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.190, time=1 hour, 21 minutes and 59.22 seconds, total_count=99204, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=11.039, cer_ctc=0.059, loss_att=6.610, acc=0.936, cer=0.041, wer=0.552, loss=7.939, time=44.54 seconds, total_count=408, gpu_max_cached_mem_GB=22.482, [att_plot] time=38.92 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 04:32:28,880 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 04:32:28,881 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/2epoch.pth
+[bmi2:0/4] 2024-07-07 04:32:28,881 (trainer:299) INFO: 13/100epoch started. Estimated time to finish: 5 days, 3 hours and 29 minutes
+[bmi2:0/4] 2024-07-07 04:36:41,306 (trainer:779) INFO: 13epoch:train:1-413batch: iter_time=0.005, forward_time=0.154, loss_ctc=21.260, loss_att=10.095, acc=0.937, loss=13.445, backward_time=0.254, grad_norm=126.264, clip=100.000, loss_scale=1.100e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.223
+[bmi2:0/4] 2024-07-07 04:40:46,668 (trainer:779) INFO: 13epoch:train:414-826batch: iter_time=0.003, forward_time=0.158, loss_ctc=20.664, loss_att=9.839, acc=0.936, loss=13.086, backward_time=0.254, grad_norm=121.743, clip=100.000, loss_scale=1.147e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 04:44:51,613 (trainer:779) INFO: 13epoch:train:827-1239batch: iter_time=0.003, forward_time=0.157, loss_ctc=20.822, loss_att=9.897, acc=0.936, loss=13.175, backward_time=0.254, grad_norm=123.022, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 04:48:57,010 (trainer:779) INFO: 13epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.159, loss_ctc=20.852, loss_att=9.898, acc=0.941, loss=13.184, backward_time=0.255, grad_norm=123.209, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 04:53:02,260 (trainer:779) INFO: 13epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.158, loss_ctc=20.435, loss_att=9.734, acc=0.935, loss=12.944, backward_time=0.255, grad_norm=120.944, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 04:57:08,011 (trainer:779) INFO: 13epoch:train:2066-2478batch: iter_time=0.004, forward_time=0.157, loss_ctc=20.577, loss_att=9.773, acc=0.939, loss=13.014, backward_time=0.254, grad_norm=125.982, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 05:01:12,656 (trainer:779) INFO: 13epoch:train:2479-2891batch: iter_time=0.003, forward_time=0.156, loss_ctc=20.571, loss_att=9.839, acc=0.938, loss=13.059, backward_time=0.253, grad_norm=126.834, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 05:05:18,135 (trainer:779) INFO: 13epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.157, loss_ctc=21.134, loss_att=10.081, acc=0.942, loss=13.397, backward_time=0.254, grad_norm=123.581, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 05:09:23,508 (trainer:779) INFO: 13epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.157, loss_ctc=20.779, loss_att=9.856, acc=0.941, loss=13.133, backward_time=0.255, grad_norm=123.040, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 05:13:28,210 (trainer:779) INFO: 13epoch:train:3718-4130batch: iter_time=7.022e-04, forward_time=0.157, loss_ctc=20.728, loss_att=9.832, acc=0.941, loss=13.101, backward_time=0.254, grad_norm=118.690, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 05:17:32,372 (trainer:779) INFO: 13epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.156, loss_ctc=20.375, loss_att=9.712, acc=0.939, loss=12.911, backward_time=0.253, grad_norm=115.603, clip=100.000, loss_scale=2.199e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 05:21:37,040 (trainer:779) INFO: 13epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.157, loss_ctc=20.422, loss_att=9.689, acc=0.939, loss=12.909, backward_time=0.254, grad_norm=121.357, clip=100.000, loss_scale=2.985e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 05:25:40,832 (trainer:779) INFO: 13epoch:train:4957-5369batch: iter_time=8.089e-04, forward_time=0.157, loss_ctc=20.806, loss_att=9.856, acc=0.941, loss=13.141, backward_time=0.254, grad_norm=118.768, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 05:29:46,505 (trainer:779) INFO: 13epoch:train:5370-5782batch: iter_time=0.004, forward_time=0.158, loss_ctc=20.773, loss_att=9.886, acc=0.938, loss=13.152, backward_time=0.255, grad_norm=127.025, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 05:33:51,057 (trainer:779) INFO: 13epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.158, loss_ctc=20.590, loss_att=9.791, acc=0.941, loss=13.031, backward_time=0.254, grad_norm=134.486, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 05:37:56,491 (trainer:779) INFO: 13epoch:train:6196-6608batch: iter_time=0.004, forward_time=0.157, loss_ctc=19.864, loss_att=9.471, acc=0.937, loss=12.589, backward_time=0.254, grad_norm=125.744, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 05:42:01,573 (trainer:779) INFO: 13epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.157, loss_ctc=20.272, loss_att=9.651, acc=0.940, loss=12.837, backward_time=0.254, grad_norm=130.862, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 05:46:08,629 (trainer:779) INFO: 13epoch:train:7022-7434batch: iter_time=0.006, forward_time=0.158, loss_ctc=19.676, loss_att=9.403, acc=0.934, loss=12.485, backward_time=0.254, grad_norm=114.541, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.196
+[bmi2:0/4] 2024-07-07 05:50:12,779 (trainer:779) INFO: 13epoch:train:7435-7847batch: iter_time=8.500e-04, forward_time=0.157, loss_ctc=20.928, loss_att=9.927, acc=0.943, loss=13.228, backward_time=0.254, grad_norm=116.518, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 05:54:16,932 (trainer:779) INFO: 13epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.157, loss_ctc=20.305, loss_att=9.638, acc=0.940, loss=12.838, backward_time=0.254, grad_norm=123.027, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 05:55:15,437 (trainer:365) INFO: 13epoch results: [train] iter_time=0.002, forward_time=0.157, loss_ctc=20.587, loss_att=9.791, acc=0.939, loss=13.030, backward_time=0.254, grad_norm=123.083, clip=100.000, loss_scale=3.011e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188, time=1 hour, 21 minutes and 52.95 seconds, total_count=107471, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.812, cer_ctc=0.055, loss_att=6.485, acc=0.939, cer=0.040, wer=0.547, loss=7.783, time=17.26 seconds, total_count=442, gpu_max_cached_mem_GB=22.482, [att_plot] time=36.34 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 05:55:20,376 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 05:55:20,376 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/3epoch.pth
+[bmi2:0/4] 2024-07-07 05:55:20,377 (trainer:299) INFO: 14/100epoch started. Estimated time to finish: 5 days, 1 hour and 56 minutes
+[bmi2:0/4] 2024-07-07 05:59:32,517 (trainer:779) INFO: 14epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=19.550, loss_att=9.220, acc=0.940, loss=12.319, backward_time=0.253, grad_norm=115.637, clip=100.000, loss_scale=4.398e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.221
+[bmi2:0/4] 2024-07-07 06:03:36,805 (trainer:779) INFO: 14epoch:train:414-826batch: iter_time=2.035e-04, forward_time=0.157, loss_ctc=19.586, loss_att=9.352, acc=0.940, loss=12.422, backward_time=0.255, grad_norm=124.023, clip=100.000, loss_scale=7.415e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 06:07:41,968 (trainer:779) INFO: 14epoch:train:827-1239batch: iter_time=1.781e-04, forward_time=0.158, loss_ctc=19.767, loss_att=9.385, acc=0.941, loss=12.500, backward_time=0.255, grad_norm=127.367, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 06:11:44,696 (trainer:779) INFO: 14epoch:train:1240-1652batch: iter_time=1.854e-04, forward_time=0.156, loss_ctc=19.847, loss_att=9.358, acc=0.943, loss=12.505, backward_time=0.253, grad_norm=123.009, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 06:15:48,437 (trainer:779) INFO: 14epoch:train:1653-2065batch: iter_time=1.864e-04, forward_time=0.158, loss_ctc=19.474, loss_att=9.152, acc=0.939, loss=12.249, backward_time=0.254, grad_norm=119.359, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 06:19:52,213 (trainer:779) INFO: 14epoch:train:2066-2478batch: iter_time=1.907e-04, forward_time=0.157, loss_ctc=19.866, loss_att=9.378, acc=0.944, loss=12.525, backward_time=0.254, grad_norm=118.068, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 06:23:56,364 (trainer:779) INFO: 14epoch:train:2479-2891batch: iter_time=1.799e-04, forward_time=0.157, loss_ctc=19.827, loss_att=9.356, acc=0.941, loss=12.497, backward_time=0.254, grad_norm=113.937, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 06:27:59,882 (trainer:779) INFO: 14epoch:train:2892-3304batch: iter_time=1.724e-04, forward_time=0.157, loss_ctc=19.732, loss_att=9.325, acc=0.943, loss=12.447, backward_time=0.254, grad_norm=117.393, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 06:32:03,237 (trainer:779) INFO: 14epoch:train:3305-3717batch: iter_time=1.790e-04, forward_time=0.157, loss_ctc=19.709, loss_att=9.316, acc=0.943, loss=12.434, backward_time=0.254, grad_norm=117.503, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 06:36:07,492 (trainer:779) INFO: 14epoch:train:3718-4130batch: iter_time=1.875e-04, forward_time=0.158, loss_ctc=19.372, loss_att=9.190, acc=0.941, loss=12.244, backward_time=0.254, grad_norm=114.849, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 06:40:11,719 (trainer:779) INFO: 14epoch:train:4131-4543batch: iter_time=1.819e-04, forward_time=0.157, loss_ctc=19.790, loss_att=9.405, acc=0.942, loss=12.521, backward_time=0.255, grad_norm=120.806, clip=100.000, loss_scale=8.796e+12, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 06:44:14,615 (trainer:779) INFO: 14epoch:train:4544-4956batch: iter_time=1.909e-04, forward_time=0.156, loss_ctc=19.626, loss_att=9.309, acc=0.942, loss=12.404, backward_time=0.253, grad_norm=108.229, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 06:48:18,045 (trainer:779) INFO: 14epoch:train:4957-5369batch: iter_time=1.800e-04, forward_time=0.157, loss_ctc=19.393, loss_att=9.169, acc=0.940, loss=12.236, backward_time=0.254, grad_norm=114.971, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 06:52:20,804 (trainer:779) INFO: 14epoch:train:5370-5782batch: iter_time=1.916e-04, forward_time=0.156, loss_ctc=19.400, loss_att=9.189, acc=0.937, loss=12.252, backward_time=0.253, grad_norm=111.078, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 06:56:24,953 (trainer:779) INFO: 14epoch:train:5783-6195batch: iter_time=1.794e-04, forward_time=0.158, loss_ctc=19.209, loss_att=9.130, acc=0.941, loss=12.154, backward_time=0.255, grad_norm=111.264, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 07:00:28,743 (trainer:779) INFO: 14epoch:train:6196-6608batch: iter_time=1.764e-04, forward_time=0.157, loss_ctc=19.430, loss_att=9.202, acc=0.944, loss=12.270, backward_time=0.254, grad_norm=116.099, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 07:04:32,995 (trainer:779) INFO: 14epoch:train:6609-7021batch: iter_time=1.888e-04, forward_time=0.158, loss_ctc=19.493, loss_att=9.205, acc=0.945, loss=12.291, backward_time=0.254, grad_norm=114.117, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 07:08:36,550 (trainer:779) INFO: 14epoch:train:7022-7434batch: iter_time=1.854e-04, forward_time=0.156, loss_ctc=19.605, loss_att=9.238, acc=0.945, loss=12.348, backward_time=0.254, grad_norm=116.309, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 07:12:40,820 (trainer:779) INFO: 14epoch:train:7435-7847batch: iter_time=1.838e-04, forward_time=0.157, loss_ctc=19.691, loss_att=9.266, acc=0.947, loss=12.393, backward_time=0.254, grad_norm=124.575, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 07:16:44,566 (trainer:779) INFO: 14epoch:train:7848-8260batch: iter_time=1.758e-04, forward_time=0.157, loss_ctc=19.623, loss_att=9.269, acc=0.944, loss=12.375, backward_time=0.254, grad_norm=114.175, clip=100.000, loss_scale=1.759e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 07:17:39,310 (trainer:365) INFO: 14epoch results: [train] iter_time=2.301e-04, forward_time=0.157, loss_ctc=19.598, loss_att=9.270, acc=0.942, loss=12.368, backward_time=0.254, grad_norm=117.120, clip=100.000, loss_scale=1.247e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.182, time=1 hour, 21 minutes and 29 seconds, total_count=115738, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.509, cer_ctc=0.054, loss_att=6.104, acc=0.942, cer=0.038, wer=0.534, loss=7.426, time=17.13 seconds, total_count=476, gpu_max_cached_mem_GB=22.482, [att_plot] time=32.8 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 07:17:44,054 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 07:17:44,055 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/4epoch.pth
+[bmi2:0/4] 2024-07-07 07:17:44,055 (trainer:299) INFO: 15/100epoch started. Estimated time to finish: 5 days, 21 minutes and 42.68 seconds
+[bmi2:0/4] 2024-07-07 07:21:56,769 (trainer:779) INFO: 15epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=18.767, loss_att=8.808, acc=0.944, loss=11.796, backward_time=0.254, grad_norm=111.978, clip=100.000, loss_scale=2.340e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.224
+[bmi2:0/4] 2024-07-07 07:26:00,133 (trainer:779) INFO: 15epoch:train:414-826batch: iter_time=1.805e-04, forward_time=0.157, loss_ctc=18.666, loss_att=8.751, acc=0.944, loss=11.725, backward_time=0.253, grad_norm=113.853, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 07:30:03,829 (trainer:779) INFO: 15epoch:train:827-1239batch: iter_time=1.745e-04, forward_time=0.157, loss_ctc=19.072, loss_att=8.936, acc=0.946, loss=11.977, backward_time=0.254, grad_norm=110.625, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 07:34:07,332 (trainer:779) INFO: 15epoch:train:1240-1652batch: iter_time=1.887e-04, forward_time=0.156, loss_ctc=18.800, loss_att=8.858, acc=0.945, loss=11.840, backward_time=0.254, grad_norm=105.871, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 07:38:12,209 (trainer:779) INFO: 15epoch:train:1653-2065batch: iter_time=1.758e-04, forward_time=0.158, loss_ctc=19.196, loss_att=8.972, acc=0.947, loss=12.039, backward_time=0.255, grad_norm=117.089, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 07:42:16,102 (trainer:779) INFO: 15epoch:train:2066-2478batch: iter_time=1.887e-04, forward_time=0.157, loss_ctc=19.067, loss_att=8.946, acc=0.948, loss=11.982, backward_time=0.254, grad_norm=119.614, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 07:46:20,235 (trainer:779) INFO: 15epoch:train:2479-2891batch: iter_time=1.856e-04, forward_time=0.158, loss_ctc=18.869, loss_att=8.869, acc=0.945, loss=11.869, backward_time=0.254, grad_norm=119.409, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 07:50:23,495 (trainer:779) INFO: 15epoch:train:2892-3304batch: iter_time=1.760e-04, forward_time=0.156, loss_ctc=18.849, loss_att=8.848, acc=0.941, loss=11.848, backward_time=0.254, grad_norm=113.100, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 07:54:25,709 (trainer:779) INFO: 15epoch:train:3305-3717batch: iter_time=1.820e-04, forward_time=0.155, loss_ctc=18.755, loss_att=8.812, acc=0.942, loss=11.795, backward_time=0.253, grad_norm=113.019, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.173
+[bmi2:0/4] 2024-07-07 07:58:31,604 (trainer:779) INFO: 15epoch:train:3718-4130batch: iter_time=1.844e-04, forward_time=0.159, loss_ctc=19.021, loss_att=8.953, acc=0.947, loss=11.973, backward_time=0.256, grad_norm=116.559, clip=100.000, loss_scale=3.518e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 08:02:35,890 (trainer:779) INFO: 15epoch:train:4131-4543batch: iter_time=1.752e-04, forward_time=0.158, loss_ctc=18.516, loss_att=8.697, acc=0.942, loss=11.643, backward_time=0.254, grad_norm=106.754, clip=100.000, loss_scale=5.790e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 08:06:40,829 (trainer:779) INFO: 15epoch:train:4544-4956batch: iter_time=1.755e-04, forward_time=0.157, loss_ctc=18.792, loss_att=8.926, acc=0.947, loss=11.885, backward_time=0.255, grad_norm=124.157, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 08:10:44,771 (trainer:779) INFO: 15epoch:train:4957-5369batch: iter_time=1.743e-04, forward_time=0.158, loss_ctc=18.576, loss_att=8.723, acc=0.945, loss=11.679, backward_time=0.255, grad_norm=105.353, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 08:14:48,990 (trainer:779) INFO: 15epoch:train:5370-5782batch: iter_time=1.825e-04, forward_time=0.157, loss_ctc=18.813, loss_att=8.840, acc=0.946, loss=11.832, backward_time=0.254, grad_norm=114.900, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 08:18:52,485 (trainer:779) INFO: 15epoch:train:5783-6195batch: iter_time=1.785e-04, forward_time=0.157, loss_ctc=18.618, loss_att=8.789, acc=0.943, loss=11.737, backward_time=0.254, grad_norm=106.062, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 08:22:56,367 (trainer:779) INFO: 15epoch:train:6196-6608batch: iter_time=1.747e-04, forward_time=0.157, loss_ctc=18.774, loss_att=8.804, acc=0.947, loss=11.795, backward_time=0.254, grad_norm=111.635, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 08:27:00,661 (trainer:779) INFO: 15epoch:train:6609-7021batch: iter_time=1.747e-04, forward_time=0.157, loss_ctc=18.596, loss_att=8.724, acc=0.945, loss=11.686, backward_time=0.254, grad_norm=113.624, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 08:31:05,402 (trainer:779) INFO: 15epoch:train:7022-7434batch: iter_time=1.855e-04, forward_time=0.158, loss_ctc=18.561, loss_att=8.714, acc=0.947, loss=11.668, backward_time=0.254, grad_norm=111.230, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 08:35:09,349 (trainer:779) INFO: 15epoch:train:7435-7847batch: iter_time=1.826e-04, forward_time=0.157, loss_ctc=18.633, loss_att=8.744, acc=0.945, loss=11.711, backward_time=0.254, grad_norm=116.654, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 08:39:13,112 (trainer:779) INFO: 15epoch:train:7848-8260batch: iter_time=1.739e-04, forward_time=0.157, loss_ctc=18.558, loss_att=8.721, acc=0.944, loss=11.672, backward_time=0.254, grad_norm=110.711, clip=100.000, loss_scale=7.037e+13, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 08:40:06,831 (trainer:365) INFO: 15epoch results: [train] iter_time=2.257e-04, forward_time=0.157, loss_ctc=18.772, loss_att=8.820, acc=0.945, loss=11.806, backward_time=0.254, grad_norm=113.120, clip=100.000, loss_scale=5.158e+13, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184, time=1 hour, 21 minutes and 33.74 seconds, total_count=124005, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.277, cer_ctc=0.052, loss_att=6.278, acc=0.943, cer=0.038, wer=0.532, loss=7.478, time=17.97 seconds, total_count=510, gpu_max_cached_mem_GB=22.482, [att_plot] time=31.07 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 08:40:10,640 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 08:40:10,641 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/6epoch.pth
+[bmi2:0/4] 2024-07-07 08:40:10,641 (trainer:299) INFO: 16/100epoch started. Estimated time to finish: 4 days, 22 hours and 49 minutes
+[bmi2:0/4] 2024-07-07 08:44:23,591 (trainer:779) INFO: 16epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=18.208, loss_att=8.514, acc=0.949, loss=11.422, backward_time=0.254, grad_norm=109.622, clip=100.000, loss_scale=1.390e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.225
+[bmi2:0/4] 2024-07-07 08:48:27,948 (trainer:779) INFO: 16epoch:train:414-826batch: iter_time=1.919e-04, forward_time=0.157, loss_ctc=18.145, loss_att=8.490, acc=0.949, loss=11.387, backward_time=0.255, grad_norm=112.582, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 08:52:29,960 (trainer:779) INFO: 16epoch:train:827-1239batch: iter_time=1.925e-04, forward_time=0.155, loss_ctc=18.300, loss_att=8.498, acc=0.949, loss=11.439, backward_time=0.252, grad_norm=109.842, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.172
+[bmi2:0/4] 2024-07-07 08:56:33,907 (trainer:779) INFO: 16epoch:train:1240-1652batch: iter_time=1.790e-04, forward_time=0.157, loss_ctc=18.340, loss_att=8.571, acc=0.949, loss=11.502, backward_time=0.254, grad_norm=108.312, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 09:00:37,396 (trainer:779) INFO: 16epoch:train:1653-2065batch: iter_time=1.868e-04, forward_time=0.156, loss_ctc=18.611, loss_att=8.652, acc=0.949, loss=11.640, backward_time=0.253, grad_norm=119.014, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 09:04:41,779 (trainer:779) INFO: 16epoch:train:2066-2478batch: iter_time=1.923e-04, forward_time=0.158, loss_ctc=18.365, loss_att=8.593, acc=0.947, loss=11.524, backward_time=0.255, grad_norm=113.673, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 09:08:46,004 (trainer:779) INFO: 16epoch:train:2479-2891batch: iter_time=1.812e-04, forward_time=0.158, loss_ctc=18.013, loss_att=8.417, acc=0.944, loss=11.296, backward_time=0.254, grad_norm=105.601, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 09:12:50,176 (trainer:779) INFO: 16epoch:train:2892-3304batch: iter_time=1.748e-04, forward_time=0.158, loss_ctc=18.275, loss_att=8.561, acc=0.948, loss=11.475, backward_time=0.255, grad_norm=106.801, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 09:16:53,724 (trainer:779) INFO: 16epoch:train:3305-3717batch: iter_time=1.789e-04, forward_time=0.157, loss_ctc=18.161, loss_att=8.509, acc=0.948, loss=11.405, backward_time=0.253, grad_norm=112.347, clip=100.000, loss_scale=1.407e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 09:20:56,901 (trainer:779) INFO: 16epoch:train:3718-4130batch: iter_time=1.809e-04, forward_time=0.156, loss_ctc=18.217, loss_att=8.600, acc=0.948, loss=11.485, backward_time=0.253, grad_norm=111.957, clip=100.000, loss_scale=1.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 09:24:59,740 (trainer:779) INFO: 16epoch:train:4131-4543batch: iter_time=1.767e-04, forward_time=0.156, loss_ctc=17.990, loss_att=8.395, acc=0.947, loss=11.273, backward_time=0.253, grad_norm=107.245, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 09:29:04,728 (trainer:779) INFO: 16epoch:train:4544-4956batch: iter_time=1.841e-04, forward_time=0.158, loss_ctc=18.147, loss_att=8.490, acc=0.946, loss=11.387, backward_time=0.255, grad_norm=109.458, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 09:33:07,688 (trainer:779) INFO: 16epoch:train:4957-5369batch: iter_time=1.802e-04, forward_time=0.156, loss_ctc=18.042, loss_att=8.499, acc=0.945, loss=11.362, backward_time=0.253, grad_norm=114.945, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 09:37:11,087 (trainer:779) INFO: 16epoch:train:5370-5782batch: iter_time=1.910e-04, forward_time=0.157, loss_ctc=17.704, loss_att=8.269, acc=0.945, loss=11.099, backward_time=0.254, grad_norm=105.825, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 09:41:14,311 (trainer:779) INFO: 16epoch:train:5783-6195batch: iter_time=1.861e-04, forward_time=0.156, loss_ctc=18.053, loss_att=8.448, acc=0.946, loss=11.329, backward_time=0.253, grad_norm=106.661, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 09:45:17,338 (trainer:779) INFO: 16epoch:train:6196-6608batch: iter_time=1.821e-04, forward_time=0.156, loss_ctc=17.750, loss_att=8.319, acc=0.946, loss=11.148, backward_time=0.253, grad_norm=106.351, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 09:49:21,441 (trainer:779) INFO: 16epoch:train:6609-7021batch: iter_time=1.773e-04, forward_time=0.158, loss_ctc=18.080, loss_att=8.452, acc=0.946, loss=11.340, backward_time=0.254, grad_norm=107.033, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 09:53:24,971 (trainer:779) INFO: 16epoch:train:7022-7434batch: iter_time=1.907e-04, forward_time=0.157, loss_ctc=18.032, loss_att=8.403, acc=0.947, loss=11.292, backward_time=0.254, grad_norm=106.869, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 09:57:29,897 (trainer:779) INFO: 16epoch:train:7435-7847batch: iter_time=1.818e-04, forward_time=0.158, loss_ctc=18.135, loss_att=8.480, acc=0.950, loss=11.376, backward_time=0.255, grad_norm=104.608, clip=100.000, loss_scale=2.815e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 10:01:32,790 (trainer:779) INFO: 16epoch:train:7848-8260batch: iter_time=1.833e-04, forward_time=0.156, loss_ctc=17.671, loss_att=8.269, acc=0.945, loss=11.090, backward_time=0.254, grad_norm=110.667, clip=100.000, loss_scale=4.514e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 10:02:32,278 (trainer:365) INFO: 16epoch results: [train] iter_time=2.399e-04, forward_time=0.157, loss_ctc=18.109, loss_att=8.470, acc=0.947, loss=11.362, backward_time=0.254, grad_norm=109.484, clip=100.000, loss_scale=2.218e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182, time=1 hour, 21 minutes and 27 seconds, total_count=132272, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.422, cer_ctc=0.051, loss_att=6.229, acc=0.941, cer=0.037, wer=0.531, loss=7.487, time=17.46 seconds, total_count=544, gpu_max_cached_mem_GB=22.482, [att_plot] time=37.17 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 10:02:36,810 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 10:02:36,811 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/5epoch.pth
+[bmi2:0/4] 2024-07-07 10:02:36,811 (trainer:299) INFO: 17/100epoch started. Estimated time to finish: 4 days, 21 hours and 17 minutes
+[bmi2:0/4] 2024-07-07 10:06:49,958 (trainer:779) INFO: 17epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=17.397, loss_att=8.107, acc=0.947, loss=10.894, backward_time=0.254, grad_norm=108.054, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.226
+[bmi2:0/4] 2024-07-07 10:10:52,317 (trainer:779) INFO: 17epoch:train:414-826batch: iter_time=1.869e-04, forward_time=0.156, loss_ctc=17.575, loss_att=8.188, acc=0.947, loss=11.004, backward_time=0.253, grad_norm=102.647, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.173
+[bmi2:0/4] 2024-07-07 10:14:55,043 (trainer:779) INFO: 17epoch:train:827-1239batch: iter_time=1.816e-04, forward_time=0.156, loss_ctc=17.424, loss_att=8.090, acc=0.947, loss=10.890, backward_time=0.253, grad_norm=108.492, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 10:18:59,196 (trainer:779) INFO: 17epoch:train:1240-1652batch: iter_time=1.780e-04, forward_time=0.158, loss_ctc=17.644, loss_att=8.243, acc=0.950, loss=11.063, backward_time=0.254, grad_norm=111.670, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 10:23:04,069 (trainer:779) INFO: 17epoch:train:1653-2065batch: iter_time=1.880e-04, forward_time=0.158, loss_ctc=17.796, loss_att=8.247, acc=0.952, loss=11.112, backward_time=0.255, grad_norm=105.864, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 10:27:09,065 (trainer:779) INFO: 17epoch:train:2066-2478batch: iter_time=1.731e-04, forward_time=0.158, loss_ctc=17.811, loss_att=8.228, acc=0.951, loss=11.103, backward_time=0.255, grad_norm=103.238, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 10:31:12,235 (trainer:779) INFO: 17epoch:train:2479-2891batch: iter_time=1.760e-04, forward_time=0.156, loss_ctc=17.742, loss_att=8.251, acc=0.950, loss=11.099, backward_time=0.254, grad_norm=113.912, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 10:35:16,097 (trainer:779) INFO: 17epoch:train:2892-3304batch: iter_time=1.713e-04, forward_time=0.158, loss_ctc=17.606, loss_att=8.161, acc=0.949, loss=10.994, backward_time=0.254, grad_norm=110.235, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 10:39:20,488 (trainer:779) INFO: 17epoch:train:3305-3717batch: iter_time=1.930e-04, forward_time=0.158, loss_ctc=17.627, loss_att=8.155, acc=0.951, loss=10.997, backward_time=0.255, grad_norm=106.396, clip=100.000, loss_scale=5.629e+14, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 10:43:25,539 (trainer:779) INFO: 17epoch:train:3718-4130batch: iter_time=1.814e-04, forward_time=0.158, loss_ctc=17.810, loss_att=8.274, acc=0.951, loss=11.135, backward_time=0.255, grad_norm=108.140, clip=100.000, loss_scale=1.088e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 10:47:28,540 (trainer:779) INFO: 17epoch:train:4131-4543batch: iter_time=1.808e-04, forward_time=0.157, loss_ctc=17.444, loss_att=8.109, acc=0.946, loss=10.910, backward_time=0.253, grad_norm=97.678, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 10:51:32,068 (trainer:779) INFO: 17epoch:train:4544-4956batch: iter_time=1.889e-04, forward_time=0.156, loss_ctc=17.411, loss_att=8.116, acc=0.944, loss=10.905, backward_time=0.254, grad_norm=107.335, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 10:55:35,373 (trainer:779) INFO: 17epoch:train:4957-5369batch: iter_time=1.854e-04, forward_time=0.156, loss_ctc=17.501, loss_att=8.147, acc=0.949, loss=10.953, backward_time=0.254, grad_norm=109.698, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 10:59:39,087 (trainer:779) INFO: 17epoch:train:5370-5782batch: iter_time=1.929e-04, forward_time=0.156, loss_ctc=17.662, loss_att=8.173, acc=0.952, loss=11.020, backward_time=0.253, grad_norm=111.198, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 11:03:43,183 (trainer:779) INFO: 17epoch:train:5783-6195batch: iter_time=1.764e-04, forward_time=0.157, loss_ctc=17.712, loss_att=8.224, acc=0.953, loss=11.071, backward_time=0.254, grad_norm=110.193, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 11:07:47,053 (trainer:779) INFO: 17epoch:train:6196-6608batch: iter_time=1.798e-04, forward_time=0.157, loss_ctc=17.497, loss_att=8.128, acc=0.948, loss=10.938, backward_time=0.254, grad_norm=104.335, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 11:11:51,053 (trainer:779) INFO: 17epoch:train:6609-7021batch: iter_time=1.957e-04, forward_time=0.157, loss_ctc=17.482, loss_att=8.106, acc=0.950, loss=10.919, backward_time=0.254, grad_norm=104.662, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 11:15:56,498 (trainer:779) INFO: 17epoch:train:7022-7434batch: iter_time=1.936e-04, forward_time=0.158, loss_ctc=17.182, loss_att=7.986, acc=0.946, loss=10.745, backward_time=0.255, grad_norm=105.198, clip=100.000, loss_scale=1.126e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 11:20:00,117 (trainer:779) INFO: 17epoch:train:7435-7847batch: iter_time=2.312e-04, forward_time=0.156, loss_ctc=17.633, loss_att=8.177, acc=0.952, loss=11.013, backward_time=0.254, grad_norm=105.961, clip=100.000, loss_scale=1.405e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 11:24:04,101 (trainer:779) INFO: 17epoch:train:7848-8260batch: iter_time=2.197e-04, forward_time=0.156, loss_ctc=17.308, loss_att=8.038, acc=0.948, loss=10.819, backward_time=0.254, grad_norm=112.630, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 11:25:01,185 (trainer:365) INFO: 17epoch results: [train] iter_time=2.293e-04, forward_time=0.157, loss_ctc=17.561, loss_att=8.157, acc=0.949, loss=10.978, backward_time=0.254, grad_norm=107.374, clip=100.000, loss_scale=9.420e+14, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183, time=1 hour, 21 minutes and 32.46 seconds, total_count=140539, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.082, cer_ctc=0.050, loss_att=5.737, acc=0.945, cer=0.035, wer=0.524, loss=7.040, time=17.13 seconds, total_count=578, gpu_max_cached_mem_GB=22.482, [att_plot] time=34.79 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 11:25:05,641 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 11:25:05,643 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/7epoch.pth
+[bmi2:0/4] 2024-07-07 11:25:05,643 (trainer:299) INFO: 18/100epoch started. Estimated time to finish: 4 days, 19 hours and 47 minutes
+[bmi2:0/4] 2024-07-07 11:29:17,876 (trainer:779) INFO: 18epoch:train:1-413batch: iter_time=9.947e-04, forward_time=0.156, loss_ctc=17.197, loss_att=7.933, acc=0.954, loss=10.712, backward_time=0.255, grad_norm=103.923, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.222
+[bmi2:0/4] 2024-07-07 11:33:22,333 (trainer:779) INFO: 18epoch:train:414-826batch: iter_time=2.333e-04, forward_time=0.157, loss_ctc=16.863, loss_att=7.776, acc=0.950, loss=10.502, backward_time=0.255, grad_norm=106.306, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 11:37:26,314 (trainer:779) INFO: 18epoch:train:827-1239batch: iter_time=2.143e-04, forward_time=0.157, loss_ctc=17.067, loss_att=7.867, acc=0.952, loss=10.627, backward_time=0.255, grad_norm=104.084, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 11:41:30,367 (trainer:779) INFO: 18epoch:train:1240-1652batch: iter_time=2.308e-04, forward_time=0.156, loss_ctc=17.165, loss_att=7.939, acc=0.953, loss=10.707, backward_time=0.254, grad_norm=102.173, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 11:45:34,321 (trainer:779) INFO: 18epoch:train:1653-2065batch: iter_time=2.202e-04, forward_time=0.156, loss_ctc=17.197, loss_att=8.016, acc=0.950, loss=10.770, backward_time=0.254, grad_norm=104.192, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 11:49:38,266 (trainer:779) INFO: 18epoch:train:2066-2478batch: iter_time=2.206e-04, forward_time=0.157, loss_ctc=16.846, loss_att=7.795, acc=0.951, loss=10.510, backward_time=0.254, grad_norm=100.705, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 11:53:42,953 (trainer:779) INFO: 18epoch:train:2479-2891batch: iter_time=2.129e-04, forward_time=0.159, loss_ctc=17.013, loss_att=7.903, acc=0.947, loss=10.636, backward_time=0.255, grad_norm=104.291, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 11:57:47,828 (trainer:779) INFO: 18epoch:train:2892-3304batch: iter_time=2.306e-04, forward_time=0.157, loss_ctc=17.026, loss_att=7.888, acc=0.952, loss=10.629, backward_time=0.255, grad_norm=106.889, clip=100.000, loss_scale=2.252e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 12:01:50,937 (trainer:779) INFO: 18epoch:train:3305-3717batch: iter_time=2.184e-04, forward_time=0.155, loss_ctc=17.239, loss_att=7.947, acc=0.953, loss=10.735, backward_time=0.253, grad_norm=101.698, clip=100.000, loss_scale=3.553e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 12:05:55,426 (trainer:779) INFO: 18epoch:train:3718-4130batch: iter_time=2.342e-04, forward_time=0.157, loss_ctc=16.932, loss_att=7.851, acc=0.946, loss=10.575, backward_time=0.255, grad_norm=102.997, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 12:09:59,670 (trainer:779) INFO: 18epoch:train:4131-4543batch: iter_time=2.300e-04, forward_time=0.157, loss_ctc=17.225, loss_att=7.974, acc=0.955, loss=10.749, backward_time=0.255, grad_norm=98.614, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 12:14:03,570 (trainer:779) INFO: 18epoch:train:4544-4956batch: iter_time=2.258e-04, forward_time=0.156, loss_ctc=16.873, loss_att=7.842, acc=0.950, loss=10.551, backward_time=0.254, grad_norm=104.318, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 12:18:06,500 (trainer:779) INFO: 18epoch:train:4957-5369batch: iter_time=2.378e-04, forward_time=0.156, loss_ctc=17.097, loss_att=7.944, acc=0.952, loss=10.690, backward_time=0.253, grad_norm=105.694, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 12:22:12,158 (trainer:779) INFO: 18epoch:train:5370-5782batch: iter_time=2.164e-04, forward_time=0.159, loss_ctc=17.071, loss_att=7.957, acc=0.953, loss=10.691, backward_time=0.256, grad_norm=106.593, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 12:26:17,245 (trainer:779) INFO: 18epoch:train:5783-6195batch: iter_time=2.222e-04, forward_time=0.158, loss_ctc=16.690, loss_att=7.749, acc=0.950, loss=10.432, backward_time=0.255, grad_norm=97.861, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 12:30:23,134 (trainer:779) INFO: 18epoch:train:6196-6608batch: iter_time=2.226e-04, forward_time=0.159, loss_ctc=16.698, loss_att=7.781, acc=0.951, loss=10.456, backward_time=0.255, grad_norm=100.286, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 12:34:28,818 (trainer:779) INFO: 18epoch:train:6609-7021batch: iter_time=2.208e-04, forward_time=0.158, loss_ctc=16.882, loss_att=7.849, acc=0.951, loss=10.559, backward_time=0.255, grad_norm=98.753, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.031, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 12:38:34,594 (trainer:779) INFO: 18epoch:train:7022-7434batch: iter_time=2.180e-04, forward_time=0.158, loss_ctc=16.571, loss_att=7.725, acc=0.949, loss=10.379, backward_time=0.256, grad_norm=91.849, clip=100.000, loss_scale=4.504e+15, optim_step_time=0.031, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 12:42:39,071 (trainer:779) INFO: 18epoch:train:7435-7847batch: iter_time=2.129e-04, forward_time=0.157, loss_ctc=16.838, loss_att=7.808, acc=0.950, loss=10.517, backward_time=0.255, grad_norm=100.192, clip=100.000, loss_scale=8.526e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 12:46:44,526 (trainer:779) INFO: 18epoch:train:7848-8260batch: iter_time=2.217e-04, forward_time=0.158, loss_ctc=16.861, loss_att=7.817, acc=0.951, loss=10.530, backward_time=0.256, grad_norm=103.859, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 12:47:42,907 (trainer:365) INFO: 18epoch results: [train] iter_time=2.619e-04, forward_time=0.157, loss_ctc=16.966, loss_att=7.867, acc=0.951, loss=10.597, backward_time=0.255, grad_norm=102.261, clip=100.000, loss_scale=3.985e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.186, time=1 hour, 21 minutes and 43.78 seconds, total_count=148806, gpu_max_cached_mem_GB=22.482, [valid] loss_ctc=10.110, cer_ctc=0.050, loss_att=5.931, acc=0.945, cer=0.036, wer=0.515, loss=7.184, time=17.53 seconds, total_count=612, gpu_max_cached_mem_GB=22.482, [att_plot] time=35.96 seconds, total_count=0, gpu_max_cached_mem_GB=22.482
+[bmi2:0/4] 2024-07-07 12:47:47,418 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 12:47:47,420 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/8epoch.pth
+[bmi2:0/4] 2024-07-07 12:47:47,420 (trainer:299) INFO: 19/100epoch started. Estimated time to finish: 4 days, 18 hours and 19 minutes
+[bmi2:0/4] 2024-07-07 12:52:01,165 (trainer:779) INFO: 19epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=16.435, loss_att=7.576, acc=0.952, loss=10.234, backward_time=0.255, grad_norm=97.193, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.229
+[bmi2:0/4] 2024-07-07 12:56:05,800 (trainer:779) INFO: 19epoch:train:414-826batch: iter_time=2.206e-04, forward_time=0.157, loss_ctc=16.478, loss_att=7.603, acc=0.955, loss=10.266, backward_time=0.255, grad_norm=100.687, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 13:00:09,673 (trainer:779) INFO: 19epoch:train:827-1239batch: iter_time=2.207e-04, forward_time=0.156, loss_ctc=16.289, loss_att=7.557, acc=0.950, loss=10.177, backward_time=0.255, grad_norm=105.984, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 13:04:13,912 (trainer:779) INFO: 19epoch:train:1240-1652batch: iter_time=2.267e-04, forward_time=0.157, loss_ctc=16.489, loss_att=7.656, acc=0.949, loss=10.306, backward_time=0.254, grad_norm=96.802, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 13:08:18,522 (trainer:779) INFO: 19epoch:train:1653-2065batch: iter_time=2.207e-04, forward_time=0.157, loss_ctc=16.283, loss_att=7.544, acc=0.953, loss=10.166, backward_time=0.255, grad_norm=99.078, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 13:12:23,440 (trainer:779) INFO: 19epoch:train:2066-2478batch: iter_time=2.244e-04, forward_time=0.157, loss_ctc=16.142, loss_att=7.480, acc=0.952, loss=10.079, backward_time=0.255, grad_norm=98.945, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 13:16:27,547 (trainer:779) INFO: 19epoch:train:2479-2891batch: iter_time=2.186e-04, forward_time=0.156, loss_ctc=16.385, loss_att=7.542, acc=0.950, loss=10.195, backward_time=0.255, grad_norm=100.861, clip=100.000, loss_scale=9.007e+15, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 13:20:33,804 (trainer:779) INFO: 19epoch:train:2892-3304batch: iter_time=2.210e-04, forward_time=0.158, loss_ctc=16.937, loss_att=7.798, acc=0.956, loss=10.540, backward_time=0.254, grad_norm=102.338, clip=100.000, loss_scale=1.101e+16, optim_step_time=0.031, optim0_lr0=0.001, train_time=1.192
+[bmi2:0/4] 2024-07-07 13:24:39,169 (trainer:779) INFO: 19epoch:train:3305-3717batch: iter_time=2.328e-04, forward_time=0.157, loss_ctc=16.351, loss_att=7.548, acc=0.951, loss=10.189, backward_time=0.254, grad_norm=104.698, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 13:28:46,478 (trainer:779) INFO: 19epoch:train:3718-4130batch: iter_time=2.366e-04, forward_time=0.157, loss_ctc=16.438, loss_att=7.522, acc=0.952, loss=10.197, backward_time=0.256, grad_norm=100.273, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.031, optim0_lr0=0.001, train_time=1.197
+[bmi2:0/4] 2024-07-07 13:32:52,064 (trainer:779) INFO: 19epoch:train:4131-4543batch: iter_time=2.255e-04, forward_time=0.156, loss_ctc=16.467, loss_att=7.602, acc=0.950, loss=10.261, backward_time=0.255, grad_norm=101.651, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.030, optim0_lr0=0.001, train_time=1.190
+[bmi2:0/4] 2024-07-07 13:36:58,381 (trainer:779) INFO: 19epoch:train:4544-4956batch: iter_time=1.805e-04, forward_time=0.159, loss_ctc=16.384, loss_att=7.591, acc=0.950, loss=10.229, backward_time=0.255, grad_norm=100.165, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.192
+[bmi2:0/4] 2024-07-07 13:41:04,427 (trainer:779) INFO: 19epoch:train:4957-5369batch: iter_time=1.813e-04, forward_time=0.159, loss_ctc=16.896, loss_att=7.747, acc=0.956, loss=10.492, backward_time=0.255, grad_norm=105.611, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.192
+[bmi2:0/4] 2024-07-07 13:45:11,846 (trainer:779) INFO: 19epoch:train:5370-5782batch: iter_time=1.797e-04, forward_time=0.160, loss_ctc=16.636, loss_att=7.682, acc=0.956, loss=10.368, backward_time=0.256, grad_norm=114.359, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.197
+[bmi2:0/4] 2024-07-07 13:49:16,412 (trainer:779) INFO: 19epoch:train:5783-6195batch: iter_time=1.784e-04, forward_time=0.158, loss_ctc=16.566, loss_att=7.659, acc=0.952, loss=10.331, backward_time=0.254, grad_norm=111.698, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 13:53:21,263 (trainer:779) INFO: 19epoch:train:6196-6608batch: iter_time=1.755e-04, forward_time=0.158, loss_ctc=16.630, loss_att=7.654, acc=0.956, loss=10.347, backward_time=0.254, grad_norm=104.206, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 13:57:25,825 (trainer:779) INFO: 19epoch:train:6609-7021batch: iter_time=1.817e-04, forward_time=0.158, loss_ctc=16.288, loss_att=7.538, acc=0.953, loss=10.163, backward_time=0.255, grad_norm=100.291, clip=100.000, loss_scale=1.801e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 14:01:30,266 (trainer:779) INFO: 19epoch:train:7022-7434batch: iter_time=1.742e-04, forward_time=0.157, loss_ctc=16.303, loss_att=7.534, acc=0.952, loss=10.165, backward_time=0.254, grad_norm=105.413, clip=100.000, loss_scale=2.767e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 14:05:33,762 (trainer:779) INFO: 19epoch:train:7435-7847batch: iter_time=1.858e-04, forward_time=0.156, loss_ctc=16.678, loss_att=7.687, acc=0.955, loss=10.384, backward_time=0.254, grad_norm=101.486, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 14:09:37,795 (trainer:779) INFO: 19epoch:train:7848-8260batch: iter_time=1.812e-04, forward_time=0.157, loss_ctc=16.458, loss_att=7.591, acc=0.952, loss=10.251, backward_time=0.254, grad_norm=103.531, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 14:10:34,644 (trainer:365) INFO: 19epoch results: [train] iter_time=2.614e-04, forward_time=0.157, loss_ctc=16.473, loss_att=7.604, acc=0.953, loss=10.265, backward_time=0.255, grad_norm=102.764, clip=100.000, loss_scale=1.681e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189, time=1 hour, 21 minutes and 55.6 seconds, total_count=157073, gpu_max_cached_mem_GB=22.527, [valid] loss_ctc=10.007, cer_ctc=0.049, loss_att=5.952, acc=0.945, cer=0.036, wer=0.511, loss=7.169, time=17.66 seconds, total_count=646, gpu_max_cached_mem_GB=22.527, [att_plot] time=33.95 seconds, total_count=0, gpu_max_cached_mem_GB=22.527
+[bmi2:0/4] 2024-07-07 14:10:38,722 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 14:10:38,723 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/9epoch.pth
+[bmi2:0/4] 2024-07-07 14:10:38,723 (trainer:299) INFO: 20/100epoch started. Estimated time to finish: 4 days, 16 hours and 52 minutes
+[bmi2:0/4] 2024-07-07 14:14:52,692 (trainer:779) INFO: 20epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=15.873, loss_att=7.276, acc=0.954, loss=9.855, backward_time=0.254, grad_norm=100.298, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.230
+[bmi2:0/4] 2024-07-07 14:18:57,217 (trainer:779) INFO: 20epoch:train:414-826batch: iter_time=1.814e-04, forward_time=0.158, loss_ctc=16.132, loss_att=7.365, acc=0.957, loss=9.995, backward_time=0.255, grad_norm=98.275, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 14:23:01,410 (trainer:779) INFO: 20epoch:train:827-1239batch: iter_time=1.843e-04, forward_time=0.158, loss_ctc=16.165, loss_att=7.395, acc=0.956, loss=10.026, backward_time=0.254, grad_norm=103.311, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 14:27:05,855 (trainer:779) INFO: 20epoch:train:1240-1652batch: iter_time=1.812e-04, forward_time=0.157, loss_ctc=16.376, loss_att=7.518, acc=0.957, loss=10.175, backward_time=0.255, grad_norm=103.929, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 14:31:09,160 (trainer:779) INFO: 20epoch:train:1653-2065batch: iter_time=1.761e-04, forward_time=0.156, loss_ctc=16.474, loss_att=7.543, acc=0.959, loss=10.222, backward_time=0.255, grad_norm=105.577, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 14:35:13,356 (trainer:779) INFO: 20epoch:train:2066-2478batch: iter_time=1.757e-04, forward_time=0.158, loss_ctc=16.066, loss_att=7.393, acc=0.953, loss=9.995, backward_time=0.254, grad_norm=96.769, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 14:39:16,965 (trainer:779) INFO: 20epoch:train:2479-2891batch: iter_time=1.912e-04, forward_time=0.157, loss_ctc=15.910, loss_att=7.362, acc=0.949, loss=9.926, backward_time=0.254, grad_norm=102.625, clip=100.000, loss_scale=3.603e+16, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 14:43:20,944 (trainer:779) INFO: 20epoch:train:2892-3304batch: iter_time=1.706e-04, forward_time=0.157, loss_ctc=16.186, loss_att=7.413, acc=0.955, loss=10.045, backward_time=0.254, grad_norm=101.042, clip=100.000, loss_scale=6.718e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 14:47:24,504 (trainer:779) INFO: 20epoch:train:3305-3717batch: iter_time=1.770e-04, forward_time=0.157, loss_ctc=16.144, loss_att=7.448, acc=0.951, loss=10.057, backward_time=0.255, grad_norm=130.003, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 14:51:28,801 (trainer:779) INFO: 20epoch:train:3718-4130batch: iter_time=1.762e-04, forward_time=0.157, loss_ctc=16.056, loss_att=7.386, acc=0.952, loss=9.987, backward_time=0.254, grad_norm=104.614, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 14:55:33,291 (trainer:779) INFO: 20epoch:train:4131-4543batch: iter_time=1.836e-04, forward_time=0.158, loss_ctc=16.143, loss_att=7.400, acc=0.956, loss=10.023, backward_time=0.255, grad_norm=97.293, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 14:59:38,322 (trainer:779) INFO: 20epoch:train:4544-4956batch: iter_time=1.845e-04, forward_time=0.158, loss_ctc=16.042, loss_att=7.372, acc=0.953, loss=9.973, backward_time=0.254, grad_norm=100.409, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 15:03:42,665 (trainer:779) INFO: 20epoch:train:4957-5369batch: iter_time=1.796e-04, forward_time=0.157, loss_ctc=16.264, loss_att=7.491, acc=0.954, loss=10.123, backward_time=0.254, grad_norm=100.937, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 15:07:46,460 (trainer:779) INFO: 20epoch:train:5370-5782batch: iter_time=1.908e-04, forward_time=0.156, loss_ctc=16.279, loss_att=7.470, acc=0.955, loss=10.113, backward_time=0.253, grad_norm=99.242, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 15:11:50,802 (trainer:779) INFO: 20epoch:train:5783-6195batch: iter_time=1.814e-04, forward_time=0.158, loss_ctc=16.185, loss_att=7.437, acc=0.955, loss=10.061, backward_time=0.255, grad_norm=107.614, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 15:15:56,541 (trainer:779) INFO: 20epoch:train:6196-6608batch: iter_time=1.799e-04, forward_time=0.158, loss_ctc=15.783, loss_att=7.270, acc=0.953, loss=9.824, backward_time=0.255, grad_norm=103.701, clip=100.000, loss_scale=7.206e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.189
+[bmi2:0/4] 2024-07-07 15:20:01,449 (trainer:779) INFO: 20epoch:train:6609-7021batch: iter_time=1.698e-04, forward_time=0.159, loss_ctc=15.873, loss_att=7.286, acc=0.951, loss=9.862, backward_time=0.254, grad_norm=98.160, clip=100.000, loss_scale=8.500e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 15:24:06,851 (trainer:779) INFO: 20epoch:train:7022-7434batch: iter_time=1.869e-04, forward_time=0.158, loss_ctc=16.168, loss_att=7.416, acc=0.955, loss=10.042, backward_time=0.254, grad_norm=100.469, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 15:28:11,404 (trainer:779) INFO: 20epoch:train:7435-7847batch: iter_time=1.851e-04, forward_time=0.158, loss_ctc=15.976, loss_att=7.379, acc=0.953, loss=9.958, backward_time=0.254, grad_norm=106.607, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 15:32:16,855 (trainer:779) INFO: 20epoch:train:7848-8260batch: iter_time=1.728e-04, forward_time=0.159, loss_ctc=15.758, loss_att=7.246, acc=0.951, loss=9.800, backward_time=0.255, grad_norm=94.102, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 15:33:18,583 (trainer:365) INFO: 20epoch results: [train] iter_time=2.264e-04, forward_time=0.158, loss_ctc=16.090, loss_att=7.392, acc=0.954, loss=10.002, backward_time=0.254, grad_norm=102.732, clip=100.000, loss_scale=7.072e+16, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186, time=1 hour, 21 minutes and 42.99 seconds, total_count=165340, gpu_max_cached_mem_GB=22.527, [valid] loss_ctc=9.947, cer_ctc=0.048, loss_att=5.929, acc=0.944, cer=0.036, wer=0.520, loss=7.134, time=17.73 seconds, total_count=680, gpu_max_cached_mem_GB=22.527, [att_plot] time=39.13 seconds, total_count=0, gpu_max_cached_mem_GB=22.527
+[bmi2:0/4] 2024-07-07 15:33:23,873 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 15:33:23,923 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/10epoch.pth
+[bmi2:0/4] 2024-07-07 15:33:23,924 (trainer:299) INFO: 21/100epoch started. Estimated time to finish: 4 days, 15 hours and 25 minutes
+[bmi2:0/4] 2024-07-07 15:37:37,981 (trainer:779) INFO: 21epoch:train:1-413batch: iter_time=9.740e-04, forward_time=0.155, loss_ctc=15.752, loss_att=7.189, acc=0.957, loss=9.758, backward_time=0.254, grad_norm=97.846, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.231
+[bmi2:0/4] 2024-07-07 15:41:41,363 (trainer:779) INFO: 21epoch:train:414-826batch: iter_time=1.853e-04, forward_time=0.156, loss_ctc=15.772, loss_att=7.250, acc=0.954, loss=9.806, backward_time=0.254, grad_norm=102.486, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 15:45:46,261 (trainer:779) INFO: 21epoch:train:827-1239batch: iter_time=1.796e-04, forward_time=0.159, loss_ctc=15.900, loss_att=7.247, acc=0.958, loss=9.843, backward_time=0.255, grad_norm=102.599, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 15:49:51,117 (trainer:779) INFO: 21epoch:train:1240-1652batch: iter_time=1.817e-04, forward_time=0.159, loss_ctc=15.903, loss_att=7.323, acc=0.956, loss=9.897, backward_time=0.255, grad_norm=112.621, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 15:53:55,483 (trainer:779) INFO: 21epoch:train:1653-2065batch: iter_time=1.815e-04, forward_time=0.157, loss_ctc=15.418, loss_att=7.070, acc=0.955, loss=9.574, backward_time=0.255, grad_norm=98.035, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 15:58:00,042 (trainer:779) INFO: 21epoch:train:2066-2478batch: iter_time=1.756e-04, forward_time=0.158, loss_ctc=15.758, loss_att=7.219, acc=0.956, loss=9.781, backward_time=0.254, grad_norm=102.531, clip=100.000, loss_scale=1.441e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 16:02:02,881 (trainer:779) INFO: 21epoch:train:2479-2891batch: iter_time=1.797e-04, forward_time=0.156, loss_ctc=15.780, loss_att=7.205, acc=0.955, loss=9.778, backward_time=0.253, grad_norm=98.394, clip=100.000, loss_scale=2.176e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 16:06:08,249 (trainer:779) INFO: 21epoch:train:2892-3304batch: iter_time=1.829e-04, forward_time=0.159, loss_ctc=15.723, loss_att=7.187, acc=0.955, loss=9.747, backward_time=0.255, grad_norm=99.592, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 16:10:12,636 (trainer:779) INFO: 21epoch:train:3305-3717batch: iter_time=1.771e-04, forward_time=0.158, loss_ctc=15.650, loss_att=7.175, acc=0.954, loss=9.717, backward_time=0.255, grad_norm=101.644, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 16:14:17,942 (trainer:779) INFO: 21epoch:train:3718-4130batch: iter_time=1.727e-04, forward_time=0.158, loss_ctc=15.899, loss_att=7.280, acc=0.955, loss=9.866, backward_time=0.255, grad_norm=95.242, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 16:18:21,887 (trainer:779) INFO: 21epoch:train:4131-4543batch: iter_time=1.866e-04, forward_time=0.157, loss_ctc=15.720, loss_att=7.168, acc=0.955, loss=9.734, backward_time=0.255, grad_norm=104.426, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 16:22:26,815 (trainer:779) INFO: 21epoch:train:4544-4956batch: iter_time=1.776e-04, forward_time=0.158, loss_ctc=15.692, loss_att=7.189, acc=0.957, loss=9.740, backward_time=0.255, grad_norm=96.940, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 16:26:30,527 (trainer:779) INFO: 21epoch:train:4957-5369batch: iter_time=1.760e-04, forward_time=0.157, loss_ctc=15.695, loss_att=7.164, acc=0.955, loss=9.723, backward_time=0.255, grad_norm=101.185, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 16:30:34,340 (trainer:779) INFO: 21epoch:train:5370-5782batch: iter_time=1.780e-04, forward_time=0.157, loss_ctc=15.531, loss_att=7.125, acc=0.953, loss=9.647, backward_time=0.253, grad_norm=93.486, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 16:34:38,580 (trainer:779) INFO: 21epoch:train:5783-6195batch: iter_time=1.807e-04, forward_time=0.158, loss_ctc=15.551, loss_att=7.085, acc=0.955, loss=9.625, backward_time=0.254, grad_norm=97.724, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 16:38:42,968 (trainer:779) INFO: 21epoch:train:6196-6608batch: iter_time=1.784e-04, forward_time=0.157, loss_ctc=15.698, loss_att=7.194, acc=0.952, loss=9.745, backward_time=0.254, grad_norm=101.514, clip=100.000, loss_scale=2.882e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 16:42:46,835 (trainer:779) INFO: 21epoch:train:6609-7021batch: iter_time=1.848e-04, forward_time=0.157, loss_ctc=15.728, loss_att=7.189, acc=0.955, loss=9.751, backward_time=0.254, grad_norm=97.531, clip=100.000, loss_scale=5.261e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 16:46:51,388 (trainer:779) INFO: 21epoch:train:7022-7434batch: iter_time=1.709e-04, forward_time=0.157, loss_ctc=15.477, loss_att=7.065, acc=0.955, loss=9.589, backward_time=0.254, grad_norm=96.308, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 16:50:54,531 (trainer:779) INFO: 21epoch:train:7435-7847batch: iter_time=1.913e-04, forward_time=0.155, loss_ctc=15.877, loss_att=7.240, acc=0.959, loss=9.831, backward_time=0.254, grad_norm=108.783, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 16:54:59,084 (trainer:779) INFO: 21epoch:train:7848-8260batch: iter_time=1.788e-04, forward_time=0.158, loss_ctc=15.419, loss_att=7.053, acc=0.955, loss=9.563, backward_time=0.255, grad_norm=97.693, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 16:55:59,763 (trainer:365) INFO: 21epoch results: [train] iter_time=2.196e-04, forward_time=0.157, loss_ctc=15.696, loss_att=7.180, acc=0.955, loss=9.735, backward_time=0.254, grad_norm=100.325, clip=100.000, loss_scale=2.968e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185, time=1 hour, 21 minutes and 39.99 seconds, total_count=173607, gpu_max_cached_mem_GB=22.527, [valid] loss_ctc=9.917, cer_ctc=0.048, loss_att=6.033, acc=0.945, cer=0.035, wer=0.512, loss=7.198, time=17.43 seconds, total_count=714, gpu_max_cached_mem_GB=22.527, [att_plot] time=38.42 seconds, total_count=0, gpu_max_cached_mem_GB=22.527
+[bmi2:0/4] 2024-07-07 16:56:03,788 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 16:56:03,846 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/12epoch.pth
+[bmi2:0/4] 2024-07-07 16:56:03,846 (trainer:299) INFO: 22/100epoch started. Estimated time to finish: 4 days, 13 hours and 58 minutes
+[bmi2:0/4] 2024-07-07 17:00:15,255 (trainer:779) INFO: 22epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=15.125, loss_att=6.899, acc=0.955, loss=9.367, backward_time=0.254, grad_norm=103.457, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.218
+[bmi2:0/4] 2024-07-07 17:04:19,554 (trainer:779) INFO: 22epoch:train:414-826batch: iter_time=1.736e-04, forward_time=0.157, loss_ctc=15.410, loss_att=7.019, acc=0.958, loss=9.536, backward_time=0.254, grad_norm=97.083, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 17:08:23,691 (trainer:779) INFO: 22epoch:train:827-1239batch: iter_time=1.828e-04, forward_time=0.158, loss_ctc=15.478, loss_att=7.094, acc=0.959, loss=9.609, backward_time=0.255, grad_norm=108.073, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 17:12:27,907 (trainer:779) INFO: 22epoch:train:1240-1652batch: iter_time=1.760e-04, forward_time=0.157, loss_ctc=15.147, loss_att=6.906, acc=0.954, loss=9.378, backward_time=0.255, grad_norm=106.415, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 17:16:31,264 (trainer:779) INFO: 22epoch:train:1653-2065batch: iter_time=1.780e-04, forward_time=0.157, loss_ctc=15.272, loss_att=6.982, acc=0.957, loss=9.469, backward_time=0.254, grad_norm=98.026, clip=100.000, loss_scale=5.765e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 17:20:36,078 (trainer:779) INFO: 22epoch:train:2066-2478batch: iter_time=1.924e-04, forward_time=0.158, loss_ctc=15.460, loss_att=7.031, acc=0.957, loss=9.560, backward_time=0.254, grad_norm=99.951, clip=100.000, loss_scale=6.656e+17, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 17:24:38,784 (trainer:779) INFO: 22epoch:train:2479-2891batch: iter_time=1.907e-04, forward_time=0.156, loss_ctc=15.418, loss_att=7.028, acc=0.956, loss=9.545, backward_time=0.253, grad_norm=97.531, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 17:28:42,548 (trainer:779) INFO: 22epoch:train:2892-3304batch: iter_time=1.757e-04, forward_time=0.157, loss_ctc=15.333, loss_att=7.006, acc=0.955, loss=9.504, backward_time=0.253, grad_norm=99.057, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 17:32:45,985 (trainer:779) INFO: 22epoch:train:3305-3717batch: iter_time=1.787e-04, forward_time=0.156, loss_ctc=15.721, loss_att=7.124, acc=0.959, loss=9.703, backward_time=0.254, grad_norm=93.889, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 17:36:49,983 (trainer:779) INFO: 22epoch:train:3718-4130batch: iter_time=1.847e-04, forward_time=0.158, loss_ctc=15.211, loss_att=6.938, acc=0.955, loss=9.420, backward_time=0.253, grad_norm=97.834, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 17:40:53,549 (trainer:779) INFO: 22epoch:train:4131-4543batch: iter_time=1.747e-04, forward_time=0.156, loss_ctc=15.575, loss_att=7.103, acc=0.957, loss=9.645, backward_time=0.254, grad_norm=109.207, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 17:44:58,196 (trainer:779) INFO: 22epoch:train:4544-4956batch: iter_time=1.878e-04, forward_time=0.157, loss_ctc=15.327, loss_att=6.994, acc=0.958, loss=9.494, backward_time=0.255, grad_norm=101.925, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 17:49:02,180 (trainer:779) INFO: 22epoch:train:4957-5369batch: iter_time=1.788e-04, forward_time=0.157, loss_ctc=15.310, loss_att=6.963, acc=0.956, loss=9.467, backward_time=0.254, grad_norm=100.757, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.182
+[bmi2:0/4] 2024-07-07 17:53:06,699 (trainer:779) INFO: 22epoch:train:5370-5782batch: iter_time=1.837e-04, forward_time=0.157, loss_ctc=15.363, loss_att=7.012, acc=0.955, loss=9.518, backward_time=0.254, grad_norm=104.441, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 17:57:11,408 (trainer:779) INFO: 22epoch:train:5783-6195batch: iter_time=1.787e-04, forward_time=0.158, loss_ctc=15.657, loss_att=7.137, acc=0.960, loss=9.693, backward_time=0.255, grad_norm=98.214, clip=100.000, loss_scale=1.153e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 18:01:16,814 (trainer:779) INFO: 22epoch:train:6196-6608batch: iter_time=1.798e-04, forward_time=0.158, loss_ctc=15.368, loss_att=6.999, acc=0.956, loss=9.510, backward_time=0.255, grad_norm=97.837, clip=100.000, loss_scale=1.693e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 18:05:20,308 (trainer:779) INFO: 22epoch:train:6609-7021batch: iter_time=1.858e-04, forward_time=0.157, loss_ctc=15.114, loss_att=6.892, acc=0.953, loss=9.359, backward_time=0.254, grad_norm=96.801, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 18:09:23,725 (trainer:779) INFO: 22epoch:train:7022-7434batch: iter_time=1.871e-04, forward_time=0.157, loss_ctc=15.371, loss_att=7.027, acc=0.955, loss=9.530, backward_time=0.253, grad_norm=95.445, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 18:13:28,871 (trainer:779) INFO: 22epoch:train:7435-7847batch: iter_time=1.762e-04, forward_time=0.159, loss_ctc=15.263, loss_att=6.967, acc=0.956, loss=9.456, backward_time=0.255, grad_norm=98.859, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.188
+[bmi2:0/4] 2024-07-07 18:17:34,098 (trainer:779) INFO: 22epoch:train:7848-8260batch: iter_time=1.846e-04, forward_time=0.158, loss_ctc=15.313, loss_att=6.970, acc=0.957, loss=9.473, backward_time=0.255, grad_norm=97.350, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 18:18:32,010 (trainer:365) INFO: 22epoch results: [train] iter_time=2.262e-04, forward_time=0.157, loss_ctc=15.361, loss_att=7.004, acc=0.956, loss=9.511, backward_time=0.254, grad_norm=100.099, clip=100.000, loss_scale=1.243e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184, time=1 hour, 21 minutes and 35.09 seconds, total_count=181874, gpu_max_cached_mem_GB=22.527, [valid] loss_ctc=9.736, cer_ctc=0.047, loss_att=5.766, acc=0.947, cer=0.034, wer=0.507, loss=6.957, time=17.03 seconds, total_count=748, gpu_max_cached_mem_GB=22.527, [att_plot] time=36.03 seconds, total_count=0, gpu_max_cached_mem_GB=22.527
+[bmi2:0/4] 2024-07-07 18:18:35,891 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 18:18:35,929 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/11epoch.pth
+[bmi2:0/4] 2024-07-07 18:18:35,929 (trainer:299) INFO: 23/100epoch started. Estimated time to finish: 4 days, 12 hours and 31 minutes
+[bmi2:0/4] 2024-07-07 18:22:48,192 (trainer:779) INFO: 23epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=14.715, loss_att=6.713, acc=0.955, loss=9.114, backward_time=0.254, grad_norm=101.230, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.222
+[bmi2:0/4] 2024-07-07 18:26:52,109 (trainer:779) INFO: 23epoch:train:414-826batch: iter_time=1.853e-04, forward_time=0.158, loss_ctc=15.107, loss_att=6.838, acc=0.960, loss=9.319, backward_time=0.254, grad_norm=95.755, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 18:30:56,592 (trainer:779) INFO: 23epoch:train:827-1239batch: iter_time=1.733e-04, forward_time=0.158, loss_ctc=14.925, loss_att=6.771, acc=0.957, loss=9.217, backward_time=0.255, grad_norm=101.394, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 18:35:01,574 (trainer:779) INFO: 23epoch:train:1240-1652batch: iter_time=1.717e-04, forward_time=0.158, loss_ctc=14.957, loss_att=6.779, acc=0.958, loss=9.232, backward_time=0.255, grad_norm=95.189, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 18:39:03,480 (trainer:779) INFO: 23epoch:train:1653-2065batch: iter_time=1.806e-04, forward_time=0.155, loss_ctc=14.884, loss_att=6.770, acc=0.958, loss=9.204, backward_time=0.253, grad_norm=90.549, clip=100.000, loss_scale=2.306e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.172
+[bmi2:0/4] 2024-07-07 18:43:07,986 (trainer:779) INFO: 23epoch:train:2066-2478batch: iter_time=1.827e-04, forward_time=0.158, loss_ctc=15.023, loss_att=6.850, acc=0.956, loss=9.302, backward_time=0.255, grad_norm=99.313, clip=100.000, loss_scale=4.144e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 18:47:10,332 (trainer:779) INFO: 23epoch:train:2479-2891batch: iter_time=1.844e-04, forward_time=0.156, loss_ctc=15.298, loss_att=6.944, acc=0.960, loss=9.451, backward_time=0.254, grad_norm=106.468, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.174
+[bmi2:0/4] 2024-07-07 18:51:13,927 (trainer:779) INFO: 23epoch:train:2892-3304batch: iter_time=1.777e-04, forward_time=0.157, loss_ctc=15.224, loss_att=6.918, acc=0.961, loss=9.410, backward_time=0.254, grad_norm=100.613, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 18:55:17,565 (trainer:779) INFO: 23epoch:train:3305-3717batch: iter_time=1.884e-04, forward_time=0.157, loss_ctc=15.064, loss_att=6.853, acc=0.959, loss=9.317, backward_time=0.254, grad_norm=106.292, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 18:59:22,900 (trainer:779) INFO: 23epoch:train:3718-4130batch: iter_time=1.777e-04, forward_time=0.158, loss_ctc=15.157, loss_att=6.992, acc=0.960, loss=9.441, backward_time=0.255, grad_norm=109.342, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 19:03:27,013 (trainer:779) INFO: 23epoch:train:4131-4543batch: iter_time=1.885e-04, forward_time=0.158, loss_ctc=14.935, loss_att=6.793, acc=0.960, loss=9.236, backward_time=0.255, grad_norm=96.374, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 19:07:31,482 (trainer:779) INFO: 23epoch:train:4544-4956batch: iter_time=1.731e-04, forward_time=0.158, loss_ctc=15.169, loss_att=6.872, acc=0.957, loss=9.361, backward_time=0.254, grad_norm=92.548, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 19:11:34,937 (trainer:779) INFO: 23epoch:train:4957-5369batch: iter_time=1.778e-04, forward_time=0.157, loss_ctc=14.685, loss_att=6.697, acc=0.951, loss=9.093, backward_time=0.254, grad_norm=96.265, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 19:15:39,319 (trainer:779) INFO: 23epoch:train:5370-5782batch: iter_time=1.754e-04, forward_time=0.157, loss_ctc=14.991, loss_att=6.782, acc=0.958, loss=9.245, backward_time=0.254, grad_norm=102.925, clip=100.000, loss_scale=4.612e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183
+[bmi2:0/4] 2024-07-07 19:19:43,929 (trainer:779) INFO: 23epoch:train:5783-6195batch: iter_time=1.749e-04, forward_time=0.159, loss_ctc=14.904, loss_att=6.801, acc=0.956, loss=9.232, backward_time=0.255, grad_norm=98.436, clip=100.000, loss_scale=5.127e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.185
+[bmi2:0/4] 2024-07-07 19:23:47,416 (trainer:779) INFO: 23epoch:train:6196-6608batch: iter_time=1.753e-04, forward_time=0.156, loss_ctc=14.982, loss_att=6.808, acc=0.958, loss=9.260, backward_time=0.254, grad_norm=95.762, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 19:27:51,128 (trainer:779) INFO: 23epoch:train:6609-7021batch: iter_time=1.804e-04, forward_time=0.157, loss_ctc=15.271, loss_att=6.933, acc=0.959, loss=9.434, backward_time=0.253, grad_norm=96.490, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 19:31:54,518 (trainer:779) INFO: 23epoch:train:7022-7434batch: iter_time=1.782e-04, forward_time=0.156, loss_ctc=14.932, loss_att=6.791, acc=0.957, loss=9.233, backward_time=0.254, grad_norm=94.945, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 19:35:59,467 (trainer:779) INFO: 23epoch:train:7435-7847batch: iter_time=1.793e-04, forward_time=0.158, loss_ctc=14.875, loss_att=6.792, acc=0.956, loss=9.217, backward_time=0.255, grad_norm=98.062, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 19:40:01,570 (trainer:779) INFO: 23epoch:train:7848-8260batch: iter_time=1.733e-04, forward_time=0.156, loss_ctc=15.080, loss_att=6.888, acc=0.956, loss=9.346, backward_time=0.252, grad_norm=97.666, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.172
+[bmi2:0/4] 2024-07-07 19:40:59,044 (trainer:365) INFO: 23epoch results: [train] iter_time=2.215e-04, forward_time=0.157, loss_ctc=15.007, loss_att=6.828, acc=0.958, loss=9.282, backward_time=0.254, grad_norm=98.786, clip=100.000, loss_scale=5.194e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.183, time=1 hour, 21 minutes and 30.58 seconds, total_count=190141, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.700, cer_ctc=0.046, loss_att=6.008, acc=0.944, cer=0.034, wer=0.504, loss=7.116, time=17.44 seconds, total_count=782, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.09 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-07 19:41:03,816 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 19:41:03,859 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/13epoch.pth
+[bmi2:0/4] 2024-07-07 19:41:03,859 (trainer:299) INFO: 24/100epoch started. Estimated time to finish: 4 days, 11 hours and 4 minutes
+[bmi2:0/4] 2024-07-07 19:45:16,828 (trainer:779) INFO: 24epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=14.364, loss_att=6.524, acc=0.956, loss=8.876, backward_time=0.254, grad_norm=93.654, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.225
+[bmi2:0/4] 2024-07-07 19:49:22,042 (trainer:779) INFO: 24epoch:train:414-826batch: iter_time=1.845e-04, forward_time=0.159, loss_ctc=14.905, loss_att=6.719, acc=0.964, loss=9.175, backward_time=0.255, grad_norm=100.835, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.187
+[bmi2:0/4] 2024-07-07 19:53:25,492 (trainer:779) INFO: 24epoch:train:827-1239batch: iter_time=1.698e-04, forward_time=0.157, loss_ctc=14.567, loss_att=6.588, acc=0.956, loss=8.982, backward_time=0.253, grad_norm=95.723, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 19:57:28,522 (trainer:779) INFO: 24epoch:train:1240-1652batch: iter_time=1.810e-04, forward_time=0.156, loss_ctc=14.808, loss_att=6.672, acc=0.960, loss=9.113, backward_time=0.254, grad_norm=95.646, clip=100.000, loss_scale=9.223e+18, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.176
+[bmi2:0/4] 2024-07-07 20:01:32,241 (trainer:779) INFO: 24epoch:train:1653-2065batch: iter_time=1.765e-04, forward_time=0.157, loss_ctc=14.755, loss_att=6.649, acc=0.957, loss=9.081, backward_time=0.254, grad_norm=96.343, clip=100.000, loss_scale=1.330e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 20:05:35,544 (trainer:779) INFO: 24epoch:train:2066-2478batch: iter_time=1.880e-04, forward_time=0.157, loss_ctc=14.994, loss_att=6.780, acc=0.960, loss=9.244, backward_time=0.253, grad_norm=103.727, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 20:09:38,052 (trainer:779) INFO: 24epoch:train:2479-2891batch: iter_time=1.783e-04, forward_time=0.156, loss_ctc=14.945, loss_att=6.737, acc=0.958, loss=9.199, backward_time=0.253, grad_norm=97.624, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 20:13:42,027 (trainer:779) INFO: 24epoch:train:2892-3304batch: iter_time=1.853e-04, forward_time=0.157, loss_ctc=14.471, loss_att=6.570, acc=0.954, loss=8.940, backward_time=0.253, grad_norm=98.406, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.181
+[bmi2:0/4] 2024-07-07 20:17:45,087 (trainer:779) INFO: 24epoch:train:3305-3717batch: iter_time=1.750e-04, forward_time=0.157, loss_ctc=14.597, loss_att=6.651, acc=0.958, loss=9.035, backward_time=0.254, grad_norm=100.393, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 20:21:47,946 (trainer:779) INFO: 24epoch:train:3718-4130batch: iter_time=1.876e-04, forward_time=0.156, loss_ctc=14.773, loss_att=6.672, acc=0.956, loss=9.102, backward_time=0.253, grad_norm=94.593, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 20:25:50,869 (trainer:779) INFO: 24epoch:train:4131-4543batch: iter_time=1.868e-04, forward_time=0.156, loss_ctc=14.966, loss_att=6.756, acc=0.956, loss=9.219, backward_time=0.253, grad_norm=95.167, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.177
+[bmi2:0/4] 2024-07-07 20:29:54,351 (trainer:779) INFO: 24epoch:train:4544-4956batch: iter_time=1.731e-04, forward_time=0.157, loss_ctc=14.754, loss_att=6.682, acc=0.960, loss=9.104, backward_time=0.254, grad_norm=93.289, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 20:33:57,857 (trainer:779) INFO: 24epoch:train:4957-5369batch: iter_time=1.915e-04, forward_time=0.157, loss_ctc=14.929, loss_att=6.765, acc=0.960, loss=9.214, backward_time=0.254, grad_norm=99.219, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 20:38:01,352 (trainer:779) INFO: 24epoch:train:5370-5782batch: iter_time=1.815e-04, forward_time=0.156, loss_ctc=14.851, loss_att=6.717, acc=0.959, loss=9.157, backward_time=0.253, grad_norm=97.845, clip=100.000, loss_scale=1.845e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 20:42:05,762 (trainer:779) INFO: 24epoch:train:5783-6195batch: iter_time=1.901e-04, forward_time=0.158, loss_ctc=14.757, loss_att=6.688, acc=0.960, loss=9.109, backward_time=0.254, grad_norm=96.626, clip=100.000, loss_scale=3.242e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.184
+[bmi2:0/4] 2024-07-07 20:46:09,482 (trainer:779) INFO: 24epoch:train:6196-6608batch: iter_time=1.861e-04, forward_time=0.156, loss_ctc=14.890, loss_att=6.735, acc=0.961, loss=9.181, backward_time=0.253, grad_norm=99.109, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 20:50:14,363 (trainer:779) INFO: 24epoch:train:6609-7021batch: iter_time=1.786e-04, forward_time=0.159, loss_ctc=14.745, loss_att=6.685, acc=0.958, loss=9.103, backward_time=0.255, grad_norm=98.774, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.186
+[bmi2:0/4] 2024-07-07 20:54:17,927 (trainer:779) INFO: 24epoch:train:7022-7434batch: iter_time=1.751e-04, forward_time=0.157, loss_ctc=14.870, loss_att=6.789, acc=0.957, loss=9.213, backward_time=0.253, grad_norm=108.501, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 20:58:20,517 (trainer:779) INFO: 24epoch:train:7435-7847batch: iter_time=1.747e-04, forward_time=0.156, loss_ctc=14.585, loss_att=6.641, acc=0.957, loss=9.024, backward_time=0.253, grad_norm=96.428, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 21:02:23,281 (trainer:779) INFO: 24epoch:train:7848-8260batch: iter_time=1.806e-04, forward_time=0.156, loss_ctc=14.765, loss_att=6.677, acc=0.960, loss=9.103, backward_time=0.253, grad_norm=93.702, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.175
+[bmi2:0/4] 2024-07-07 21:03:20,963 (trainer:365) INFO: 24epoch results: [train] iter_time=2.327e-04, forward_time=0.157, loss_ctc=14.763, loss_att=6.684, acc=0.958, loss=9.108, backward_time=0.254, grad_norm=97.773, clip=100.000, loss_scale=2.167e+19, optim_step_time=0.028, optim0_lr0=0.001, train_time=1.181, time=1 hour, 21 minutes and 24.08 seconds, total_count=198408, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.555, cer_ctc=0.046, loss_att=5.574, acc=0.948, cer=0.034, wer=0.502, loss=6.769, time=17.7 seconds, total_count=816, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.33 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-07 21:03:26,255 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 21:03:26,320 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/16epoch.pth
+[bmi2:0/4] 2024-07-07 21:03:26,321 (trainer:299) INFO: 25/100epoch started. Estimated time to finish: 4 days, 9 hours and 37 minutes
+[bmi2:0/4] 2024-07-07 21:07:39,120 (trainer:779) INFO: 25epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=14.387, loss_att=6.477, acc=0.958, loss=8.850, backward_time=0.253, grad_norm=100.661, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.225
+[bmi2:0/4] 2024-07-07 21:11:42,590 (trainer:779) INFO: 25epoch:train:414-826batch: iter_time=1.820e-04, forward_time=0.158, loss_ctc=14.602, loss_att=6.565, acc=0.959, loss=8.976, backward_time=0.254, grad_norm=117.077, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.178
+[bmi2:0/4] 2024-07-07 21:15:46,214 (trainer:779) INFO: 25epoch:train:827-1239batch: iter_time=1.860e-04, forward_time=0.158, loss_ctc=14.589, loss_att=6.633, acc=0.960, loss=9.020, backward_time=0.254, grad_norm=102.207, clip=100.000, loss_scale=3.689e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.180
+[bmi2:0/4] 2024-07-07 21:19:49,904 (trainer:779) INFO: 25epoch:train:1240-1652batch: iter_time=1.848e-04, forward_time=0.157, loss_ctc=14.729, loss_att=6.645, acc=0.962, loss=9.070, backward_time=0.254, grad_norm=101.691, clip=100.000, loss_scale=4.010e+19, optim_step_time=0.029, optim0_lr0=0.001, train_time=1.179
+[bmi2:0/4] 2024-07-07 21:23:54,053 (trainer:779) INFO: 25epoch:train:1653-2065batch: iter_time=1.788e-04, forward_time=0.157, loss_ctc=14.691, loss_att=6.603, acc=0.959, loss=9.030, backward_time=0.254, grad_norm=104.160, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.028, optim0_lr0=9.994e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 21:27:56,956 (trainer:779) INFO: 25epoch:train:2066-2478batch: iter_time=1.751e-04, forward_time=0.156, loss_ctc=14.657, loss_att=6.607, acc=0.959, loss=9.022, backward_time=0.253, grad_norm=94.346, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.984e-04, train_time=1.175
+[bmi2:0/4] 2024-07-07 21:32:00,725 (trainer:779) INFO: 25epoch:train:2479-2891batch: iter_time=1.829e-04, forward_time=0.158, loss_ctc=14.633, loss_att=6.602, acc=0.962, loss=9.011, backward_time=0.254, grad_norm=101.999, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.973e-04, train_time=1.181
+[bmi2:0/4] 2024-07-07 21:36:04,639 (trainer:779) INFO: 25epoch:train:2892-3304batch: iter_time=1.762e-04, forward_time=0.157, loss_ctc=14.382, loss_att=6.497, acc=0.957, loss=8.862, backward_time=0.254, grad_norm=91.030, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.963e-04, train_time=1.180
+[bmi2:0/4] 2024-07-07 21:40:08,326 (trainer:779) INFO: 25epoch:train:3305-3717batch: iter_time=1.752e-04, forward_time=0.157, loss_ctc=14.345, loss_att=6.501, acc=0.958, loss=8.854, backward_time=0.254, grad_norm=102.005, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.028, optim0_lr0=9.953e-04, train_time=1.180
+[bmi2:0/4] 2024-07-07 21:44:12,885 (trainer:779) INFO: 25epoch:train:3718-4130batch: iter_time=1.823e-04, forward_time=0.157, loss_ctc=14.664, loss_att=6.587, acc=0.963, loss=9.010, backward_time=0.255, grad_norm=97.153, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.943e-04, train_time=1.184
+[bmi2:0/4] 2024-07-07 21:48:17,029 (trainer:779) INFO: 25epoch:train:4131-4543batch: iter_time=1.784e-04, forward_time=0.157, loss_ctc=14.670, loss_att=6.621, acc=0.963, loss=9.036, backward_time=0.254, grad_norm=92.976, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.933e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 21:52:20,258 (trainer:779) INFO: 25epoch:train:4544-4956batch: iter_time=1.821e-04, forward_time=0.156, loss_ctc=14.386, loss_att=6.491, acc=0.960, loss=8.859, backward_time=0.253, grad_norm=103.981, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.029, optim0_lr0=9.923e-04, train_time=1.177
+[bmi2:0/4] 2024-07-07 21:56:24,337 (trainer:779) INFO: 25epoch:train:4957-5369batch: iter_time=1.842e-04, forward_time=0.157, loss_ctc=14.227, loss_att=6.460, acc=0.953, loss=8.790, backward_time=0.255, grad_norm=99.896, clip=100.000, loss_scale=7.379e+19, optim_step_time=0.028, optim0_lr0=9.912e-04, train_time=1.182
+[bmi2:0/4] 2024-07-07 22:00:28,538 (trainer:779) INFO: 25epoch:train:5370-5782batch: iter_time=1.800e-04, forward_time=0.157, loss_ctc=14.388, loss_att=6.490, acc=0.958, loss=8.859, backward_time=0.254, grad_norm=96.194, clip=100.000, loss_scale=1.034e+20, optim_step_time=0.028, optim0_lr0=9.902e-04, train_time=1.182
+[bmi2:0/4] 2024-07-07 22:04:31,325 (trainer:779) INFO: 25epoch:train:5783-6195batch: iter_time=1.731e-04, forward_time=0.156, loss_ctc=14.270, loss_att=6.422, acc=0.958, loss=8.777, backward_time=0.253, grad_norm=98.395, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.892e-04, train_time=1.176
+[bmi2:0/4] 2024-07-07 22:08:34,733 (trainer:779) INFO: 25epoch:train:6196-6608batch: iter_time=1.746e-04, forward_time=0.156, loss_ctc=14.307, loss_att=6.475, acc=0.960, loss=8.824, backward_time=0.253, grad_norm=94.159, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.882e-04, train_time=1.178
+[bmi2:0/4] 2024-07-07 22:12:38,261 (trainer:779) INFO: 25epoch:train:6609-7021batch: iter_time=1.938e-04, forward_time=0.158, loss_ctc=14.428, loss_att=6.522, acc=0.956, loss=8.894, backward_time=0.254, grad_norm=101.638, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.872e-04, train_time=1.180
+[bmi2:0/4] 2024-07-07 22:16:42,329 (trainer:779) INFO: 25epoch:train:7022-7434batch: iter_time=1.762e-04, forward_time=0.156, loss_ctc=14.772, loss_att=6.686, acc=0.963, loss=9.112, backward_time=0.254, grad_norm=105.018, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.863e-04, train_time=1.181
+[bmi2:0/4] 2024-07-07 22:20:46,285 (trainer:779) INFO: 25epoch:train:7435-7847batch: iter_time=1.814e-04, forward_time=0.157, loss_ctc=14.499, loss_att=6.573, acc=0.957, loss=8.951, backward_time=0.254, grad_norm=100.954, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.853e-04, train_time=1.182
+[bmi2:0/4] 2024-07-07 22:24:50,802 (trainer:779) INFO: 25epoch:train:7848-8260batch: iter_time=1.841e-04, forward_time=0.158, loss_ctc=14.546, loss_att=6.589, acc=0.961, loss=8.976, backward_time=0.255, grad_norm=103.840, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.843e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 22:25:47,187 (trainer:365) INFO: 25epoch results: [train] iter_time=2.221e-04, forward_time=0.157, loss_ctc=14.505, loss_att=6.551, acc=0.959, loss=8.937, backward_time=0.254, grad_norm=100.478, clip=100.000, loss_scale=9.023e+19, optim_step_time=0.029, optim0_lr0=9.938e-04, train_time=1.183, time=1 hour, 21 minutes and 29.24 seconds, total_count=206675, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.885, cer_ctc=0.047, loss_att=5.758, acc=0.947, cer=0.034, wer=0.509, loss=6.997, time=16.97 seconds, total_count=850, gpu_max_cached_mem_GB=22.529, [att_plot] time=34.66 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-07 22:25:51,907 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-07 22:25:51,949 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/14epoch.pth
+[bmi2:0/4] 2024-07-07 22:25:51,950 (trainer:299) INFO: 26/100epoch started. Estimated time to finish: 4 days, 8 hours and 11 minutes
+[bmi2:0/4] 2024-07-07 22:30:03,536 (trainer:779) INFO: 26epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=14.062, loss_att=6.330, acc=0.958, loss=8.650, backward_time=0.253, grad_norm=101.267, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.833e-04, train_time=1.219
+[bmi2:0/4] 2024-07-07 22:34:07,954 (trainer:779) INFO: 26epoch:train:414-826batch: iter_time=1.895e-04, forward_time=0.158, loss_ctc=14.342, loss_att=6.452, acc=0.960, loss=8.819, backward_time=0.254, grad_norm=96.220, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.029, optim0_lr0=9.823e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 22:38:11,160 (trainer:779) INFO: 26epoch:train:827-1239batch: iter_time=1.760e-04, forward_time=0.156, loss_ctc=14.295, loss_att=6.453, acc=0.961, loss=8.806, backward_time=0.254, grad_norm=102.981, clip=100.000, loss_scale=1.476e+20, optim_step_time=0.028, optim0_lr0=9.813e-04, train_time=1.178
+[bmi2:0/4] 2024-07-07 22:42:13,339 (trainer:779) INFO: 26epoch:train:1240-1652batch: iter_time=1.887e-04, forward_time=0.156, loss_ctc=14.370, loss_att=6.448, acc=0.963, loss=8.825, backward_time=0.253, grad_norm=95.414, clip=100.000, loss_scale=2.552e+20, optim_step_time=0.028, optim0_lr0=9.803e-04, train_time=1.172
+[bmi2:0/4] 2024-07-07 22:46:17,164 (trainer:779) INFO: 26epoch:train:1653-2065batch: iter_time=1.848e-04, forward_time=0.158, loss_ctc=14.325, loss_att=6.467, acc=0.960, loss=8.824, backward_time=0.254, grad_norm=110.191, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.794e-04, train_time=1.181
+[bmi2:0/4] 2024-07-07 22:50:20,783 (trainer:779) INFO: 26epoch:train:2066-2478batch: iter_time=1.707e-04, forward_time=0.157, loss_ctc=14.037, loss_att=6.338, acc=0.959, loss=8.648, backward_time=0.254, grad_norm=94.425, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.784e-04, train_time=1.179
+[bmi2:0/4] 2024-07-07 22:54:25,615 (trainer:779) INFO: 26epoch:train:2479-2891batch: iter_time=1.819e-04, forward_time=0.158, loss_ctc=14.233, loss_att=6.405, acc=0.961, loss=8.754, backward_time=0.256, grad_norm=90.223, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.774e-04, train_time=1.186
+[bmi2:0/4] 2024-07-07 22:58:30,249 (trainer:779) INFO: 26epoch:train:2892-3304batch: iter_time=1.770e-04, forward_time=0.157, loss_ctc=14.039, loss_att=6.350, acc=0.960, loss=8.657, backward_time=0.255, grad_norm=93.497, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.765e-04, train_time=1.184
+[bmi2:0/4] 2024-07-07 23:02:34,499 (trainer:779) INFO: 26epoch:train:3305-3717batch: iter_time=1.790e-04, forward_time=0.157, loss_ctc=14.230, loss_att=6.396, acc=0.963, loss=8.746, backward_time=0.254, grad_norm=90.401, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.755e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 23:06:39,474 (trainer:779) INFO: 26epoch:train:3718-4130batch: iter_time=1.703e-04, forward_time=0.159, loss_ctc=14.006, loss_att=6.323, acc=0.959, loss=8.628, backward_time=0.254, grad_norm=87.746, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.746e-04, train_time=1.186
+[bmi2:0/4] 2024-07-07 23:10:43,669 (trainer:779) INFO: 26epoch:train:4131-4543batch: iter_time=1.725e-04, forward_time=0.157, loss_ctc=14.214, loss_att=6.427, acc=0.958, loss=8.763, backward_time=0.255, grad_norm=88.323, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.736e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 23:14:48,317 (trainer:779) INFO: 26epoch:train:4544-4956batch: iter_time=1.859e-04, forward_time=0.157, loss_ctc=14.117, loss_att=6.369, acc=0.960, loss=8.693, backward_time=0.254, grad_norm=91.780, clip=100.000, loss_scale=2.951e+20, optim_step_time=0.029, optim0_lr0=9.727e-04, train_time=1.184
+[bmi2:0/4] 2024-07-07 23:18:50,310 (trainer:779) INFO: 26epoch:train:4957-5369batch: iter_time=1.916e-04, forward_time=0.155, loss_ctc=14.252, loss_att=6.398, acc=0.963, loss=8.754, backward_time=0.253, grad_norm=96.392, clip=100.000, loss_scale=3.080e+20, optim_step_time=0.028, optim0_lr0=9.717e-04, train_time=1.172
+[bmi2:0/4] 2024-07-07 23:22:53,692 (trainer:779) INFO: 26epoch:train:5370-5782batch: iter_time=1.750e-04, forward_time=0.157, loss_ctc=14.266, loss_att=6.431, acc=0.962, loss=8.781, backward_time=0.253, grad_norm=95.490, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.708e-04, train_time=1.178
+[bmi2:0/4] 2024-07-07 23:26:57,555 (trainer:779) INFO: 26epoch:train:5783-6195batch: iter_time=1.803e-04, forward_time=0.157, loss_ctc=14.253, loss_att=6.409, acc=0.963, loss=8.762, backward_time=0.254, grad_norm=93.123, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.028, optim0_lr0=9.698e-04, train_time=1.181
+[bmi2:0/4] 2024-07-07 23:31:01,787 (trainer:779) INFO: 26epoch:train:6196-6608batch: iter_time=1.900e-04, forward_time=0.157, loss_ctc=14.294, loss_att=6.428, acc=0.962, loss=8.788, backward_time=0.254, grad_norm=106.582, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.689e-04, train_time=1.182
+[bmi2:0/4] 2024-07-07 23:35:05,728 (trainer:779) INFO: 26epoch:train:6609-7021batch: iter_time=1.741e-04, forward_time=0.158, loss_ctc=14.230, loss_att=6.454, acc=0.960, loss=8.787, backward_time=0.254, grad_norm=96.656, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.679e-04, train_time=1.182
+[bmi2:0/4] 2024-07-07 23:39:08,628 (trainer:779) INFO: 26epoch:train:7022-7434batch: iter_time=1.847e-04, forward_time=0.156, loss_ctc=14.019, loss_att=6.326, acc=0.954, loss=8.634, backward_time=0.253, grad_norm=111.030, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.670e-04, train_time=1.175
+[bmi2:0/4] 2024-07-07 23:43:12,476 (trainer:779) INFO: 26epoch:train:7435-7847batch: iter_time=1.777e-04, forward_time=0.157, loss_ctc=14.249, loss_att=6.430, acc=0.958, loss=8.775, backward_time=0.254, grad_norm=112.595, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.028, optim0_lr0=9.661e-04, train_time=1.181
+[bmi2:0/4] 2024-07-07 23:47:16,920 (trainer:779) INFO: 26epoch:train:7848-8260batch: iter_time=1.893e-04, forward_time=0.158, loss_ctc=14.540, loss_att=6.575, acc=0.960, loss=8.964, backward_time=0.254, grad_norm=110.635, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.028, optim0_lr0=9.651e-04, train_time=1.183
+[bmi2:0/4] 2024-07-07 23:48:13,831 (trainer:365) INFO: 26epoch results: [train] iter_time=2.325e-04, forward_time=0.157, loss_ctc=14.216, loss_att=6.409, acc=0.960, loss=8.751, backward_time=0.254, grad_norm=98.252, clip=100.000, loss_scale=3.752e+20, optim_step_time=0.029, optim0_lr0=9.741e-04, train_time=1.183, time=1 hour, 21 minutes and 29.76 seconds, total_count=214942, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.686, cer_ctc=0.046, loss_att=5.609, acc=0.948, cer=0.034, wer=0.499, loss=6.832, time=17.48 seconds, total_count=884, gpu_max_cached_mem_GB=22.529, [att_plot] time=34.64 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-07 23:48:18,694 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-07 23:48:18,750 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/15epoch.pth
+[bmi2:0/4] 2024-07-07 23:48:18,750 (trainer:299) INFO: 27/100epoch started. Estimated time to finish: 4 days, 6 hours and 45 minutes
+[bmi2:0/4] 2024-07-07 23:52:31,337 (trainer:779) INFO: 27epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=14.131, loss_att=6.335, acc=0.965, loss=8.674, backward_time=0.254, grad_norm=105.214, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.642e-04, train_time=1.223
+[bmi2:0/4] 2024-07-07 23:56:35,835 (trainer:779) INFO: 27epoch:train:414-826batch: iter_time=1.732e-04, forward_time=0.158, loss_ctc=13.858, loss_att=6.284, acc=0.961, loss=8.556, backward_time=0.254, grad_norm=104.499, clip=100.000, loss_scale=5.903e+20, optim_step_time=0.029, optim0_lr0=9.633e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 00:00:39,292 (trainer:779) INFO: 27epoch:train:827-1239batch: iter_time=1.781e-04, forward_time=0.157, loss_ctc=14.119, loss_att=6.357, acc=0.963, loss=8.685, backward_time=0.254, grad_norm=110.879, clip=100.000, loss_scale=8.109e+20, optim_step_time=0.029, optim0_lr0=9.624e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 00:04:43,381 (trainer:779) INFO: 27epoch:train:1240-1652batch: iter_time=1.851e-04, forward_time=0.157, loss_ctc=13.740, loss_att=6.216, acc=0.959, loss=8.473, backward_time=0.255, grad_norm=104.942, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.614e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 00:08:47,275 (trainer:779) INFO: 27epoch:train:1653-2065batch: iter_time=1.797e-04, forward_time=0.157, loss_ctc=14.143, loss_att=6.343, acc=0.963, loss=8.683, backward_time=0.254, grad_norm=91.194, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.605e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 00:12:52,302 (trainer:779) INFO: 27epoch:train:2066-2478batch: iter_time=1.793e-04, forward_time=0.158, loss_ctc=14.163, loss_att=6.355, acc=0.962, loss=8.698, backward_time=0.255, grad_norm=102.193, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.596e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 00:16:55,838 (trainer:779) INFO: 27epoch:train:2479-2891batch: iter_time=1.759e-04, forward_time=0.156, loss_ctc=14.073, loss_att=6.300, acc=0.960, loss=8.632, backward_time=0.254, grad_norm=98.580, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.587e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 00:20:59,907 (trainer:779) INFO: 27epoch:train:2892-3304batch: iter_time=1.801e-04, forward_time=0.157, loss_ctc=14.120, loss_att=6.344, acc=0.962, loss=8.677, backward_time=0.254, grad_norm=96.123, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.578e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 00:25:03,566 (trainer:779) INFO: 27epoch:train:3305-3717batch: iter_time=1.761e-04, forward_time=0.158, loss_ctc=13.874, loss_att=6.258, acc=0.961, loss=8.543, backward_time=0.254, grad_norm=94.501, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.569e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 00:29:07,676 (trainer:779) INFO: 27epoch:train:3718-4130batch: iter_time=1.783e-04, forward_time=0.158, loss_ctc=13.857, loss_att=6.248, acc=0.957, loss=8.531, backward_time=0.254, grad_norm=96.473, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.560e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 00:33:12,253 (trainer:779) INFO: 27epoch:train:4131-4543batch: iter_time=1.811e-04, forward_time=0.158, loss_ctc=14.195, loss_att=6.349, acc=0.962, loss=8.703, backward_time=0.255, grad_norm=104.187, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.551e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 00:37:16,120 (trainer:779) INFO: 27epoch:train:4544-4956batch: iter_time=1.711e-04, forward_time=0.158, loss_ctc=13.771, loss_att=6.181, acc=0.959, loss=8.458, backward_time=0.254, grad_norm=101.359, clip=100.000, loss_scale=1.181e+21, optim_step_time=0.029, optim0_lr0=9.542e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 00:41:20,080 (trainer:779) INFO: 27epoch:train:4957-5369batch: iter_time=1.853e-04, forward_time=0.157, loss_ctc=13.764, loss_att=6.203, acc=0.959, loss=8.471, backward_time=0.255, grad_norm=95.922, clip=100.000, loss_scale=1.994e+21, optim_step_time=0.028, optim0_lr0=9.533e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 00:45:25,179 (trainer:779) INFO: 27epoch:train:5370-5782batch: iter_time=1.768e-04, forward_time=0.158, loss_ctc=14.071, loss_att=6.349, acc=0.961, loss=8.665, backward_time=0.255, grad_norm=94.411, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.524e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 00:49:28,041 (trainer:779) INFO: 27epoch:train:5783-6195batch: iter_time=1.763e-04, forward_time=0.156, loss_ctc=14.072, loss_att=6.345, acc=0.960, loss=8.663, backward_time=0.253, grad_norm=95.575, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.515e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 00:53:31,259 (trainer:779) INFO: 27epoch:train:6196-6608batch: iter_time=1.871e-04, forward_time=0.156, loss_ctc=13.868, loss_att=6.274, acc=0.959, loss=8.552, backward_time=0.253, grad_norm=97.471, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.028, optim0_lr0=9.506e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 00:57:36,272 (trainer:779) INFO: 27epoch:train:6609-7021batch: iter_time=1.755e-04, forward_time=0.158, loss_ctc=14.301, loss_att=6.410, acc=0.961, loss=8.778, backward_time=0.255, grad_norm=104.912, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.028, optim0_lr0=9.497e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 01:01:40,898 (trainer:779) INFO: 27epoch:train:7022-7434batch: iter_time=1.745e-04, forward_time=0.158, loss_ctc=13.980, loss_att=6.261, acc=0.961, loss=8.577, backward_time=0.254, grad_norm=103.791, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.488e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 01:05:46,500 (trainer:779) INFO: 27epoch:train:7435-7847batch: iter_time=1.813e-04, forward_time=0.159, loss_ctc=13.913, loss_att=6.243, acc=0.958, loss=8.544, backward_time=0.255, grad_norm=100.959, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.480e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 01:09:51,590 (trainer:779) INFO: 27epoch:train:7848-8260batch: iter_time=1.698e-04, forward_time=0.158, loss_ctc=13.898, loss_att=6.257, acc=0.963, loss=8.549, backward_time=0.255, grad_norm=98.322, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.471e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 01:10:51,715 (trainer:365) INFO: 27epoch results: [train] iter_time=2.240e-04, forward_time=0.157, loss_ctc=13.995, loss_att=6.295, acc=0.961, loss=8.605, backward_time=0.254, grad_norm=100.047, clip=100.000, loss_scale=1.558e+21, optim_step_time=0.029, optim0_lr0=9.556e-04, train_time=1.184, time=1 hour, 21 minutes and 37.59 seconds, total_count=223209, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.604, cer_ctc=0.045, loss_att=5.765, acc=0.946, cer=0.033, wer=0.501, loss=6.917, time=17.86 seconds, total_count=918, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.51 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 01:10:56,898 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 01:10:56,941 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/20epoch.pth
+[bmi2:0/4] 2024-07-08 01:10:56,941 (trainer:299) INFO: 28/100epoch started. Estimated time to finish: 4 days, 5 hours and 20 minutes
+[bmi2:0/4] 2024-07-08 01:15:10,181 (trainer:779) INFO: 28epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=13.683, loss_att=6.118, acc=0.962, loss=8.388, backward_time=0.254, grad_norm=95.394, clip=100.000, loss_scale=2.361e+21, optim_step_time=0.029, optim0_lr0=9.462e-04, train_time=1.227
+[bmi2:0/4] 2024-07-08 01:19:14,272 (trainer:779) INFO: 28epoch:train:414-826batch: iter_time=1.770e-04, forward_time=0.157, loss_ctc=13.769, loss_att=6.188, acc=0.961, loss=8.462, backward_time=0.254, grad_norm=94.479, clip=100.000, loss_scale=2.407e+21, optim_step_time=0.029, optim0_lr0=9.453e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 01:23:19,250 (trainer:779) INFO: 28epoch:train:827-1239batch: iter_time=1.903e-04, forward_time=0.158, loss_ctc=13.753, loss_att=6.135, acc=0.965, loss=8.421, backward_time=0.255, grad_norm=94.987, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.445e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 01:27:23,085 (trainer:779) INFO: 28epoch:train:1240-1652batch: iter_time=1.922e-04, forward_time=0.157, loss_ctc=13.603, loss_att=6.103, acc=0.962, loss=8.353, backward_time=0.254, grad_norm=88.300, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.028, optim0_lr0=9.436e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 01:31:26,140 (trainer:779) INFO: 28epoch:train:1653-2065batch: iter_time=1.736e-04, forward_time=0.156, loss_ctc=14.000, loss_att=6.264, acc=0.966, loss=8.585, backward_time=0.254, grad_norm=94.684, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.427e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 01:35:30,551 (trainer:779) INFO: 28epoch:train:2066-2478batch: iter_time=1.798e-04, forward_time=0.158, loss_ctc=13.640, loss_att=6.133, acc=0.963, loss=8.385, backward_time=0.255, grad_norm=97.551, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.419e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 01:39:34,031 (trainer:779) INFO: 28epoch:train:2479-2891batch: iter_time=1.765e-04, forward_time=0.157, loss_ctc=13.714, loss_att=6.144, acc=0.962, loss=8.415, backward_time=0.254, grad_norm=93.410, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.410e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 01:43:38,415 (trainer:779) INFO: 28epoch:train:2892-3304batch: iter_time=1.830e-04, forward_time=0.158, loss_ctc=13.687, loss_att=6.137, acc=0.958, loss=8.402, backward_time=0.255, grad_norm=94.231, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.401e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 01:47:42,866 (trainer:779) INFO: 28epoch:train:3305-3717batch: iter_time=1.757e-04, forward_time=0.157, loss_ctc=13.797, loss_att=6.204, acc=0.962, loss=8.482, backward_time=0.255, grad_norm=94.978, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.393e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 01:51:48,344 (trainer:779) INFO: 28epoch:train:3718-4130batch: iter_time=1.742e-04, forward_time=0.159, loss_ctc=13.695, loss_att=6.140, acc=0.963, loss=8.407, backward_time=0.255, grad_norm=101.628, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.384e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 01:55:51,883 (trainer:779) INFO: 28epoch:train:4131-4543batch: iter_time=1.724e-04, forward_time=0.157, loss_ctc=14.031, loss_att=6.256, acc=0.962, loss=8.589, backward_time=0.253, grad_norm=104.186, clip=100.000, loss_scale=4.722e+21, optim_step_time=0.029, optim0_lr0=9.376e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 01:59:54,814 (trainer:779) INFO: 28epoch:train:4544-4956batch: iter_time=1.810e-04, forward_time=0.156, loss_ctc=13.805, loss_att=6.172, acc=0.959, loss=8.462, backward_time=0.253, grad_norm=102.372, clip=100.000, loss_scale=6.296e+21, optim_step_time=0.029, optim0_lr0=9.367e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 02:03:59,621 (trainer:779) INFO: 28epoch:train:4957-5369batch: iter_time=1.842e-04, forward_time=0.158, loss_ctc=13.699, loss_att=6.153, acc=0.961, loss=8.417, backward_time=0.255, grad_norm=105.117, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.359e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 02:08:03,451 (trainer:779) INFO: 28epoch:train:5370-5782batch: iter_time=1.817e-04, forward_time=0.157, loss_ctc=13.633, loss_att=6.153, acc=0.958, loss=8.397, backward_time=0.254, grad_norm=98.221, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.350e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 02:12:06,537 (trainer:779) INFO: 28epoch:train:5783-6195batch: iter_time=1.881e-04, forward_time=0.157, loss_ctc=14.084, loss_att=6.291, acc=0.965, loss=8.629, backward_time=0.254, grad_norm=106.423, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.342e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 02:16:10,874 (trainer:779) INFO: 28epoch:train:6196-6608batch: iter_time=1.846e-04, forward_time=0.158, loss_ctc=13.978, loss_att=6.247, acc=0.961, loss=8.567, backward_time=0.254, grad_norm=124.518, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.333e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 02:20:15,323 (trainer:779) INFO: 28epoch:train:6609-7021batch: iter_time=1.786e-04, forward_time=0.159, loss_ctc=13.813, loss_att=6.194, acc=0.961, loss=8.480, backward_time=0.254, grad_norm=104.108, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.325e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 02:24:18,170 (trainer:779) INFO: 28epoch:train:7022-7434batch: iter_time=1.787e-04, forward_time=0.156, loss_ctc=14.114, loss_att=6.298, acc=0.962, loss=8.643, backward_time=0.254, grad_norm=102.284, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.317e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 02:28:21,135 (trainer:779) INFO: 28epoch:train:7435-7847batch: iter_time=1.911e-04, forward_time=0.156, loss_ctc=13.887, loss_att=6.204, acc=0.960, loss=8.509, backward_time=0.253, grad_norm=94.792, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.308e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 02:32:24,459 (trainer:779) INFO: 28epoch:train:7848-8260batch: iter_time=1.787e-04, forward_time=0.156, loss_ctc=13.798, loss_att=6.183, acc=0.959, loss=8.467, backward_time=0.253, grad_norm=95.493, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.300e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 02:33:19,059 (trainer:365) INFO: 28epoch results: [train] iter_time=2.287e-04, forward_time=0.157, loss_ctc=13.809, loss_att=6.186, acc=0.962, loss=8.473, backward_time=0.254, grad_norm=99.370, clip=100.000, loss_scale=6.459e+21, optim_step_time=0.029, optim0_lr0=9.380e-04, train_time=1.183, time=1 hour, 21 minutes and 32.32 seconds, total_count=231476, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.711, cer_ctc=0.046, loss_att=5.725, acc=0.947, cer=0.034, wer=0.503, loss=6.921, time=17.49 seconds, total_count=952, gpu_max_cached_mem_GB=22.529, [att_plot] time=32.3 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 02:33:23,355 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 02:33:23,423 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/23epoch.pth
+[bmi2:0/4] 2024-07-08 02:33:23,423 (trainer:299) INFO: 29/100epoch started. Estimated time to finish: 4 days, 3 hours and 54 minutes
+[bmi2:0/4] 2024-07-08 02:37:35,985 (trainer:779) INFO: 29epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=13.428, loss_att=5.993, acc=0.962, loss=8.224, backward_time=0.253, grad_norm=97.225, clip=100.000, loss_scale=9.445e+21, optim_step_time=0.029, optim0_lr0=9.292e-04, train_time=1.224
+[bmi2:0/4] 2024-07-08 02:41:38,050 (trainer:779) INFO: 29epoch:train:414-826batch: iter_time=1.758e-04, forward_time=0.155, loss_ctc=13.600, loss_att=6.064, acc=0.963, loss=8.325, backward_time=0.253, grad_norm=95.535, clip=100.000, loss_scale=1.570e+22, optim_step_time=0.028, optim0_lr0=9.283e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 02:45:41,845 (trainer:779) INFO: 29epoch:train:827-1239batch: iter_time=1.913e-04, forward_time=0.158, loss_ctc=13.619, loss_att=6.088, acc=0.960, loss=8.347, backward_time=0.254, grad_norm=100.835, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.275e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 02:49:47,694 (trainer:779) INFO: 29epoch:train:1240-1652batch: iter_time=1.764e-04, forward_time=0.159, loss_ctc=13.577, loss_att=6.052, acc=0.965, loss=8.309, backward_time=0.256, grad_norm=93.614, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.267e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 02:53:51,849 (trainer:779) INFO: 29epoch:train:1653-2065batch: iter_time=1.909e-04, forward_time=0.158, loss_ctc=13.454, loss_att=6.064, acc=0.957, loss=8.281, backward_time=0.255, grad_norm=110.819, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.259e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 02:57:55,740 (trainer:779) INFO: 29epoch:train:2066-2478batch: iter_time=1.779e-04, forward_time=0.156, loss_ctc=13.738, loss_att=6.123, acc=0.967, loss=8.408, backward_time=0.255, grad_norm=107.053, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.028, optim0_lr0=9.251e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 03:02:00,308 (trainer:779) INFO: 29epoch:train:2479-2891batch: iter_time=1.782e-04, forward_time=0.159, loss_ctc=13.623, loss_att=6.088, acc=0.961, loss=8.349, backward_time=0.255, grad_norm=103.117, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.242e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 03:06:04,194 (trainer:779) INFO: 29epoch:train:2892-3304batch: iter_time=1.840e-04, forward_time=0.157, loss_ctc=13.570, loss_att=6.053, acc=0.960, loss=8.308, backward_time=0.254, grad_norm=104.858, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.234e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 03:10:08,755 (trainer:779) INFO: 29epoch:train:3305-3717batch: iter_time=1.782e-04, forward_time=0.158, loss_ctc=13.708, loss_att=6.117, acc=0.963, loss=8.394, backward_time=0.255, grad_norm=99.880, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.226e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 03:14:12,760 (trainer:779) INFO: 29epoch:train:3718-4130batch: iter_time=1.818e-04, forward_time=0.158, loss_ctc=13.539, loss_att=6.051, acc=0.962, loss=8.297, backward_time=0.254, grad_norm=99.220, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.218e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 03:18:16,425 (trainer:779) INFO: 29epoch:train:4131-4543batch: iter_time=1.691e-04, forward_time=0.157, loss_ctc=13.702, loss_att=6.096, acc=0.963, loss=8.378, backward_time=0.254, grad_norm=93.261, clip=100.000, loss_scale=1.889e+22, optim_step_time=0.029, optim0_lr0=9.210e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 03:22:20,872 (trainer:779) INFO: 29epoch:train:4544-4956batch: iter_time=1.625e-04, forward_time=0.157, loss_ctc=13.654, loss_att=6.114, acc=0.963, loss=8.376, backward_time=0.254, grad_norm=90.698, clip=100.000, loss_scale=3.732e+22, optim_step_time=0.028, optim0_lr0=9.202e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 03:26:24,180 (trainer:779) INFO: 29epoch:train:4957-5369batch: iter_time=1.773e-04, forward_time=0.157, loss_ctc=13.507, loss_att=6.043, acc=0.963, loss=8.282, backward_time=0.253, grad_norm=94.146, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.194e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 03:30:27,901 (trainer:779) INFO: 29epoch:train:5370-5782batch: iter_time=1.830e-04, forward_time=0.157, loss_ctc=13.641, loss_att=6.101, acc=0.963, loss=8.363, backward_time=0.254, grad_norm=92.642, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.186e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 03:34:31,365 (trainer:779) INFO: 29epoch:train:5783-6195batch: iter_time=1.784e-04, forward_time=0.158, loss_ctc=13.661, loss_att=6.150, acc=0.963, loss=8.403, backward_time=0.254, grad_norm=97.763, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.178e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 03:38:35,034 (trainer:779) INFO: 29epoch:train:6196-6608batch: iter_time=1.927e-04, forward_time=0.157, loss_ctc=13.564, loss_att=6.058, acc=0.963, loss=8.310, backward_time=0.253, grad_norm=94.208, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.170e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 03:42:38,021 (trainer:779) INFO: 29epoch:train:6609-7021batch: iter_time=1.787e-04, forward_time=0.157, loss_ctc=13.655, loss_att=6.070, acc=0.963, loss=8.346, backward_time=0.254, grad_norm=103.209, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.162e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 03:46:41,923 (trainer:779) INFO: 29epoch:train:7022-7434batch: iter_time=1.902e-04, forward_time=0.156, loss_ctc=13.533, loss_att=6.022, acc=0.963, loss=8.275, backward_time=0.254, grad_norm=95.478, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.154e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 03:50:46,521 (trainer:779) INFO: 29epoch:train:7435-7847batch: iter_time=1.784e-04, forward_time=0.159, loss_ctc=13.424, loss_att=5.994, acc=0.962, loss=8.223, backward_time=0.254, grad_norm=95.875, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.146e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 03:54:50,644 (trainer:779) INFO: 29epoch:train:7848-8260batch: iter_time=1.812e-04, forward_time=0.157, loss_ctc=13.661, loss_att=6.104, acc=0.963, loss=8.371, backward_time=0.254, grad_norm=99.303, clip=100.000, loss_scale=3.778e+22, optim_step_time=0.029, optim0_lr0=9.138e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 03:55:44,916 (trainer:365) INFO: 29epoch results: [train] iter_time=2.354e-04, forward_time=0.157, loss_ctc=13.592, loss_att=6.072, acc=0.962, loss=8.328, backward_time=0.254, grad_norm=98.430, clip=100.000, loss_scale=2.675e+22, optim_step_time=0.029, optim0_lr0=9.214e-04, train_time=1.183, time=1 hour, 21 minutes and 32.11 seconds, total_count=239743, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.632, cer_ctc=0.046, loss_att=5.643, acc=0.949, cer=0.033, wer=0.495, loss=6.840, time=17.23 seconds, total_count=986, gpu_max_cached_mem_GB=22.529, [att_plot] time=32.14 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 03:55:50,110 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-08 03:55:50,154 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/19epoch.pth
+[bmi2:0/4] 2024-07-08 03:55:50,154 (trainer:299) INFO: 30/100epoch started. Estimated time to finish: 4 days, 2 hours and 29 minutes
+[bmi2:0/4] 2024-07-08 04:00:01,373 (trainer:779) INFO: 30epoch:train:1-413batch: iter_time=9.423e-04, forward_time=0.156, loss_ctc=13.110, loss_att=5.848, acc=0.962, loss=8.027, backward_time=0.253, grad_norm=99.448, clip=100.000, loss_scale=4.933e+22, optim_step_time=0.029, optim0_lr0=9.130e-04, train_time=1.217
+[bmi2:0/4] 2024-07-08 04:04:03,917 (trainer:779) INFO: 30epoch:train:414-826batch: iter_time=1.917e-04, forward_time=0.156, loss_ctc=13.248, loss_att=5.892, acc=0.961, loss=8.099, backward_time=0.253, grad_norm=91.540, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.028, optim0_lr0=9.122e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 04:08:07,806 (trainer:779) INFO: 30epoch:train:827-1239batch: iter_time=1.772e-04, forward_time=0.157, loss_ctc=13.354, loss_att=5.941, acc=0.964, loss=8.165, backward_time=0.254, grad_norm=98.032, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.114e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 04:12:12,374 (trainer:779) INFO: 30epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.157, loss_ctc=13.438, loss_att=6.019, acc=0.963, loss=8.245, backward_time=0.254, grad_norm=96.713, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.107e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 04:16:17,170 (trainer:779) INFO: 30epoch:train:1653-2065batch: iter_time=0.004, forward_time=0.157, loss_ctc=13.236, loss_att=5.897, acc=0.963, loss=8.098, backward_time=0.254, grad_norm=94.972, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.099e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 04:20:23,830 (trainer:779) INFO: 30epoch:train:2066-2478batch: iter_time=0.008, forward_time=0.158, loss_ctc=13.262, loss_att=5.901, acc=0.961, loss=8.109, backward_time=0.254, grad_norm=93.636, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.091e-04, train_time=1.194
+[bmi2:0/4] 2024-07-08 04:24:28,988 (trainer:779) INFO: 30epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.160, loss_ctc=13.294, loss_att=5.921, acc=0.964, loss=8.133, backward_time=0.256, grad_norm=92.467, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.083e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 04:28:32,976 (trainer:779) INFO: 30epoch:train:2892-3304batch: iter_time=9.205e-04, forward_time=0.157, loss_ctc=13.544, loss_att=5.994, acc=0.965, loss=8.259, backward_time=0.254, grad_norm=90.785, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.076e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 04:32:38,736 (trainer:779) INFO: 30epoch:train:3305-3717batch: iter_time=0.003, forward_time=0.159, loss_ctc=13.417, loss_att=5.975, acc=0.966, loss=8.208, backward_time=0.255, grad_norm=91.531, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.068e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 04:36:42,898 (trainer:779) INFO: 30epoch:train:3718-4130batch: iter_time=0.003, forward_time=0.157, loss_ctc=13.428, loss_att=5.972, acc=0.963, loss=8.209, backward_time=0.254, grad_norm=89.485, clip=100.000, loss_scale=7.556e+22, optim_step_time=0.029, optim0_lr0=9.060e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 04:40:48,637 (trainer:779) INFO: 30epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.158, loss_ctc=13.335, loss_att=5.950, acc=0.964, loss=8.166, backward_time=0.255, grad_norm=100.849, clip=100.000, loss_scale=1.225e+23, optim_step_time=0.028, optim0_lr0=9.053e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 04:44:54,750 (trainer:779) INFO: 30epoch:train:4544-4956batch: iter_time=0.005, forward_time=0.158, loss_ctc=13.342, loss_att=5.932, acc=0.961, loss=8.155, backward_time=0.254, grad_norm=91.345, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.045e-04, train_time=1.191
+[bmi2:0/4] 2024-07-08 04:48:59,313 (trainer:779) INFO: 30epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.158, loss_ctc=13.382, loss_att=5.969, acc=0.965, loss=8.193, backward_time=0.254, grad_norm=92.751, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.037e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 04:53:05,243 (trainer:779) INFO: 30epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.159, loss_ctc=13.491, loss_att=6.009, acc=0.962, loss=8.254, backward_time=0.255, grad_norm=100.866, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.030e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 04:57:12,423 (trainer:779) INFO: 30epoch:train:5783-6195batch: iter_time=0.005, forward_time=0.160, loss_ctc=13.464, loss_att=5.968, acc=0.964, loss=8.217, backward_time=0.256, grad_norm=93.692, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.022e-04, train_time=1.197
+[bmi2:0/4] 2024-07-08 05:01:17,118 (trainer:779) INFO: 30epoch:train:6196-6608batch: iter_time=0.003, forward_time=0.157, loss_ctc=13.245, loss_att=5.896, acc=0.963, loss=8.101, backward_time=0.254, grad_norm=92.567, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.015e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 05:05:22,265 (trainer:779) INFO: 30epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.159, loss_ctc=13.420, loss_att=5.986, acc=0.964, loss=8.216, backward_time=0.254, grad_norm=98.767, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=9.007e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 05:09:27,000 (trainer:779) INFO: 30epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.157, loss_ctc=13.362, loss_att=5.967, acc=0.963, loss=8.185, backward_time=0.255, grad_norm=92.251, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=8.999e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 05:13:33,161 (trainer:779) INFO: 30epoch:train:7435-7847batch: iter_time=0.004, forward_time=0.158, loss_ctc=13.432, loss_att=5.975, acc=0.962, loss=8.212, backward_time=0.255, grad_norm=96.652, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=8.992e-04, train_time=1.192
+[bmi2:0/4] 2024-07-08 05:17:38,501 (trainer:779) INFO: 30epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.159, loss_ctc=13.389, loss_att=5.963, acc=0.962, loss=8.191, backward_time=0.255, grad_norm=102.402, clip=100.000, loss_scale=1.511e+23, optim_step_time=0.029, optim0_lr0=8.984e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 05:19:09,701 (trainer:365) INFO: 30epoch results: [train] iter_time=0.003, forward_time=0.158, loss_ctc=13.360, loss_att=5.949, acc=0.963, loss=8.172, backward_time=0.254, grad_norm=95.066, clip=100.000, loss_scale=1.106e+23, optim_step_time=0.029, optim0_lr0=9.057e-04, train_time=1.188, time=1 hour, 21 minutes and 53.57 seconds, total_count=248010, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.824, cer_ctc=0.045, loss_att=5.644, acc=0.947, cer=0.034, wer=0.498, loss=6.898, time=45.8 seconds, total_count=1020, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.18 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 05:19:14,367 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 05:19:14,411 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/18epoch.pth
+[bmi2:0/4] 2024-07-08 05:19:14,412 (trainer:299) INFO: 31/100epoch started. Estimated time to finish: 4 days, 1 hour and 6 minutes
+[bmi2:0/4] 2024-07-08 05:23:27,951 (trainer:779) INFO: 31epoch:train:1-413batch: iter_time=9.287e-04, forward_time=0.157, loss_ctc=12.974, loss_att=5.752, acc=0.962, loss=7.919, backward_time=0.254, grad_norm=101.072, clip=100.000, loss_scale=2.949e+23, optim_step_time=0.029, optim0_lr0=8.977e-04, train_time=1.228
+[bmi2:0/4] 2024-07-08 05:27:32,779 (trainer:779) INFO: 31epoch:train:414-826batch: iter_time=7.912e-04, forward_time=0.158, loss_ctc=13.209, loss_att=5.845, acc=0.966, loss=8.054, backward_time=0.254, grad_norm=104.868, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.969e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 05:31:37,284 (trainer:779) INFO: 31epoch:train:827-1239batch: iter_time=0.004, forward_time=0.157, loss_ctc=12.929, loss_att=5.790, acc=0.958, loss=7.931, backward_time=0.253, grad_norm=96.745, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.962e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 05:35:41,271 (trainer:779) INFO: 31epoch:train:1240-1652batch: iter_time=3.745e-04, forward_time=0.157, loss_ctc=13.172, loss_att=5.837, acc=0.966, loss=8.037, backward_time=0.254, grad_norm=91.696, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.955e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 05:39:46,988 (trainer:779) INFO: 31epoch:train:1653-2065batch: iter_time=8.250e-04, forward_time=0.158, loss_ctc=13.222, loss_att=5.876, acc=0.966, loss=8.080, backward_time=0.256, grad_norm=95.085, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.947e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 05:43:52,295 (trainer:779) INFO: 31epoch:train:2066-2478batch: iter_time=5.679e-04, forward_time=0.158, loss_ctc=13.360, loss_att=5.936, acc=0.966, loss=8.163, backward_time=0.255, grad_norm=93.962, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.940e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 05:47:56,722 (trainer:779) INFO: 31epoch:train:2479-2891batch: iter_time=1.760e-04, forward_time=0.158, loss_ctc=13.229, loss_att=5.888, acc=0.966, loss=8.090, backward_time=0.255, grad_norm=92.977, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.932e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 05:52:01,149 (trainer:779) INFO: 31epoch:train:2892-3304batch: iter_time=3.405e-04, forward_time=0.157, loss_ctc=13.205, loss_att=5.855, acc=0.965, loss=8.060, backward_time=0.255, grad_norm=91.729, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.925e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 05:56:04,992 (trainer:779) INFO: 31epoch:train:3305-3717batch: iter_time=7.505e-04, forward_time=0.157, loss_ctc=13.246, loss_att=5.929, acc=0.965, loss=8.124, backward_time=0.254, grad_norm=92.001, clip=100.000, loss_scale=3.022e+23, optim_step_time=0.029, optim0_lr0=8.918e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 06:00:08,702 (trainer:779) INFO: 31epoch:train:3718-4130batch: iter_time=5.678e-04, forward_time=0.156, loss_ctc=13.104, loss_att=5.828, acc=0.962, loss=8.011, backward_time=0.253, grad_norm=87.168, clip=100.000, loss_scale=3.825e+23, optim_step_time=0.029, optim0_lr0=8.910e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 06:04:12,032 (trainer:779) INFO: 31epoch:train:4131-4543batch: iter_time=1.752e-04, forward_time=0.157, loss_ctc=13.079, loss_att=5.843, acc=0.961, loss=8.013, backward_time=0.254, grad_norm=91.917, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.028, optim0_lr0=8.903e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 06:08:16,557 (trainer:779) INFO: 31epoch:train:4544-4956batch: iter_time=4.353e-04, forward_time=0.158, loss_ctc=13.120, loss_att=5.841, acc=0.962, loss=8.025, backward_time=0.254, grad_norm=88.631, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.896e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 06:12:20,670 (trainer:779) INFO: 31epoch:train:4957-5369batch: iter_time=1.819e-04, forward_time=0.158, loss_ctc=13.379, loss_att=5.938, acc=0.966, loss=8.170, backward_time=0.255, grad_norm=97.311, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.889e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 06:16:25,092 (trainer:779) INFO: 31epoch:train:5370-5782batch: iter_time=1.740e-04, forward_time=0.157, loss_ctc=13.085, loss_att=5.859, acc=0.965, loss=8.027, backward_time=0.254, grad_norm=91.424, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.881e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 06:20:28,023 (trainer:779) INFO: 31epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.156, loss_ctc=12.746, loss_att=5.717, acc=0.959, loss=7.825, backward_time=0.254, grad_norm=88.118, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.874e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 06:24:33,254 (trainer:779) INFO: 31epoch:train:6196-6608batch: iter_time=1.736e-04, forward_time=0.158, loss_ctc=12.940, loss_att=5.766, acc=0.963, loss=7.918, backward_time=0.255, grad_norm=93.251, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.867e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 06:28:37,273 (trainer:779) INFO: 31epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.157, loss_ctc=13.096, loss_att=5.834, acc=0.962, loss=8.013, backward_time=0.254, grad_norm=96.464, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.860e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 06:32:41,847 (trainer:779) INFO: 31epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.157, loss_ctc=13.025, loss_att=5.818, acc=0.962, loss=7.980, backward_time=0.253, grad_norm=90.972, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.853e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 06:36:45,997 (trainer:779) INFO: 31epoch:train:7435-7847batch: iter_time=1.762e-04, forward_time=0.156, loss_ctc=13.002, loss_att=5.787, acc=0.966, loss=7.952, backward_time=0.254, grad_norm=90.936, clip=100.000, loss_scale=6.045e+23, optim_step_time=0.029, optim0_lr0=8.845e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 06:40:49,745 (trainer:779) INFO: 31epoch:train:7848-8260batch: iter_time=1.803e-04, forward_time=0.157, loss_ctc=13.251, loss_att=5.869, acc=0.967, loss=8.084, backward_time=0.254, grad_norm=91.074, clip=100.000, loss_scale=9.549e+23, optim_step_time=0.029, optim0_lr0=8.838e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 06:41:47,217 (trainer:365) INFO: 31epoch results: [train] iter_time=7.799e-04, forward_time=0.157, loss_ctc=13.116, loss_att=5.840, acc=0.964, loss=8.023, backward_time=0.254, grad_norm=93.382, clip=100.000, loss_scale=4.751e+23, optim_step_time=0.029, optim0_lr0=8.907e-04, train_time=1.185, time=1 hour, 21 minutes and 40.52 seconds, total_count=256277, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.886, cer_ctc=0.045, loss_att=6.519, acc=0.941, cer=0.035, wer=0.504, loss=7.529, time=16.75 seconds, total_count=1054, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.53 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 06:41:51,347 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 06:41:51,348 (trainer:299) INFO: 32/100epoch started. Estimated time to finish: 3 days, 23 hours and 42 minutes
+[bmi2:0/4] 2024-07-08 06:46:03,238 (trainer:779) INFO: 32epoch:train:1-413batch: iter_time=9.838e-04, forward_time=0.156, loss_ctc=12.820, loss_att=5.734, acc=0.964, loss=7.860, backward_time=0.253, grad_norm=99.058, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.028, optim0_lr0=8.831e-04, train_time=1.220
+[bmi2:0/4] 2024-07-08 06:50:07,696 (trainer:779) INFO: 32epoch:train:414-826batch: iter_time=1.845e-04, forward_time=0.158, loss_ctc=12.891, loss_att=5.721, acc=0.968, loss=7.872, backward_time=0.255, grad_norm=90.475, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.029, optim0_lr0=8.824e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 06:54:11,720 (trainer:779) INFO: 32epoch:train:827-1239batch: iter_time=1.891e-04, forward_time=0.158, loss_ctc=12.803, loss_att=5.690, acc=0.963, loss=7.824, backward_time=0.255, grad_norm=94.491, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.028, optim0_lr0=8.817e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 06:58:15,610 (trainer:779) INFO: 32epoch:train:1240-1652batch: iter_time=1.793e-04, forward_time=0.157, loss_ctc=12.848, loss_att=5.706, acc=0.966, loss=7.848, backward_time=0.254, grad_norm=90.359, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.028, optim0_lr0=8.810e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 07:02:19,656 (trainer:779) INFO: 32epoch:train:1653-2065batch: iter_time=1.850e-04, forward_time=0.158, loss_ctc=12.904, loss_att=5.757, acc=0.967, loss=7.901, backward_time=0.254, grad_norm=89.999, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.029, optim0_lr0=8.803e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 07:06:24,068 (trainer:779) INFO: 32epoch:train:2066-2478batch: iter_time=1.783e-04, forward_time=0.157, loss_ctc=12.830, loss_att=5.713, acc=0.962, loss=7.848, backward_time=0.254, grad_norm=95.548, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.029, optim0_lr0=8.796e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 07:10:27,945 (trainer:779) INFO: 32epoch:train:2479-2891batch: iter_time=1.815e-04, forward_time=0.157, loss_ctc=12.808, loss_att=5.687, acc=0.965, loss=7.823, backward_time=0.254, grad_norm=97.163, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.029, optim0_lr0=8.789e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 07:14:33,368 (trainer:779) INFO: 32epoch:train:2892-3304batch: iter_time=1.825e-04, forward_time=0.158, loss_ctc=12.882, loss_att=5.719, acc=0.962, loss=7.868, backward_time=0.255, grad_norm=87.730, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.028, optim0_lr0=8.782e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 07:18:37,606 (trainer:779) INFO: 32epoch:train:3305-3717batch: iter_time=1.707e-04, forward_time=0.157, loss_ctc=12.842, loss_att=5.725, acc=0.965, loss=7.860, backward_time=0.254, grad_norm=93.362, clip=100.000, loss_scale=1.209e+24, optim_step_time=0.029, optim0_lr0=8.775e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 07:22:41,037 (trainer:779) INFO: 32epoch:train:3718-4130batch: iter_time=1.793e-04, forward_time=0.156, loss_ctc=12.902, loss_att=5.717, acc=0.965, loss=7.872, backward_time=0.253, grad_norm=99.505, clip=100.000, loss_scale=2.307e+24, optim_step_time=0.028, optim0_lr0=8.768e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 07:26:44,484 (trainer:779) INFO: 32epoch:train:4131-4543batch: iter_time=1.969e-04, forward_time=0.157, loss_ctc=12.842, loss_att=5.693, acc=0.963, loss=7.838, backward_time=0.254, grad_norm=96.342, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.028, optim0_lr0=8.761e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 07:30:49,106 (trainer:779) INFO: 32epoch:train:4544-4956batch: iter_time=1.824e-04, forward_time=0.157, loss_ctc=12.972, loss_att=5.743, acc=0.967, loss=7.912, backward_time=0.254, grad_norm=92.195, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.029, optim0_lr0=8.754e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 07:34:53,224 (trainer:779) INFO: 32epoch:train:4957-5369batch: iter_time=1.787e-04, forward_time=0.158, loss_ctc=12.934, loss_att=5.777, acc=0.965, loss=7.924, backward_time=0.254, grad_norm=93.262, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.029, optim0_lr0=8.747e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 07:38:57,065 (trainer:779) INFO: 32epoch:train:5370-5782batch: iter_time=1.828e-04, forward_time=0.156, loss_ctc=12.751, loss_att=5.672, acc=0.965, loss=7.796, backward_time=0.254, grad_norm=90.270, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.029, optim0_lr0=8.740e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 07:43:01,033 (trainer:779) INFO: 32epoch:train:5783-6195batch: iter_time=1.772e-04, forward_time=0.157, loss_ctc=12.881, loss_att=5.725, acc=0.966, loss=7.872, backward_time=0.254, grad_norm=89.362, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.028, optim0_lr0=8.733e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 07:47:05,908 (trainer:779) INFO: 32epoch:train:6196-6608batch: iter_time=1.776e-04, forward_time=0.158, loss_ctc=13.144, loss_att=5.831, acc=0.961, loss=8.025, backward_time=0.254, grad_norm=99.435, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.029, optim0_lr0=8.726e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 07:51:09,180 (trainer:779) INFO: 32epoch:train:6609-7021batch: iter_time=1.907e-04, forward_time=0.158, loss_ctc=13.005, loss_att=5.762, acc=0.964, loss=7.935, backward_time=0.254, grad_norm=92.802, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.029, optim0_lr0=8.719e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 07:55:13,059 (trainer:779) INFO: 32epoch:train:7022-7434batch: iter_time=1.811e-04, forward_time=0.157, loss_ctc=13.137, loss_att=5.824, acc=0.966, loss=8.018, backward_time=0.254, grad_norm=93.220, clip=100.000, loss_scale=2.418e+24, optim_step_time=0.028, optim0_lr0=8.713e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 07:59:15,624 (trainer:779) INFO: 32epoch:train:7435-7847batch: iter_time=1.869e-04, forward_time=0.156, loss_ctc=12.836, loss_att=5.755, acc=0.962, loss=7.879, backward_time=0.254, grad_norm=99.058, clip=100.000, loss_scale=2.958e+24, optim_step_time=0.029, optim0_lr0=8.706e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 08:03:18,822 (trainer:779) INFO: 32epoch:train:7848-8260batch: iter_time=1.940e-04, forward_time=0.156, loss_ctc=12.732, loss_att=5.666, acc=0.965, loss=7.786, backward_time=0.253, grad_norm=87.620, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.699e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 08:04:13,845 (trainer:365) INFO: 32epoch results: [train] iter_time=2.230e-04, forward_time=0.157, loss_ctc=12.888, loss_att=5.731, acc=0.964, loss=7.878, backward_time=0.254, grad_norm=93.549, clip=100.000, loss_scale=2.019e+24, optim_step_time=0.029, optim0_lr0=8.764e-04, train_time=1.183, time=1 hour, 21 minutes and 32.22 seconds, total_count=264544, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.439, cer_ctc=0.043, loss_att=5.607, acc=0.949, cer=0.033, wer=0.493, loss=6.757, time=17.27 seconds, total_count=1088, gpu_max_cached_mem_GB=22.529, [att_plot] time=33.01 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 08:04:18,991 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-08 08:04:19,102 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/21epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/31epoch.pth
+[bmi2:0/4] 2024-07-08 08:04:19,103 (trainer:299) INFO: 33/100epoch started. Estimated time to finish: 3 days, 22 hours and 17 minutes
+[bmi2:0/4] 2024-07-08 08:08:32,065 (trainer:779) INFO: 33epoch:train:1-413batch: iter_time=9.856e-04, forward_time=0.156, loss_ctc=12.611, loss_att=5.594, acc=0.964, loss=7.699, backward_time=0.254, grad_norm=85.825, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.029, optim0_lr0=8.692e-04, train_time=1.226
+[bmi2:0/4] 2024-07-08 08:12:35,441 (trainer:779) INFO: 33epoch:train:414-826batch: iter_time=1.840e-04, forward_time=0.157, loss_ctc=12.638, loss_att=5.591, acc=0.968, loss=7.705, backward_time=0.254, grad_norm=88.951, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.685e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 08:16:36,586 (trainer:779) INFO: 33epoch:train:827-1239batch: iter_time=1.849e-04, forward_time=0.155, loss_ctc=12.541, loss_att=5.582, acc=0.963, loss=7.670, backward_time=0.252, grad_norm=93.352, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.678e-04, train_time=1.168
+[bmi2:0/4] 2024-07-08 08:20:40,526 (trainer:779) INFO: 33epoch:train:1240-1652batch: iter_time=1.821e-04, forward_time=0.157, loss_ctc=12.807, loss_att=5.694, acc=0.967, loss=7.828, backward_time=0.254, grad_norm=87.858, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.672e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 08:24:43,665 (trainer:779) INFO: 33epoch:train:1653-2065batch: iter_time=1.877e-04, forward_time=0.156, loss_ctc=12.587, loss_att=5.557, acc=0.966, loss=7.666, backward_time=0.255, grad_norm=87.990, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.665e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 08:28:47,916 (trainer:779) INFO: 33epoch:train:2066-2478batch: iter_time=1.884e-04, forward_time=0.157, loss_ctc=12.499, loss_att=5.552, acc=0.965, loss=7.636, backward_time=0.255, grad_norm=87.668, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.029, optim0_lr0=8.658e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 08:33:01,279 (trainer:779) INFO: 33epoch:train:2479-2891batch: iter_time=1.965e-04, forward_time=0.169, loss_ctc=12.724, loss_att=5.663, acc=0.964, loss=7.781, backward_time=0.267, grad_norm=87.328, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.029, optim0_lr0=8.652e-04, train_time=1.227
+[bmi2:0/4] 2024-07-08 08:37:10,468 (trainer:779) INFO: 33epoch:train:2892-3304batch: iter_time=1.907e-04, forward_time=0.163, loss_ctc=12.676, loss_att=5.617, acc=0.964, loss=7.734, backward_time=0.261, grad_norm=86.158, clip=100.000, loss_scale=4.836e+24, optim_step_time=0.028, optim0_lr0=8.645e-04, train_time=1.206
+[bmi2:0/4] 2024-07-08 08:41:37,561 (trainer:779) INFO: 33epoch:train:3305-3717batch: iter_time=1.908e-04, forward_time=0.185, loss_ctc=12.771, loss_att=5.646, acc=0.966, loss=7.784, backward_time=0.284, grad_norm=91.958, clip=100.000, loss_scale=7.512e+24, optim_step_time=0.030, optim0_lr0=8.638e-04, train_time=1.294
+[bmi2:0/4] 2024-07-08 08:46:01,369 (trainer:779) INFO: 33epoch:train:3718-4130batch: iter_time=1.774e-04, forward_time=0.180, loss_ctc=12.549, loss_att=5.593, acc=0.962, loss=7.680, backward_time=0.279, grad_norm=87.723, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.029, optim0_lr0=8.632e-04, train_time=1.277
+[bmi2:0/4] 2024-07-08 08:50:22,029 (trainer:779) INFO: 33epoch:train:4131-4543batch: iter_time=2.309e-04, forward_time=0.176, loss_ctc=12.646, loss_att=5.628, acc=0.966, loss=7.733, backward_time=0.275, grad_norm=93.874, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.625e-04, train_time=1.262
+[bmi2:0/4] 2024-07-08 08:54:31,097 (trainer:779) INFO: 33epoch:train:4544-4956batch: iter_time=2.226e-04, forward_time=0.164, loss_ctc=12.802, loss_att=5.652, acc=0.966, loss=7.797, backward_time=0.259, grad_norm=92.112, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.618e-04, train_time=1.206
+[bmi2:0/4] 2024-07-08 08:58:35,173 (trainer:779) INFO: 33epoch:train:4957-5369batch: iter_time=1.944e-04, forward_time=0.158, loss_ctc=12.601, loss_att=5.587, acc=0.962, loss=7.691, backward_time=0.254, grad_norm=86.124, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.031, optim0_lr0=8.612e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 09:02:38,836 (trainer:779) INFO: 33epoch:train:5370-5782batch: iter_time=1.938e-04, forward_time=0.156, loss_ctc=12.856, loss_att=5.678, acc=0.968, loss=7.832, backward_time=0.254, grad_norm=88.346, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.605e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 09:06:43,609 (trainer:779) INFO: 33epoch:train:5783-6195batch: iter_time=1.890e-04, forward_time=0.158, loss_ctc=12.842, loss_att=5.683, acc=0.969, loss=7.830, backward_time=0.255, grad_norm=86.764, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.599e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 09:10:46,787 (trainer:779) INFO: 33epoch:train:6196-6608batch: iter_time=1.900e-04, forward_time=0.157, loss_ctc=12.544, loss_att=5.567, acc=0.963, loss=7.660, backward_time=0.253, grad_norm=90.321, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.592e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 09:14:49,812 (trainer:779) INFO: 33epoch:train:6609-7021batch: iter_time=1.973e-04, forward_time=0.156, loss_ctc=12.607, loss_att=5.592, acc=0.966, loss=7.696, backward_time=0.254, grad_norm=88.847, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.586e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 09:18:55,341 (trainer:779) INFO: 33epoch:train:7022-7434batch: iter_time=2.005e-04, forward_time=0.159, loss_ctc=12.644, loss_att=5.692, acc=0.964, loss=7.778, backward_time=0.255, grad_norm=105.500, clip=100.000, loss_scale=9.671e+24, optim_step_time=0.030, optim0_lr0=8.579e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 09:22:58,528 (trainer:779) INFO: 33epoch:train:7435-7847batch: iter_time=1.858e-04, forward_time=0.157, loss_ctc=12.714, loss_att=5.653, acc=0.966, loss=7.771, backward_time=0.255, grad_norm=89.098, clip=100.000, loss_scale=1.808e+25, optim_step_time=0.030, optim0_lr0=8.572e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 09:27:02,827 (trainer:779) INFO: 33epoch:train:7848-8260batch: iter_time=1.889e-04, forward_time=0.157, loss_ctc=12.732, loss_att=5.614, acc=0.967, loss=7.750, backward_time=0.253, grad_norm=83.107, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.566e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 09:28:02,005 (trainer:365) INFO: 33epoch results: [train] iter_time=2.330e-04, forward_time=0.162, loss_ctc=12.668, loss_att=5.621, acc=0.965, loss=7.735, backward_time=0.259, grad_norm=89.465, clip=100.000, loss_scale=8.541e+24, optim_step_time=0.029, optim0_lr0=8.629e-04, train_time=1.202, time=1 hour, 22 minutes and 48.92 seconds, total_count=272811, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.691, cer_ctc=0.044, loss_att=5.709, acc=0.948, cer=0.033, wer=0.495, loss=6.904, time=17.45 seconds, total_count=1122, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.53 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 09:28:06,308 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 09:28:06,386 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/17epoch.pth
+[bmi2:0/4] 2024-07-08 09:28:06,387 (trainer:299) INFO: 34/100epoch started. Estimated time to finish: 3 days, 20 hours and 55 minutes
+[bmi2:0/4] 2024-07-08 09:32:19,976 (trainer:779) INFO: 34epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=12.348, loss_att=5.480, acc=0.966, loss=7.540, backward_time=0.254, grad_norm=100.196, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.559e-04, train_time=1.228
+[bmi2:0/4] 2024-07-08 09:36:23,380 (trainer:779) INFO: 34epoch:train:414-826batch: iter_time=1.858e-04, forward_time=0.157, loss_ctc=12.236, loss_att=5.452, acc=0.962, loss=7.487, backward_time=0.253, grad_norm=88.384, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.553e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 09:40:26,205 (trainer:779) INFO: 34epoch:train:827-1239batch: iter_time=2.047e-04, forward_time=0.156, loss_ctc=12.328, loss_att=5.451, acc=0.966, loss=7.514, backward_time=0.254, grad_norm=84.227, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.546e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 09:44:30,247 (trainer:779) INFO: 34epoch:train:1240-1652batch: iter_time=1.846e-04, forward_time=0.157, loss_ctc=12.530, loss_att=5.536, acc=0.969, loss=7.634, backward_time=0.254, grad_norm=83.163, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.540e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 09:48:33,646 (trainer:779) INFO: 34epoch:train:1653-2065batch: iter_time=1.880e-04, forward_time=0.157, loss_ctc=12.552, loss_att=5.518, acc=0.969, loss=7.628, backward_time=0.254, grad_norm=88.851, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.534e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 09:52:38,384 (trainer:779) INFO: 34epoch:train:2066-2478batch: iter_time=1.858e-04, forward_time=0.159, loss_ctc=12.362, loss_att=5.458, acc=0.965, loss=7.529, backward_time=0.254, grad_norm=80.751, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.527e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 09:56:41,530 (trainer:779) INFO: 34epoch:train:2479-2891batch: iter_time=1.868e-04, forward_time=0.157, loss_ctc=12.332, loss_att=5.492, acc=0.963, loss=7.544, backward_time=0.254, grad_norm=86.144, clip=100.000, loss_scale=1.934e+25, optim_step_time=0.030, optim0_lr0=8.521e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 10:00:46,434 (trainer:779) INFO: 34epoch:train:2892-3304batch: iter_time=1.952e-04, forward_time=0.158, loss_ctc=12.330, loss_att=5.491, acc=0.964, loss=7.543, backward_time=0.254, grad_norm=78.856, clip=100.000, loss_scale=2.317e+25, optim_step_time=0.030, optim0_lr0=8.514e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 10:04:49,732 (trainer:779) INFO: 34epoch:train:3305-3717batch: iter_time=2.030e-04, forward_time=0.157, loss_ctc=12.401, loss_att=5.503, acc=0.968, loss=7.572, backward_time=0.254, grad_norm=85.930, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.508e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 10:08:54,207 (trainer:779) INFO: 34epoch:train:3718-4130batch: iter_time=1.813e-04, forward_time=0.158, loss_ctc=12.504, loss_att=5.528, acc=0.964, loss=7.621, backward_time=0.253, grad_norm=90.158, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.502e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 10:12:58,397 (trainer:779) INFO: 34epoch:train:4131-4543batch: iter_time=1.985e-04, forward_time=0.157, loss_ctc=12.628, loss_att=5.631, acc=0.966, loss=7.730, backward_time=0.255, grad_norm=93.852, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.495e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 10:17:04,317 (trainer:779) INFO: 34epoch:train:4544-4956batch: iter_time=1.882e-04, forward_time=0.159, loss_ctc=12.579, loss_att=5.567, acc=0.967, loss=7.670, backward_time=0.255, grad_norm=84.205, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.489e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 10:21:08,471 (trainer:779) INFO: 34epoch:train:4957-5369batch: iter_time=1.935e-04, forward_time=0.158, loss_ctc=12.546, loss_att=5.544, acc=0.966, loss=7.645, backward_time=0.254, grad_norm=90.436, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.483e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 10:25:12,050 (trainer:779) INFO: 34epoch:train:5370-5782batch: iter_time=1.948e-04, forward_time=0.157, loss_ctc=12.401, loss_att=5.452, acc=0.964, loss=7.537, backward_time=0.253, grad_norm=85.032, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.476e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 10:29:15,942 (trainer:779) INFO: 34epoch:train:5783-6195batch: iter_time=1.955e-04, forward_time=0.157, loss_ctc=12.563, loss_att=5.559, acc=0.968, loss=7.660, backward_time=0.254, grad_norm=82.326, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.470e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 10:33:20,426 (trainer:779) INFO: 34epoch:train:6196-6608batch: iter_time=1.909e-04, forward_time=0.158, loss_ctc=12.463, loss_att=5.553, acc=0.965, loss=7.626, backward_time=0.254, grad_norm=78.785, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.464e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 10:37:25,494 (trainer:779) INFO: 34epoch:train:6609-7021batch: iter_time=1.896e-04, forward_time=0.159, loss_ctc=12.347, loss_att=5.466, acc=0.968, loss=7.530, backward_time=0.255, grad_norm=85.954, clip=100.000, loss_scale=3.869e+25, optim_step_time=0.030, optim0_lr0=8.458e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 10:41:29,144 (trainer:779) INFO: 34epoch:train:7022-7434batch: iter_time=1.957e-04, forward_time=0.156, loss_ctc=12.360, loss_att=5.485, acc=0.967, loss=7.548, backward_time=0.254, grad_norm=83.252, clip=100.000, loss_scale=5.850e+25, optim_step_time=0.030, optim0_lr0=8.451e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 10:45:32,868 (trainer:779) INFO: 34epoch:train:7435-7847batch: iter_time=1.997e-04, forward_time=0.157, loss_ctc=12.238, loss_att=5.432, acc=0.965, loss=7.474, backward_time=0.254, grad_norm=81.292, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.030, optim0_lr0=8.445e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 10:49:38,465 (trainer:779) INFO: 34epoch:train:7848-8260batch: iter_time=1.945e-04, forward_time=0.159, loss_ctc=12.388, loss_att=5.501, acc=0.969, loss=7.567, backward_time=0.256, grad_norm=86.041, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.030, optim0_lr0=8.439e-04, train_time=1.189
+[bmi2:0/4] 2024-07-08 10:50:37,081 (trainer:365) INFO: 34epoch results: [train] iter_time=2.404e-04, forward_time=0.157, loss_ctc=12.420, loss_att=5.504, acc=0.966, loss=7.579, backward_time=0.254, grad_norm=85.866, clip=100.000, loss_scale=3.603e+25, optim_step_time=0.030, optim0_lr0=8.499e-04, train_time=1.184, time=1 hour, 21 minutes and 36.9 seconds, total_count=281078, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.808, cer_ctc=0.044, loss_att=5.727, acc=0.947, cer=0.034, wer=0.497, loss=6.951, time=16.97 seconds, total_count=1156, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.81 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 10:50:40,902 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 10:50:40,953 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/27epoch.pth
+[bmi2:0/4] 2024-07-08 10:50:40,953 (trainer:299) INFO: 35/100epoch started. Estimated time to finish: 3 days, 19 hours and 30 minutes
+[bmi2:0/4] 2024-07-08 10:54:54,567 (trainer:779) INFO: 35epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=12.089, loss_att=5.316, acc=0.966, loss=7.348, backward_time=0.255, grad_norm=83.010, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.030, optim0_lr0=8.433e-04, train_time=1.229
+[bmi2:0/4] 2024-07-08 10:58:58,213 (trainer:779) INFO: 35epoch:train:414-826batch: iter_time=2.059e-04, forward_time=0.158, loss_ctc=11.941, loss_att=5.296, acc=0.964, loss=7.289, backward_time=0.254, grad_norm=79.123, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.030, optim0_lr0=8.427e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 11:03:02,125 (trainer:779) INFO: 35epoch:train:827-1239batch: iter_time=2.054e-04, forward_time=0.157, loss_ctc=12.285, loss_att=5.431, acc=0.968, loss=7.487, backward_time=0.253, grad_norm=81.570, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.031, optim0_lr0=8.420e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 11:07:07,384 (trainer:779) INFO: 35epoch:train:1240-1652batch: iter_time=1.880e-04, forward_time=0.158, loss_ctc=12.213, loss_att=5.399, acc=0.966, loss=7.443, backward_time=0.255, grad_norm=87.137, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.030, optim0_lr0=8.414e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 11:11:12,130 (trainer:779) INFO: 35epoch:train:1653-2065batch: iter_time=2.255e-04, forward_time=0.159, loss_ctc=12.241, loss_att=5.411, acc=0.969, loss=7.460, backward_time=0.254, grad_norm=81.942, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.032, optim0_lr0=8.408e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 11:15:32,761 (trainer:779) INFO: 35epoch:train:2066-2478batch: iter_time=7.575e-04, forward_time=0.181, loss_ctc=12.064, loss_att=5.369, acc=0.962, loss=7.378, backward_time=0.253, grad_norm=81.314, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.041, optim0_lr0=8.402e-04, train_time=1.261
+[bmi2:0/4] 2024-07-08 11:20:06,267 (trainer:779) INFO: 35epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.196, loss_ctc=12.104, loss_att=5.341, acc=0.968, loss=7.370, backward_time=0.256, grad_norm=84.017, clip=100.000, loss_scale=7.737e+25, optim_step_time=0.046, optim0_lr0=8.396e-04, train_time=1.325
+[bmi2:0/4] 2024-07-08 11:24:40,810 (trainer:779) INFO: 35epoch:train:2892-3304batch: iter_time=7.442e-04, forward_time=0.197, loss_ctc=12.023, loss_att=5.345, acc=0.966, loss=7.348, backward_time=0.255, grad_norm=80.450, clip=100.000, loss_scale=1.424e+26, optim_step_time=0.048, optim0_lr0=8.390e-04, train_time=1.329
+[bmi2:0/4] 2024-07-08 11:28:58,248 (trainer:779) INFO: 35epoch:train:3305-3717batch: iter_time=7.468e-04, forward_time=0.173, loss_ctc=12.139, loss_att=5.431, acc=0.966, loss=7.444, backward_time=0.253, grad_norm=92.261, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.037, optim0_lr0=8.384e-04, train_time=1.247
+[bmi2:0/4] 2024-07-08 11:33:02,559 (trainer:779) INFO: 35epoch:train:3718-4130batch: iter_time=2.240e-04, forward_time=0.159, loss_ctc=12.200, loss_att=5.412, acc=0.968, loss=7.448, backward_time=0.253, grad_norm=80.427, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.031, optim0_lr0=8.378e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 11:37:05,654 (trainer:779) INFO: 35epoch:train:4131-4543batch: iter_time=3.375e-04, forward_time=0.160, loss_ctc=12.184, loss_att=5.403, acc=0.970, loss=7.437, backward_time=0.251, grad_norm=90.903, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.032, optim0_lr0=8.371e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 11:41:07,524 (trainer:779) INFO: 35epoch:train:4544-4956batch: iter_time=5.739e-04, forward_time=0.160, loss_ctc=12.297, loss_att=5.440, acc=0.965, loss=7.497, backward_time=0.248, grad_norm=82.039, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.033, optim0_lr0=8.365e-04, train_time=1.170
+[bmi2:0/4] 2024-07-08 11:45:08,231 (trainer:779) INFO: 35epoch:train:4957-5369batch: iter_time=6.073e-04, forward_time=0.160, loss_ctc=12.125, loss_att=5.403, acc=0.965, loss=7.419, backward_time=0.248, grad_norm=79.216, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.032, optim0_lr0=8.359e-04, train_time=1.166
+[bmi2:0/4] 2024-07-08 11:49:11,296 (trainer:779) INFO: 35epoch:train:5370-5782batch: iter_time=4.813e-04, forward_time=0.161, loss_ctc=12.347, loss_att=5.448, acc=0.967, loss=7.518, backward_time=0.249, grad_norm=83.609, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.032, optim0_lr0=8.353e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 11:53:14,522 (trainer:779) INFO: 35epoch:train:5783-6195batch: iter_time=4.574e-04, forward_time=0.161, loss_ctc=12.424, loss_att=5.451, acc=0.971, loss=7.543, backward_time=0.250, grad_norm=81.970, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.032, optim0_lr0=8.347e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 11:57:17,676 (trainer:779) INFO: 35epoch:train:6196-6608batch: iter_time=5.867e-04, forward_time=0.161, loss_ctc=12.155, loss_att=5.382, acc=0.968, loss=7.414, backward_time=0.250, grad_norm=83.525, clip=100.000, loss_scale=1.547e+26, optim_step_time=0.032, optim0_lr0=8.341e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 12:01:19,725 (trainer:779) INFO: 35epoch:train:6609-7021batch: iter_time=6.328e-04, forward_time=0.160, loss_ctc=12.389, loss_att=5.447, acc=0.968, loss=7.530, backward_time=0.249, grad_norm=87.786, clip=100.000, loss_scale=1.788e+26, optim_step_time=0.033, optim0_lr0=8.335e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 12:05:21,750 (trainer:779) INFO: 35epoch:train:7022-7434batch: iter_time=6.875e-04, forward_time=0.160, loss_ctc=12.289, loss_att=5.457, acc=0.966, loss=7.507, backward_time=0.249, grad_norm=85.988, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.329e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 12:09:24,802 (trainer:779) INFO: 35epoch:train:7435-7847batch: iter_time=4.676e-04, forward_time=0.161, loss_ctc=12.248, loss_att=5.426, acc=0.967, loss=7.473, backward_time=0.249, grad_norm=78.839, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.323e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 12:13:27,237 (trainer:779) INFO: 35epoch:train:7848-8260batch: iter_time=6.162e-04, forward_time=0.160, loss_ctc=12.189, loss_att=5.390, acc=0.965, loss=7.430, backward_time=0.249, grad_norm=84.222, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.317e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 12:14:28,513 (trainer:365) INFO: 35epoch results: [train] iter_time=5.629e-04, forward_time=0.165, loss_ctc=12.197, loss_att=5.400, acc=0.967, loss=7.439, backward_time=0.252, grad_norm=83.481, clip=100.000, loss_scale=1.516e+26, optim_step_time=0.034, optim0_lr0=8.375e-04, train_time=1.202, time=1 hour, 22 minutes and 51.2 seconds, total_count=289345, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.571, cer_ctc=0.043, loss_att=5.665, acc=0.949, cer=0.032, wer=0.492, loss=6.837, time=18.66 seconds, total_count=1190, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.7 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 12:14:33,644 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-08 12:14:33,707 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/34epoch.pth
+[bmi2:0/4] 2024-07-08 12:14:33,708 (trainer:299) INFO: 36/100epoch started. Estimated time to finish: 3 days, 18 hours and 8 minutes
+[bmi2:0/4] 2024-07-08 12:18:46,261 (trainer:779) INFO: 36epoch:train:1-413batch: iter_time=0.002, forward_time=0.159, loss_ctc=11.957, loss_att=5.271, acc=0.968, loss=7.277, backward_time=0.249, grad_norm=77.195, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.311e-04, train_time=1.224
+[bmi2:0/4] 2024-07-08 12:22:49,503 (trainer:779) INFO: 36epoch:train:414-826batch: iter_time=4.557e-04, forward_time=0.162, loss_ctc=12.047, loss_att=5.324, acc=0.963, loss=7.341, backward_time=0.250, grad_norm=86.768, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.306e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 12:26:51,536 (trainer:779) INFO: 36epoch:train:827-1239batch: iter_time=4.942e-04, forward_time=0.160, loss_ctc=11.936, loss_att=5.257, acc=0.968, loss=7.261, backward_time=0.249, grad_norm=84.435, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.300e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 12:30:54,749 (trainer:779) INFO: 36epoch:train:1240-1652batch: iter_time=3.959e-04, forward_time=0.160, loss_ctc=12.073, loss_att=5.329, acc=0.970, loss=7.352, backward_time=0.250, grad_norm=83.886, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.294e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 12:34:56,629 (trainer:779) INFO: 36epoch:train:1653-2065batch: iter_time=4.199e-04, forward_time=0.161, loss_ctc=12.020, loss_att=5.297, acc=0.967, loss=7.314, backward_time=0.249, grad_norm=81.499, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.033, optim0_lr0=8.288e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 12:38:59,882 (trainer:779) INFO: 36epoch:train:2066-2478batch: iter_time=3.335e-04, forward_time=0.161, loss_ctc=11.867, loss_att=5.252, acc=0.962, loss=7.237, backward_time=0.249, grad_norm=77.767, clip=100.000, loss_scale=3.095e+26, optim_step_time=0.032, optim0_lr0=8.282e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 12:43:02,385 (trainer:779) INFO: 36epoch:train:2479-2891batch: iter_time=4.352e-04, forward_time=0.161, loss_ctc=12.008, loss_att=5.279, acc=0.967, loss=7.298, backward_time=0.250, grad_norm=83.323, clip=100.000, loss_scale=4.597e+26, optim_step_time=0.032, optim0_lr0=8.276e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 12:47:05,278 (trainer:779) INFO: 36epoch:train:2892-3304batch: iter_time=5.257e-04, forward_time=0.162, loss_ctc=12.026, loss_att=5.317, acc=0.970, loss=7.330, backward_time=0.250, grad_norm=83.772, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.270e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 12:51:06,385 (trainer:779) INFO: 36epoch:train:3305-3717batch: iter_time=5.026e-04, forward_time=0.160, loss_ctc=11.993, loss_att=5.298, acc=0.967, loss=7.306, backward_time=0.248, grad_norm=85.725, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.264e-04, train_time=1.168
+[bmi2:0/4] 2024-07-08 12:55:09,680 (trainer:779) INFO: 36epoch:train:3718-4130batch: iter_time=4.843e-04, forward_time=0.162, loss_ctc=12.061, loss_att=5.319, acc=0.968, loss=7.342, backward_time=0.250, grad_norm=76.765, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.259e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 12:59:12,575 (trainer:779) INFO: 36epoch:train:4131-4543batch: iter_time=6.473e-04, forward_time=0.161, loss_ctc=12.023, loss_att=5.324, acc=0.968, loss=7.334, backward_time=0.250, grad_norm=80.874, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.253e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 13:03:15,849 (trainer:779) INFO: 36epoch:train:4544-4956batch: iter_time=4.111e-04, forward_time=0.161, loss_ctc=12.105, loss_att=5.348, acc=0.969, loss=7.375, backward_time=0.250, grad_norm=83.943, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.247e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 13:07:18,186 (trainer:779) INFO: 36epoch:train:4957-5369batch: iter_time=4.696e-04, forward_time=0.161, loss_ctc=11.986, loss_att=5.316, acc=0.971, loss=7.317, backward_time=0.250, grad_norm=81.427, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.241e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 13:11:21,447 (trainer:779) INFO: 36epoch:train:5370-5782batch: iter_time=6.580e-04, forward_time=0.162, loss_ctc=12.002, loss_att=5.327, acc=0.968, loss=7.329, backward_time=0.250, grad_norm=80.895, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.033, optim0_lr0=8.235e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 13:15:23,174 (trainer:779) INFO: 36epoch:train:5783-6195batch: iter_time=5.428e-04, forward_time=0.160, loss_ctc=11.983, loss_att=5.278, acc=0.967, loss=7.289, backward_time=0.248, grad_norm=87.390, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.033, optim0_lr0=8.230e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 13:19:25,272 (trainer:779) INFO: 36epoch:train:6196-6608batch: iter_time=5.206e-04, forward_time=0.160, loss_ctc=12.048, loss_att=5.324, acc=0.969, loss=7.342, backward_time=0.249, grad_norm=82.501, clip=100.000, loss_scale=6.190e+26, optim_step_time=0.032, optim0_lr0=8.224e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 13:23:26,968 (trainer:779) INFO: 36epoch:train:6609-7021batch: iter_time=4.868e-04, forward_time=0.161, loss_ctc=12.015, loss_att=5.319, acc=0.966, loss=7.328, backward_time=0.249, grad_norm=79.211, clip=100.000, loss_scale=1.115e+27, optim_step_time=0.032, optim0_lr0=8.218e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 13:27:29,555 (trainer:779) INFO: 36epoch:train:7022-7434batch: iter_time=5.304e-04, forward_time=0.161, loss_ctc=11.910, loss_att=5.314, acc=0.964, loss=7.293, backward_time=0.249, grad_norm=82.578, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.212e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 13:31:31,183 (trainer:779) INFO: 36epoch:train:7435-7847batch: iter_time=5.652e-04, forward_time=0.161, loss_ctc=12.125, loss_att=5.361, acc=0.970, loss=7.390, backward_time=0.248, grad_norm=80.746, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.033, optim0_lr0=8.207e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 13:35:33,356 (trainer:779) INFO: 36epoch:train:7848-8260batch: iter_time=4.215e-04, forward_time=0.161, loss_ctc=11.908, loss_att=5.275, acc=0.966, loss=7.265, backward_time=0.249, grad_norm=78.505, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.201e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 13:36:34,919 (trainer:365) INFO: 36epoch results: [train] iter_time=5.457e-04, forward_time=0.161, loss_ctc=12.004, loss_att=5.306, acc=0.967, loss=7.316, backward_time=0.249, grad_norm=81.967, clip=100.000, loss_scale=6.363e+26, optim_step_time=0.032, optim0_lr0=8.256e-04, train_time=1.176, time=1 hour, 21 minutes and 4.51 seconds, total_count=297612, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.761, cer_ctc=0.043, loss_att=5.608, acc=0.949, cer=0.034, wer=0.497, loss=6.854, time=18.26 seconds, total_count=1224, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.44 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 13:36:40,375 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 13:36:40,455 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/30epoch.pth
+[bmi2:0/4] 2024-07-08 13:36:40,455 (trainer:299) INFO: 37/100epoch started. Estimated time to finish: 3 days, 16 hours and 43 minutes
+[bmi2:0/4] 2024-07-08 13:40:53,260 (trainer:779) INFO: 37epoch:train:1-413batch: iter_time=0.002, forward_time=0.160, loss_ctc=11.850, loss_att=5.218, acc=0.969, loss=7.208, backward_time=0.249, grad_norm=82.353, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.195e-04, train_time=1.225
+[bmi2:0/4] 2024-07-08 13:44:55,392 (trainer:779) INFO: 37epoch:train:414-826batch: iter_time=6.684e-04, forward_time=0.160, loss_ctc=11.886, loss_att=5.255, acc=0.969, loss=7.245, backward_time=0.249, grad_norm=81.134, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.190e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 13:48:58,872 (trainer:779) INFO: 37epoch:train:827-1239batch: iter_time=5.113e-04, forward_time=0.162, loss_ctc=11.890, loss_att=5.211, acc=0.970, loss=7.215, backward_time=0.250, grad_norm=85.294, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.033, optim0_lr0=8.184e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 13:53:02,740 (trainer:779) INFO: 37epoch:train:1240-1652batch: iter_time=5.485e-04, forward_time=0.161, loss_ctc=11.691, loss_att=5.178, acc=0.966, loss=7.132, backward_time=0.250, grad_norm=84.375, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.178e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 13:57:04,861 (trainer:779) INFO: 37epoch:train:1653-2065batch: iter_time=6.002e-04, forward_time=0.161, loss_ctc=11.928, loss_att=5.259, acc=0.965, loss=7.260, backward_time=0.250, grad_norm=88.241, clip=100.000, loss_scale=1.238e+27, optim_step_time=0.032, optim0_lr0=8.173e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 14:01:08,731 (trainer:779) INFO: 37epoch:train:2066-2478batch: iter_time=4.741e-04, forward_time=0.161, loss_ctc=11.884, loss_att=5.238, acc=0.970, loss=7.232, backward_time=0.251, grad_norm=82.872, clip=100.000, loss_scale=1.399e+27, optim_step_time=0.032, optim0_lr0=8.167e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 14:05:10,027 (trainer:779) INFO: 37epoch:train:2479-2891batch: iter_time=6.209e-04, forward_time=0.161, loss_ctc=11.789, loss_att=5.195, acc=0.966, loss=7.173, backward_time=0.248, grad_norm=79.427, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.161e-04, train_time=1.169
+[bmi2:0/4] 2024-07-08 14:09:13,167 (trainer:779) INFO: 37epoch:train:2892-3304batch: iter_time=5.122e-04, forward_time=0.161, loss_ctc=11.956, loss_att=5.295, acc=0.970, loss=7.294, backward_time=0.250, grad_norm=85.249, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.156e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 14:13:16,167 (trainer:779) INFO: 37epoch:train:3305-3717batch: iter_time=5.816e-04, forward_time=0.161, loss_ctc=11.802, loss_att=5.224, acc=0.967, loss=7.197, backward_time=0.250, grad_norm=82.412, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.150e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 14:17:19,510 (trainer:779) INFO: 37epoch:train:3718-4130batch: iter_time=5.156e-04, forward_time=0.162, loss_ctc=11.906, loss_att=5.238, acc=0.969, loss=7.238, backward_time=0.250, grad_norm=77.276, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.033, optim0_lr0=8.145e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 14:21:21,367 (trainer:779) INFO: 37epoch:train:4131-4543batch: iter_time=5.393e-04, forward_time=0.160, loss_ctc=11.990, loss_att=5.358, acc=0.967, loss=7.348, backward_time=0.249, grad_norm=83.945, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.031, optim0_lr0=8.139e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 14:25:24,029 (trainer:779) INFO: 37epoch:train:4544-4956batch: iter_time=6.177e-04, forward_time=0.161, loss_ctc=11.798, loss_att=5.245, acc=0.964, loss=7.211, backward_time=0.249, grad_norm=80.935, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.133e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 14:29:25,483 (trainer:779) INFO: 37epoch:train:4957-5369batch: iter_time=4.325e-04, forward_time=0.160, loss_ctc=11.837, loss_att=5.233, acc=0.966, loss=7.214, backward_time=0.248, grad_norm=94.737, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.128e-04, train_time=1.170
+[bmi2:0/4] 2024-07-08 14:33:28,279 (trainer:779) INFO: 37epoch:train:5370-5782batch: iter_time=4.072e-04, forward_time=0.160, loss_ctc=12.015, loss_att=5.300, acc=0.971, loss=7.315, backward_time=0.249, grad_norm=81.845, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.122e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 14:37:31,131 (trainer:779) INFO: 37epoch:train:5783-6195batch: iter_time=4.745e-04, forward_time=0.161, loss_ctc=11.870, loss_att=5.232, acc=0.968, loss=7.223, backward_time=0.250, grad_norm=85.242, clip=100.000, loss_scale=2.476e+27, optim_step_time=0.032, optim0_lr0=8.117e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 14:41:33,158 (trainer:779) INFO: 37epoch:train:6196-6608batch: iter_time=5.853e-04, forward_time=0.160, loss_ctc=11.650, loss_att=5.152, acc=0.966, loss=7.102, backward_time=0.248, grad_norm=82.634, clip=100.000, loss_scale=3.576e+27, optim_step_time=0.032, optim0_lr0=8.111e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 14:45:35,121 (trainer:779) INFO: 37epoch:train:6609-7021batch: iter_time=4.883e-04, forward_time=0.160, loss_ctc=11.914, loss_att=5.283, acc=0.968, loss=7.272, backward_time=0.249, grad_norm=82.340, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.033, optim0_lr0=8.106e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 14:49:36,720 (trainer:779) INFO: 37epoch:train:7022-7434batch: iter_time=4.426e-04, forward_time=0.159, loss_ctc=11.870, loss_att=5.254, acc=0.966, loss=7.239, backward_time=0.248, grad_norm=82.293, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.100e-04, train_time=1.169
+[bmi2:0/4] 2024-07-08 14:53:38,279 (trainer:779) INFO: 37epoch:train:7435-7847batch: iter_time=5.146e-04, forward_time=0.160, loss_ctc=12.071, loss_att=5.314, acc=0.969, loss=7.341, backward_time=0.249, grad_norm=80.870, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.095e-04, train_time=1.170
+[bmi2:0/4] 2024-07-08 14:57:41,851 (trainer:779) INFO: 37epoch:train:7848-8260batch: iter_time=4.085e-04, forward_time=0.162, loss_ctc=11.795, loss_att=5.195, acc=0.968, loss=7.175, backward_time=0.249, grad_norm=84.716, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.089e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 14:58:43,918 (trainer:365) INFO: 37epoch results: [train] iter_time=5.721e-04, forward_time=0.161, loss_ctc=11.868, loss_att=5.243, acc=0.968, loss=7.231, backward_time=0.249, grad_norm=83.390, clip=100.000, loss_scale=2.665e+27, optim_step_time=0.032, optim0_lr0=8.142e-04, train_time=1.177, time=1 hour, 21 minutes and 6.37 seconds, total_count=305879, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.514, cer_ctc=0.043, loss_att=5.880, acc=0.948, cer=0.033, wer=0.491, loss=6.970, time=18.43 seconds, total_count=1258, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.66 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 14:58:49,553 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 14:58:49,621 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/25epoch.pth
+[bmi2:0/4] 2024-07-08 14:58:49,621 (trainer:299) INFO: 38/100epoch started. Estimated time to finish: 3 days, 15 hours and 18 minutes
+[bmi2:0/4] 2024-07-08 15:03:02,612 (trainer:779) INFO: 38epoch:train:1-413batch: iter_time=0.001, forward_time=0.160, loss_ctc=11.637, loss_att=5.093, acc=0.969, loss=7.057, backward_time=0.249, grad_norm=82.901, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.033, optim0_lr0=8.084e-04, train_time=1.226
+[bmi2:0/4] 2024-07-08 15:07:05,523 (trainer:779) INFO: 38epoch:train:414-826batch: iter_time=3.369e-04, forward_time=0.162, loss_ctc=11.723, loss_att=5.136, acc=0.967, loss=7.112, backward_time=0.249, grad_norm=85.035, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.078e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 15:11:07,119 (trainer:779) INFO: 38epoch:train:827-1239batch: iter_time=4.269e-04, forward_time=0.161, loss_ctc=11.717, loss_att=5.150, acc=0.969, loss=7.120, backward_time=0.249, grad_norm=86.219, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.073e-04, train_time=1.170
+[bmi2:0/4] 2024-07-08 15:15:09,884 (trainer:779) INFO: 38epoch:train:1240-1652batch: iter_time=4.794e-04, forward_time=0.161, loss_ctc=11.821, loss_att=5.248, acc=0.969, loss=7.220, backward_time=0.249, grad_norm=74.709, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.033, optim0_lr0=8.068e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 15:19:12,225 (trainer:779) INFO: 38epoch:train:1653-2065batch: iter_time=4.340e-04, forward_time=0.161, loss_ctc=11.741, loss_att=5.161, acc=0.969, loss=7.135, backward_time=0.249, grad_norm=85.306, clip=100.000, loss_scale=4.952e+27, optim_step_time=0.032, optim0_lr0=8.062e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 15:23:13,741 (trainer:779) INFO: 38epoch:train:2066-2478batch: iter_time=4.017e-04, forward_time=0.160, loss_ctc=11.820, loss_att=5.255, acc=0.966, loss=7.225, backward_time=0.248, grad_norm=87.802, clip=100.000, loss_scale=8.779e+27, optim_step_time=0.032, optim0_lr0=8.057e-04, train_time=1.169
+[bmi2:0/4] 2024-07-08 15:27:16,246 (trainer:779) INFO: 38epoch:train:2479-2891batch: iter_time=6.325e-04, forward_time=0.161, loss_ctc=11.896, loss_att=5.218, acc=0.972, loss=7.221, backward_time=0.250, grad_norm=85.651, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.051e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 15:31:20,176 (trainer:779) INFO: 38epoch:train:2892-3304batch: iter_time=3.711e-04, forward_time=0.162, loss_ctc=11.630, loss_att=5.179, acc=0.967, loss=7.114, backward_time=0.251, grad_norm=81.794, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.032, optim0_lr0=8.046e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 15:35:21,952 (trainer:779) INFO: 38epoch:train:3305-3717batch: iter_time=5.326e-04, forward_time=0.161, loss_ctc=11.664, loss_att=5.153, acc=0.963, loss=7.106, backward_time=0.249, grad_norm=81.627, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.041e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 15:39:24,480 (trainer:779) INFO: 38epoch:train:3718-4130batch: iter_time=4.164e-04, forward_time=0.161, loss_ctc=11.792, loss_att=5.170, acc=0.968, loss=7.157, backward_time=0.250, grad_norm=81.993, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.035e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 15:43:27,175 (trainer:779) INFO: 38epoch:train:4131-4543batch: iter_time=5.150e-04, forward_time=0.162, loss_ctc=11.806, loss_att=5.205, acc=0.969, loss=7.185, backward_time=0.250, grad_norm=81.863, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.030e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 15:47:30,125 (trainer:779) INFO: 38epoch:train:4544-4956batch: iter_time=3.814e-04, forward_time=0.162, loss_ctc=11.635, loss_att=5.131, acc=0.967, loss=7.082, backward_time=0.249, grad_norm=82.287, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.025e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 15:51:33,458 (trainer:779) INFO: 38epoch:train:4957-5369batch: iter_time=5.031e-04, forward_time=0.162, loss_ctc=11.838, loss_att=5.200, acc=0.970, loss=7.191, backward_time=0.251, grad_norm=78.709, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.033, optim0_lr0=8.019e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 15:55:36,947 (trainer:779) INFO: 38epoch:train:5370-5782batch: iter_time=6.090e-04, forward_time=0.162, loss_ctc=11.708, loss_att=5.140, acc=0.970, loss=7.111, backward_time=0.250, grad_norm=84.016, clip=100.000, loss_scale=9.904e+27, optim_step_time=0.032, optim0_lr0=8.014e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 15:59:40,150 (trainer:779) INFO: 38epoch:train:5783-6195batch: iter_time=4.610e-04, forward_time=0.161, loss_ctc=11.862, loss_att=5.219, acc=0.970, loss=7.212, backward_time=0.251, grad_norm=82.358, clip=100.000, loss_scale=1.077e+28, optim_step_time=0.032, optim0_lr0=8.009e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 16:03:42,073 (trainer:779) INFO: 38epoch:train:6196-6608batch: iter_time=4.822e-04, forward_time=0.161, loss_ctc=11.792, loss_att=5.187, acc=0.968, loss=7.168, backward_time=0.249, grad_norm=78.804, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.032, optim0_lr0=8.003e-04, train_time=1.171
+[bmi2:0/4] 2024-07-08 16:07:44,597 (trainer:779) INFO: 38epoch:train:6609-7021batch: iter_time=5.607e-04, forward_time=0.161, loss_ctc=11.786, loss_att=5.207, acc=0.966, loss=7.181, backward_time=0.249, grad_norm=83.286, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.033, optim0_lr0=7.998e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 16:11:47,251 (trainer:779) INFO: 38epoch:train:7022-7434batch: iter_time=4.965e-04, forward_time=0.162, loss_ctc=11.714, loss_att=5.185, acc=0.968, loss=7.144, backward_time=0.250, grad_norm=84.014, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.033, optim0_lr0=7.993e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 16:15:49,361 (trainer:779) INFO: 38epoch:train:7435-7847batch: iter_time=4.438e-04, forward_time=0.160, loss_ctc=11.705, loss_att=5.162, acc=0.969, loss=7.125, backward_time=0.249, grad_norm=81.368, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.033, optim0_lr0=7.987e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 16:19:51,534 (trainer:779) INFO: 38epoch:train:7848-8260batch: iter_time=6.183e-04, forward_time=0.161, loss_ctc=11.673, loss_att=5.129, acc=0.967, loss=7.092, backward_time=0.249, grad_norm=75.773, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.032, optim0_lr0=7.982e-04, train_time=1.172
+[bmi2:0/4] 2024-07-08 16:20:52,927 (trainer:365) INFO: 38epoch results: [train] iter_time=5.163e-04, forward_time=0.161, loss_ctc=11.747, loss_att=5.176, acc=0.968, loss=7.147, backward_time=0.249, grad_norm=82.274, clip=100.000, loss_scale=1.114e+28, optim_step_time=0.032, optim0_lr0=8.033e-04, train_time=1.177, time=1 hour, 21 minutes and 7.01 seconds, total_count=314146, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.677, cer_ctc=0.043, loss_att=5.767, acc=0.948, cer=0.033, wer=0.493, loss=6.940, time=18.26 seconds, total_count=1292, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.03 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 16:20:58,595 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 16:20:58,675 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/22epoch.pth
+[bmi2:0/4] 2024-07-08 16:20:58,676 (trainer:299) INFO: 39/100epoch started. Estimated time to finish: 3 days, 13 hours and 53 minutes
+[bmi2:0/4] 2024-07-08 16:25:11,411 (trainer:779) INFO: 39epoch:train:1-413batch: iter_time=0.002, forward_time=0.160, loss_ctc=11.526, loss_att=5.085, acc=0.968, loss=7.017, backward_time=0.248, grad_norm=76.931, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.033, optim0_lr0=7.977e-04, train_time=1.224
+[bmi2:0/4] 2024-07-08 16:29:13,233 (trainer:779) INFO: 39epoch:train:414-826batch: iter_time=3.863e-04, forward_time=0.161, loss_ctc=11.619, loss_att=5.089, acc=0.969, loss=7.048, backward_time=0.249, grad_norm=78.665, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.032, optim0_lr0=7.972e-04, train_time=1.170
+[bmi2:0/4] 2024-07-08 16:33:15,634 (trainer:779) INFO: 39epoch:train:827-1239batch: iter_time=5.842e-04, forward_time=0.161, loss_ctc=11.431, loss_att=5.033, acc=0.967, loss=6.952, backward_time=0.249, grad_norm=83.303, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.032, optim0_lr0=7.966e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 16:37:20,386 (trainer:779) INFO: 39epoch:train:1240-1652batch: iter_time=2.334e-04, forward_time=0.160, loss_ctc=11.650, loss_att=5.104, acc=0.969, loss=7.067, backward_time=0.252, grad_norm=84.982, clip=100.000, loss_scale=1.981e+28, optim_step_time=0.031, optim0_lr0=7.961e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 16:41:24,206 (trainer:779) INFO: 39epoch:train:1653-2065batch: iter_time=1.985e-04, forward_time=0.158, loss_ctc=11.543, loss_att=5.083, acc=0.967, loss=7.021, backward_time=0.254, grad_norm=82.608, clip=100.000, loss_scale=2.808e+28, optim_step_time=0.030, optim0_lr0=7.956e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 16:45:29,310 (trainer:779) INFO: 39epoch:train:2066-2478batch: iter_time=1.804e-04, forward_time=0.158, loss_ctc=11.576, loss_att=5.087, acc=0.968, loss=7.034, backward_time=0.255, grad_norm=82.968, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.951e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 16:49:34,507 (trainer:779) INFO: 39epoch:train:2479-2891batch: iter_time=1.969e-04, forward_time=0.158, loss_ctc=11.554, loss_att=5.048, acc=0.969, loss=7.000, backward_time=0.256, grad_norm=75.147, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.946e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 16:53:39,313 (trainer:779) INFO: 39epoch:train:2892-3304batch: iter_time=2.020e-04, forward_time=0.158, loss_ctc=11.669, loss_att=5.106, acc=0.968, loss=7.075, backward_time=0.254, grad_norm=80.645, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.940e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 16:57:42,861 (trainer:779) INFO: 39epoch:train:3305-3717batch: iter_time=2.066e-04, forward_time=0.157, loss_ctc=11.527, loss_att=5.071, acc=0.968, loss=7.008, backward_time=0.254, grad_norm=83.706, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.935e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 17:01:47,626 (trainer:779) INFO: 39epoch:train:3718-4130batch: iter_time=1.905e-04, forward_time=0.158, loss_ctc=11.378, loss_att=5.021, acc=0.967, loss=6.928, backward_time=0.254, grad_norm=83.646, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.930e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 17:05:51,434 (trainer:779) INFO: 39epoch:train:4131-4543batch: iter_time=2.055e-04, forward_time=0.157, loss_ctc=11.471, loss_att=5.069, acc=0.965, loss=6.990, backward_time=0.255, grad_norm=86.761, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.925e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 17:09:56,081 (trainer:779) INFO: 39epoch:train:4544-4956batch: iter_time=1.987e-04, forward_time=0.158, loss_ctc=11.760, loss_att=5.187, acc=0.971, loss=7.159, backward_time=0.254, grad_norm=88.615, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.920e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 17:13:58,900 (trainer:779) INFO: 39epoch:train:4957-5369batch: iter_time=2.777e-04, forward_time=0.158, loss_ctc=11.763, loss_att=5.147, acc=0.970, loss=7.132, backward_time=0.252, grad_norm=83.626, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.032, optim0_lr0=7.915e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 17:18:02,578 (trainer:779) INFO: 39epoch:train:5370-5782batch: iter_time=2.016e-04, forward_time=0.157, loss_ctc=11.851, loss_att=5.208, acc=0.970, loss=7.201, backward_time=0.253, grad_norm=88.497, clip=100.000, loss_scale=3.961e+28, optim_step_time=0.030, optim0_lr0=7.910e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 17:22:05,715 (trainer:779) INFO: 39epoch:train:5783-6195batch: iter_time=1.885e-04, forward_time=0.157, loss_ctc=11.783, loss_att=5.139, acc=0.971, loss=7.132, backward_time=0.253, grad_norm=79.374, clip=100.000, loss_scale=6.865e+28, optim_step_time=0.030, optim0_lr0=7.904e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 17:26:10,071 (trainer:779) INFO: 39epoch:train:6196-6608batch: iter_time=1.996e-04, forward_time=0.158, loss_ctc=11.621, loss_att=5.134, acc=0.967, loss=7.080, backward_time=0.255, grad_norm=82.453, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.899e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 17:30:13,953 (trainer:779) INFO: 39epoch:train:6609-7021batch: iter_time=1.945e-04, forward_time=0.158, loss_ctc=11.651, loss_att=5.113, acc=0.968, loss=7.074, backward_time=0.254, grad_norm=84.728, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.894e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 17:34:18,972 (trainer:779) INFO: 39epoch:train:7022-7434batch: iter_time=2.040e-04, forward_time=0.159, loss_ctc=11.698, loss_att=5.138, acc=0.970, loss=7.106, backward_time=0.255, grad_norm=82.221, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.889e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 17:38:24,065 (trainer:779) INFO: 39epoch:train:7435-7847batch: iter_time=2.046e-04, forward_time=0.159, loss_ctc=11.598, loss_att=5.154, acc=0.970, loss=7.087, backward_time=0.255, grad_norm=86.917, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.884e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 17:42:26,471 (trainer:779) INFO: 39epoch:train:7848-8260batch: iter_time=2.014e-04, forward_time=0.155, loss_ctc=11.784, loss_att=5.163, acc=0.968, loss=7.149, backward_time=0.252, grad_norm=84.840, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.879e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 17:43:25,315 (trainer:365) INFO: 39epoch results: [train] iter_time=3.034e-04, forward_time=0.158, loss_ctc=11.621, loss_att=5.108, acc=0.969, loss=7.062, backward_time=0.253, grad_norm=83.035, clip=100.000, loss_scale=4.646e+28, optim_step_time=0.030, optim0_lr0=7.928e-04, train_time=1.183, time=1 hour, 21 minutes and 33.06 seconds, total_count=322413, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.664, cer_ctc=0.043, loss_att=5.616, acc=0.950, cer=0.032, wer=0.483, loss=6.830, time=16.96 seconds, total_count=1326, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.61 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 17:43:30,423 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-08 17:43:30,508 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/28epoch.pth
+[bmi2:0/4] 2024-07-08 17:43:30,508 (trainer:299) INFO: 40/100epoch started. Estimated time to finish: 3 days, 12 hours and 29 minutes
+[bmi2:0/4] 2024-07-08 17:47:43,760 (trainer:779) INFO: 40epoch:train:1-413batch: iter_time=9.307e-04, forward_time=0.157, loss_ctc=11.513, loss_att=5.122, acc=0.974, loss=7.039, backward_time=0.254, grad_norm=94.037, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.874e-04, train_time=1.227
+[bmi2:0/4] 2024-07-08 17:51:46,118 (trainer:779) INFO: 40epoch:train:414-826batch: iter_time=1.891e-04, forward_time=0.157, loss_ctc=11.441, loss_att=5.025, acc=0.966, loss=6.950, backward_time=0.252, grad_norm=77.550, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.869e-04, train_time=1.173
+[bmi2:0/4] 2024-07-08 17:55:48,863 (trainer:779) INFO: 40epoch:train:827-1239batch: iter_time=1.910e-04, forward_time=0.156, loss_ctc=11.415, loss_att=5.004, acc=0.967, loss=6.928, backward_time=0.253, grad_norm=78.922, clip=100.000, loss_scale=7.923e+28, optim_step_time=0.030, optim0_lr0=7.864e-04, train_time=1.176
+[bmi2:0/4] 2024-07-08 18:01:03,985 (trainer:779) INFO: 40epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.190, loss_ctc=11.396, loss_att=5.032, acc=0.964, loss=6.941, backward_time=0.270, grad_norm=89.954, clip=100.000, loss_scale=8.420e+28, optim_step_time=0.044, optim0_lr0=7.859e-04, train_time=1.524
+[bmi2:0/4] 2024-07-08 18:16:27,116 (trainer:779) INFO: 40epoch:train:1653-2065batch: iter_time=0.016, forward_time=0.696, loss_ctc=11.450, loss_att=5.034, acc=0.971, loss=6.959, backward_time=0.714, grad_norm=81.432, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.218, optim0_lr0=7.854e-04, train_time=4.478
+[bmi2:0/4] 2024-07-08 18:20:36,903 (trainer:779) INFO: 40epoch:train:2066-2478batch: iter_time=3.614e-04, forward_time=0.167, loss_ctc=11.486, loss_att=5.046, acc=0.971, loss=6.978, backward_time=0.254, grad_norm=81.091, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.034, optim0_lr0=7.849e-04, train_time=1.209
+[bmi2:0/4] 2024-07-08 18:24:39,882 (trainer:779) INFO: 40epoch:train:2479-2891batch: iter_time=2.092e-04, forward_time=0.157, loss_ctc=11.422, loss_att=4.999, acc=0.970, loss=6.926, backward_time=0.253, grad_norm=81.557, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.031, optim0_lr0=7.844e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 18:28:44,171 (trainer:779) INFO: 40epoch:train:2892-3304batch: iter_time=2.412e-04, forward_time=0.159, loss_ctc=11.463, loss_att=5.047, acc=0.969, loss=6.972, backward_time=0.253, grad_norm=84.113, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.030, optim0_lr0=7.839e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 18:32:48,001 (trainer:779) INFO: 40epoch:train:3305-3717batch: iter_time=2.104e-04, forward_time=0.158, loss_ctc=11.509, loss_att=5.050, acc=0.972, loss=6.988, backward_time=0.253, grad_norm=79.641, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.029, optim0_lr0=7.834e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 18:36:51,277 (trainer:779) INFO: 40epoch:train:3718-4130batch: iter_time=2.007e-04, forward_time=0.157, loss_ctc=11.496, loss_att=5.034, acc=0.969, loss=6.973, backward_time=0.253, grad_norm=77.732, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.028, optim0_lr0=7.829e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 18:40:54,899 (trainer:779) INFO: 40epoch:train:4131-4543batch: iter_time=2.071e-04, forward_time=0.158, loss_ctc=11.426, loss_att=5.040, acc=0.967, loss=6.956, backward_time=0.253, grad_norm=81.513, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.029, optim0_lr0=7.824e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 18:44:59,033 (trainer:779) INFO: 40epoch:train:4544-4956batch: iter_time=2.084e-04, forward_time=0.158, loss_ctc=11.517, loss_att=5.062, acc=0.968, loss=6.998, backward_time=0.253, grad_norm=86.984, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.030, optim0_lr0=7.819e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 18:49:02,989 (trainer:779) INFO: 40epoch:train:4957-5369batch: iter_time=1.963e-04, forward_time=0.159, loss_ctc=11.380, loss_att=5.024, acc=0.967, loss=6.931, backward_time=0.253, grad_norm=82.232, clip=100.000, loss_scale=1.585e+29, optim_step_time=0.028, optim0_lr0=7.814e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 18:53:06,278 (trainer:779) INFO: 40epoch:train:5370-5782batch: iter_time=2.102e-04, forward_time=0.156, loss_ctc=11.514, loss_att=5.043, acc=0.971, loss=6.985, backward_time=0.253, grad_norm=82.551, clip=100.000, loss_scale=2.182e+29, optim_step_time=0.028, optim0_lr0=7.809e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 18:57:09,724 (trainer:779) INFO: 40epoch:train:5783-6195batch: iter_time=2.216e-04, forward_time=0.157, loss_ctc=11.417, loss_att=5.015, acc=0.969, loss=6.935, backward_time=0.253, grad_norm=79.273, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.028, optim0_lr0=7.804e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 19:01:14,221 (trainer:779) INFO: 40epoch:train:6196-6608batch: iter_time=1.857e-04, forward_time=0.157, loss_ctc=11.412, loss_att=5.015, acc=0.969, loss=6.934, backward_time=0.253, grad_norm=75.785, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.028, optim0_lr0=7.799e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 19:05:17,688 (trainer:779) INFO: 40epoch:train:6609-7021batch: iter_time=1.937e-04, forward_time=0.156, loss_ctc=11.407, loss_att=5.040, acc=0.966, loss=6.950, backward_time=0.253, grad_norm=79.259, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.027, optim0_lr0=7.795e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 19:09:21,350 (trainer:779) INFO: 40epoch:train:7022-7434batch: iter_time=2.204e-04, forward_time=0.158, loss_ctc=11.551, loss_att=5.069, acc=0.972, loss=7.013, backward_time=0.252, grad_norm=82.309, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.029, optim0_lr0=7.790e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 19:13:25,788 (trainer:779) INFO: 40epoch:train:7435-7847batch: iter_time=2.055e-04, forward_time=0.158, loss_ctc=11.452, loss_att=5.031, acc=0.968, loss=6.957, backward_time=0.254, grad_norm=81.617, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.029, optim0_lr0=7.785e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 19:17:29,520 (trainer:779) INFO: 40epoch:train:7848-8260batch: iter_time=1.941e-04, forward_time=0.157, loss_ctc=11.591, loss_att=5.057, acc=0.970, loss=7.017, backward_time=0.253, grad_norm=81.519, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.029, optim0_lr0=7.780e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 19:18:28,392 (trainer:365) INFO: 40epoch results: [train] iter_time=0.001, forward_time=0.186, loss_ctc=11.462, loss_att=5.039, acc=0.969, loss=6.966, backward_time=0.277, grad_norm=81.958, clip=100.000, loss_scale=1.935e+29, optim_step_time=0.039, optim0_lr0=7.827e-04, train_time=1.365, time=1 hour, 34 minutes and 4.12 seconds, total_count=330680, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.665, cer_ctc=0.043, loss_att=5.881, acc=0.948, cer=0.033, wer=0.491, loss=7.016, time=17.8 seconds, total_count=1360, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.96 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 19:18:33,280 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 19:18:33,333 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/24epoch.pth
+[bmi2:0/4] 2024-07-08 19:18:33,333 (trainer:299) INFO: 41/100epoch started. Estimated time to finish: 3 days, 11 hours and 24 minutes
+[bmi2:0/4] 2024-07-08 19:22:45,802 (trainer:779) INFO: 41epoch:train:1-413batch: iter_time=0.001, forward_time=0.158, loss_ctc=11.225, loss_att=4.929, acc=0.969, loss=6.818, backward_time=0.252, grad_norm=83.496, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.029, optim0_lr0=7.775e-04, train_time=1.223
+[bmi2:0/4] 2024-07-08 19:26:49,670 (trainer:779) INFO: 41epoch:train:414-826batch: iter_time=2.747e-04, forward_time=0.159, loss_ctc=11.272, loss_att=4.925, acc=0.972, loss=6.829, backward_time=0.253, grad_norm=78.004, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.030, optim0_lr0=7.770e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 19:30:53,833 (trainer:779) INFO: 41epoch:train:827-1239batch: iter_time=2.014e-04, forward_time=0.160, loss_ctc=11.294, loss_att=4.928, acc=0.971, loss=6.838, backward_time=0.253, grad_norm=77.015, clip=100.000, loss_scale=3.169e+29, optim_step_time=0.031, optim0_lr0=7.765e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 19:34:57,380 (trainer:779) INFO: 41epoch:train:1240-1652batch: iter_time=1.998e-04, forward_time=0.158, loss_ctc=11.341, loss_att=4.964, acc=0.973, loss=6.877, backward_time=0.254, grad_norm=79.976, clip=100.000, loss_scale=5.404e+29, optim_step_time=0.028, optim0_lr0=7.760e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 19:39:01,541 (trainer:779) INFO: 41epoch:train:1653-2065batch: iter_time=2.517e-04, forward_time=0.159, loss_ctc=11.348, loss_att=4.972, acc=0.969, loss=6.885, backward_time=0.253, grad_norm=78.073, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.030, optim0_lr0=7.756e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 19:43:05,572 (trainer:779) INFO: 41epoch:train:2066-2478batch: iter_time=2.015e-04, forward_time=0.158, loss_ctc=11.430, loss_att=5.032, acc=0.969, loss=6.952, backward_time=0.254, grad_norm=82.592, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.029, optim0_lr0=7.751e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 19:47:09,140 (trainer:779) INFO: 41epoch:train:2479-2891batch: iter_time=2.120e-04, forward_time=0.158, loss_ctc=11.440, loss_att=5.005, acc=0.971, loss=6.936, backward_time=0.254, grad_norm=80.185, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.029, optim0_lr0=7.746e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 19:51:14,806 (trainer:779) INFO: 41epoch:train:2892-3304batch: iter_time=2.074e-04, forward_time=0.161, loss_ctc=11.309, loss_att=4.991, acc=0.967, loss=6.886, backward_time=0.255, grad_norm=78.914, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.030, optim0_lr0=7.741e-04, train_time=1.189
+[bmi2:0/4] 2024-07-08 19:55:19,608 (trainer:779) INFO: 41epoch:train:3305-3717batch: iter_time=2.460e-04, forward_time=0.159, loss_ctc=11.373, loss_att=4.999, acc=0.969, loss=6.911, backward_time=0.255, grad_norm=79.977, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.030, optim0_lr0=7.736e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 19:59:24,521 (trainer:779) INFO: 41epoch:train:3718-4130batch: iter_time=1.920e-04, forward_time=0.159, loss_ctc=11.263, loss_att=4.946, acc=0.972, loss=6.841, backward_time=0.254, grad_norm=78.427, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.030, optim0_lr0=7.732e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 20:03:28,779 (trainer:779) INFO: 41epoch:train:4131-4543batch: iter_time=2.361e-04, forward_time=0.160, loss_ctc=11.328, loss_att=4.995, acc=0.964, loss=6.895, backward_time=0.254, grad_norm=79.325, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.030, optim0_lr0=7.727e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 20:07:33,182 (trainer:779) INFO: 41epoch:train:4544-4956batch: iter_time=3.139e-04, forward_time=0.158, loss_ctc=11.237, loss_att=4.967, acc=0.967, loss=6.848, backward_time=0.252, grad_norm=74.559, clip=100.000, loss_scale=6.338e+29, optim_step_time=0.032, optim0_lr0=7.722e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 20:11:37,418 (trainer:779) INFO: 41epoch:train:4957-5369batch: iter_time=2.359e-04, forward_time=0.159, loss_ctc=11.313, loss_att=4.947, acc=0.967, loss=6.857, backward_time=0.253, grad_norm=79.047, clip=100.000, loss_scale=6.461e+29, optim_step_time=0.032, optim0_lr0=7.717e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 20:15:42,130 (trainer:779) INFO: 41epoch:train:5370-5782batch: iter_time=2.356e-04, forward_time=0.160, loss_ctc=11.333, loss_att=4.950, acc=0.971, loss=6.865, backward_time=0.252, grad_norm=80.313, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.032, optim0_lr0=7.713e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 20:19:46,372 (trainer:779) INFO: 41epoch:train:5783-6195batch: iter_time=2.518e-04, forward_time=0.160, loss_ctc=11.554, loss_att=5.053, acc=0.969, loss=7.003, backward_time=0.252, grad_norm=85.276, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.033, optim0_lr0=7.708e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 20:23:51,406 (trainer:779) INFO: 41epoch:train:6196-6608batch: iter_time=2.327e-04, forward_time=0.160, loss_ctc=11.192, loss_att=4.908, acc=0.970, loss=6.793, backward_time=0.252, grad_norm=75.321, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.033, optim0_lr0=7.703e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 20:27:55,828 (trainer:779) INFO: 41epoch:train:6609-7021batch: iter_time=2.247e-04, forward_time=0.160, loss_ctc=11.410, loss_att=5.024, acc=0.968, loss=6.940, backward_time=0.252, grad_norm=82.100, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.032, optim0_lr0=7.698e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 20:32:01,333 (trainer:779) INFO: 41epoch:train:7022-7434batch: iter_time=2.273e-04, forward_time=0.160, loss_ctc=11.310, loss_att=4.961, acc=0.972, loss=6.865, backward_time=0.253, grad_norm=77.330, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.033, optim0_lr0=7.694e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 20:36:05,788 (trainer:779) INFO: 41epoch:train:7435-7847batch: iter_time=2.199e-04, forward_time=0.160, loss_ctc=11.357, loss_att=4.993, acc=0.968, loss=6.902, backward_time=0.252, grad_norm=80.555, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.032, optim0_lr0=7.689e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 20:40:10,612 (trainer:779) INFO: 41epoch:train:7848-8260batch: iter_time=2.299e-04, forward_time=0.161, loss_ctc=11.453, loss_att=5.017, acc=0.970, loss=6.948, backward_time=0.253, grad_norm=81.257, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.032, optim0_lr0=7.684e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 20:41:11,447 (trainer:365) INFO: 41epoch results: [train] iter_time=2.702e-04, forward_time=0.159, loss_ctc=11.339, loss_att=4.975, acc=0.969, loss=6.885, backward_time=0.253, grad_norm=79.587, clip=100.000, loss_scale=8.045e+29, optim_step_time=0.031, optim0_lr0=7.729e-04, train_time=1.186, time=1 hour, 21 minutes and 42.49 seconds, total_count=338947, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.536, cer_ctc=0.042, loss_att=5.656, acc=0.949, cer=0.033, wer=0.493, loss=6.820, time=17.47 seconds, total_count=1394, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.14 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 20:41:16,425 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 20:41:16,469 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/33epoch.pth
+[bmi2:0/4] 2024-07-08 20:41:16,469 (trainer:299) INFO: 42/100epoch started. Estimated time to finish: 3 days, 10 hours and 15.94 seconds
+[bmi2:0/4] 2024-07-08 20:45:29,751 (trainer:779) INFO: 42epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=11.125, loss_att=4.883, acc=0.969, loss=6.755, backward_time=0.253, grad_norm=76.364, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.033, optim0_lr0=7.680e-04, train_time=1.227
+[bmi2:0/4] 2024-07-08 20:49:35,153 (trainer:779) INFO: 42epoch:train:414-826batch: iter_time=2.279e-04, forward_time=0.160, loss_ctc=11.132, loss_att=4.876, acc=0.967, loss=6.753, backward_time=0.253, grad_norm=79.889, clip=100.000, loss_scale=1.268e+30, optim_step_time=0.031, optim0_lr0=7.675e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 20:53:38,968 (trainer:779) INFO: 42epoch:train:827-1239batch: iter_time=2.363e-04, forward_time=0.159, loss_ctc=11.199, loss_att=4.876, acc=0.971, loss=6.773, backward_time=0.252, grad_norm=75.588, clip=100.000, loss_scale=1.711e+30, optim_step_time=0.032, optim0_lr0=7.670e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 20:57:43,873 (trainer:779) INFO: 42epoch:train:1240-1652batch: iter_time=2.261e-04, forward_time=0.159, loss_ctc=11.134, loss_att=4.880, acc=0.970, loss=6.757, backward_time=0.253, grad_norm=79.428, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.032, optim0_lr0=7.666e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 21:01:48,835 (trainer:779) INFO: 42epoch:train:1653-2065batch: iter_time=2.414e-04, forward_time=0.158, loss_ctc=11.183, loss_att=4.918, acc=0.968, loss=6.797, backward_time=0.254, grad_norm=81.177, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.031, optim0_lr0=7.661e-04, train_time=1.186
+[bmi2:0/4] 2024-07-08 21:05:53,327 (trainer:779) INFO: 42epoch:train:2066-2478batch: iter_time=2.328e-04, forward_time=0.159, loss_ctc=11.143, loss_att=4.867, acc=0.972, loss=6.750, backward_time=0.254, grad_norm=77.688, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.030, optim0_lr0=7.656e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 21:09:56,355 (trainer:779) INFO: 42epoch:train:2479-2891batch: iter_time=2.500e-04, forward_time=0.157, loss_ctc=11.278, loss_att=4.944, acc=0.972, loss=6.844, backward_time=0.252, grad_norm=78.598, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.032, optim0_lr0=7.652e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 21:14:00,618 (trainer:779) INFO: 42epoch:train:2892-3304batch: iter_time=2.325e-04, forward_time=0.159, loss_ctc=11.219, loss_att=4.912, acc=0.969, loss=6.804, backward_time=0.253, grad_norm=83.996, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.031, optim0_lr0=7.647e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 21:18:04,725 (trainer:779) INFO: 42epoch:train:3305-3717batch: iter_time=2.408e-04, forward_time=0.159, loss_ctc=11.310, loss_att=4.956, acc=0.970, loss=6.862, backward_time=0.252, grad_norm=80.987, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.031, optim0_lr0=7.643e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 21:22:09,967 (trainer:779) INFO: 42epoch:train:3718-4130batch: iter_time=2.226e-04, forward_time=0.160, loss_ctc=11.275, loss_att=4.929, acc=0.970, loss=6.833, backward_time=0.253, grad_norm=76.700, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.033, optim0_lr0=7.638e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 21:26:13,838 (trainer:779) INFO: 42epoch:train:4131-4543batch: iter_time=2.433e-04, forward_time=0.158, loss_ctc=11.425, loss_att=4.999, acc=0.972, loss=6.927, backward_time=0.252, grad_norm=77.620, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.031, optim0_lr0=7.633e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 21:30:17,351 (trainer:779) INFO: 42epoch:train:4544-4956batch: iter_time=2.292e-04, forward_time=0.159, loss_ctc=11.059, loss_att=4.849, acc=0.968, loss=6.712, backward_time=0.251, grad_norm=76.233, clip=100.000, loss_scale=2.535e+30, optim_step_time=0.032, optim0_lr0=7.629e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 21:34:21,892 (trainer:779) INFO: 42epoch:train:4957-5369batch: iter_time=2.271e-04, forward_time=0.160, loss_ctc=11.188, loss_att=4.880, acc=0.972, loss=6.773, backward_time=0.253, grad_norm=78.500, clip=100.000, loss_scale=4.221e+30, optim_step_time=0.032, optim0_lr0=7.624e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 21:38:27,131 (trainer:779) INFO: 42epoch:train:5370-5782batch: iter_time=2.345e-04, forward_time=0.159, loss_ctc=11.194, loss_att=4.874, acc=0.969, loss=6.770, backward_time=0.253, grad_norm=79.822, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.031, optim0_lr0=7.620e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 21:42:31,560 (trainer:779) INFO: 42epoch:train:5783-6195batch: iter_time=2.307e-04, forward_time=0.159, loss_ctc=11.156, loss_att=4.908, acc=0.966, loss=6.782, backward_time=0.254, grad_norm=76.988, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.029, optim0_lr0=7.615e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 21:46:38,145 (trainer:779) INFO: 42epoch:train:6196-6608batch: iter_time=2.233e-04, forward_time=0.160, loss_ctc=11.249, loss_att=4.923, acc=0.971, loss=6.821, backward_time=0.256, grad_norm=76.112, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.030, optim0_lr0=7.610e-04, train_time=1.193
+[bmi2:0/4] 2024-07-08 21:50:43,803 (trainer:779) INFO: 42epoch:train:6609-7021batch: iter_time=2.604e-04, forward_time=0.160, loss_ctc=11.212, loss_att=4.888, acc=0.969, loss=6.785, backward_time=0.255, grad_norm=76.933, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.030, optim0_lr0=7.606e-04, train_time=1.190
+[bmi2:0/4] 2024-07-08 21:54:49,274 (trainer:779) INFO: 42epoch:train:7022-7434batch: iter_time=2.117e-04, forward_time=0.158, loss_ctc=11.485, loss_att=5.011, acc=0.972, loss=6.953, backward_time=0.254, grad_norm=82.794, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.030, optim0_lr0=7.601e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 21:58:53,809 (trainer:779) INFO: 42epoch:train:7435-7847batch: iter_time=2.055e-04, forward_time=0.159, loss_ctc=11.329, loss_att=4.929, acc=0.972, loss=6.849, backward_time=0.255, grad_norm=80.763, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.030, optim0_lr0=7.597e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 22:02:57,575 (trainer:779) INFO: 42epoch:train:7848-8260batch: iter_time=2.395e-04, forward_time=0.158, loss_ctc=11.209, loss_att=4.915, acc=0.966, loss=6.803, backward_time=0.254, grad_norm=78.767, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.029, optim0_lr0=7.592e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 22:03:56,037 (trainer:365) INFO: 42epoch results: [train] iter_time=2.799e-04, forward_time=0.159, loss_ctc=11.224, loss_att=4.911, acc=0.970, loss=6.805, backward_time=0.253, grad_norm=78.755, clip=100.000, loss_scale=3.340e+30, optim_step_time=0.031, optim0_lr0=7.636e-04, train_time=1.187, time=1 hour, 21 minutes and 46.11 seconds, total_count=347214, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.569, cer_ctc=0.042, loss_att=5.812, acc=0.948, cer=0.033, wer=0.491, loss=6.939, time=17.98 seconds, total_count=1428, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.47 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 22:04:00,635 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 22:04:00,713 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/40epoch.pth
+[bmi2:0/4] 2024-07-08 22:04:00,713 (trainer:299) INFO: 43/100epoch started. Estimated time to finish: 3 days, 8 hours and 35 minutes
+[bmi2:0/4] 2024-07-08 22:08:15,763 (trainer:779) INFO: 43epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=10.947, loss_att=4.785, acc=0.970, loss=6.634, backward_time=0.255, grad_norm=78.248, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.030, optim0_lr0=7.588e-04, train_time=1.235
+[bmi2:0/4] 2024-07-08 22:12:19,680 (trainer:779) INFO: 43epoch:train:414-826batch: iter_time=2.018e-04, forward_time=0.158, loss_ctc=11.039, loss_att=4.820, acc=0.971, loss=6.686, backward_time=0.255, grad_norm=77.710, clip=100.000, loss_scale=5.071e+30, optim_step_time=0.029, optim0_lr0=7.583e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 22:16:26,909 (trainer:779) INFO: 43epoch:train:827-1239batch: iter_time=2.579e-04, forward_time=0.162, loss_ctc=11.123, loss_att=4.886, acc=0.971, loss=6.757, backward_time=0.256, grad_norm=77.968, clip=100.000, loss_scale=1.012e+31, optim_step_time=0.031, optim0_lr0=7.579e-04, train_time=1.198
+[bmi2:0/4] 2024-07-08 22:20:33,419 (trainer:779) INFO: 43epoch:train:1240-1652batch: iter_time=2.444e-04, forward_time=0.160, loss_ctc=11.086, loss_att=4.842, acc=0.973, loss=6.715, backward_time=0.255, grad_norm=86.117, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.031, optim0_lr0=7.574e-04, train_time=1.193
+[bmi2:0/4] 2024-07-08 22:24:36,938 (trainer:779) INFO: 43epoch:train:1653-2065batch: iter_time=2.374e-04, forward_time=0.159, loss_ctc=11.061, loss_att=4.860, acc=0.968, loss=6.720, backward_time=0.253, grad_norm=77.428, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.032, optim0_lr0=7.570e-04, train_time=1.180
+[bmi2:0/4] 2024-07-08 22:28:43,014 (trainer:779) INFO: 43epoch:train:2066-2478batch: iter_time=2.357e-04, forward_time=0.160, loss_ctc=10.880, loss_att=4.787, acc=0.967, loss=6.615, backward_time=0.254, grad_norm=75.736, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.034, optim0_lr0=7.565e-04, train_time=1.191
+[bmi2:0/4] 2024-07-08 22:32:48,208 (trainer:779) INFO: 43epoch:train:2479-2891batch: iter_time=2.381e-04, forward_time=0.161, loss_ctc=11.087, loss_att=4.840, acc=0.969, loss=6.714, backward_time=0.254, grad_norm=77.801, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.032, optim0_lr0=7.561e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 22:36:52,822 (trainer:779) INFO: 43epoch:train:2892-3304batch: iter_time=2.200e-04, forward_time=0.159, loss_ctc=11.173, loss_att=4.874, acc=0.969, loss=6.763, backward_time=0.253, grad_norm=69.394, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.030, optim0_lr0=7.556e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 22:40:56,610 (trainer:779) INFO: 43epoch:train:3305-3717batch: iter_time=2.383e-04, forward_time=0.158, loss_ctc=11.242, loss_att=4.913, acc=0.971, loss=6.811, backward_time=0.254, grad_norm=89.397, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.031, optim0_lr0=7.552e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 22:45:01,229 (trainer:779) INFO: 43epoch:train:3718-4130batch: iter_time=2.652e-04, forward_time=0.159, loss_ctc=11.039, loss_att=4.865, acc=0.969, loss=6.717, backward_time=0.252, grad_norm=87.525, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.031, optim0_lr0=7.547e-04, train_time=1.184
+[bmi2:0/4] 2024-07-08 22:49:04,949 (trainer:779) INFO: 43epoch:train:4131-4543batch: iter_time=2.222e-04, forward_time=0.158, loss_ctc=11.179, loss_att=4.896, acc=0.970, loss=6.781, backward_time=0.253, grad_norm=85.504, clip=100.000, loss_scale=1.014e+31, optim_step_time=0.031, optim0_lr0=7.543e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 22:53:09,905 (trainer:779) INFO: 43epoch:train:4544-4956batch: iter_time=2.351e-04, forward_time=0.158, loss_ctc=11.154, loss_att=4.871, acc=0.974, loss=6.756, backward_time=0.253, grad_norm=80.847, clip=100.000, loss_scale=1.328e+31, optim_step_time=0.032, optim0_lr0=7.539e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 22:57:12,856 (trainer:779) INFO: 43epoch:train:4957-5369batch: iter_time=2.208e-04, forward_time=0.159, loss_ctc=11.283, loss_att=4.882, acc=0.970, loss=6.802, backward_time=0.252, grad_norm=81.184, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.031, optim0_lr0=7.534e-04, train_time=1.177
+[bmi2:0/4] 2024-07-08 23:01:18,952 (trainer:779) INFO: 43epoch:train:5370-5782batch: iter_time=2.208e-04, forward_time=0.160, loss_ctc=11.009, loss_att=4.838, acc=0.967, loss=6.689, backward_time=0.254, grad_norm=83.778, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.032, optim0_lr0=7.530e-04, train_time=1.191
+[bmi2:0/4] 2024-07-08 23:05:22,377 (trainer:779) INFO: 43epoch:train:5783-6195batch: iter_time=3.483e-04, forward_time=0.159, loss_ctc=11.301, loss_att=4.921, acc=0.969, loss=6.835, backward_time=0.252, grad_norm=78.836, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.032, optim0_lr0=7.525e-04, train_time=1.179
+[bmi2:0/4] 2024-07-08 23:09:26,853 (trainer:779) INFO: 43epoch:train:6196-6608batch: iter_time=2.146e-04, forward_time=0.159, loss_ctc=11.218, loss_att=4.922, acc=0.971, loss=6.811, backward_time=0.252, grad_norm=79.372, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.032, optim0_lr0=7.521e-04, train_time=1.183
+[bmi2:0/4] 2024-07-08 23:13:30,833 (trainer:779) INFO: 43epoch:train:6609-7021batch: iter_time=2.235e-04, forward_time=0.159, loss_ctc=11.194, loss_att=4.885, acc=0.971, loss=6.778, backward_time=0.253, grad_norm=78.996, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.032, optim0_lr0=7.517e-04, train_time=1.182
+[bmi2:0/4] 2024-07-08 23:17:37,910 (trainer:779) INFO: 43epoch:train:7022-7434batch: iter_time=2.243e-04, forward_time=0.163, loss_ctc=11.378, loss_att=4.953, acc=0.974, loss=6.881, backward_time=0.255, grad_norm=85.955, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.032, optim0_lr0=7.512e-04, train_time=1.195
+[bmi2:0/4] 2024-07-08 23:21:42,993 (trainer:779) INFO: 43epoch:train:7435-7847batch: iter_time=2.794e-04, forward_time=0.159, loss_ctc=11.122, loss_att=4.855, acc=0.968, loss=6.735, backward_time=0.255, grad_norm=81.951, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.029, optim0_lr0=7.508e-04, train_time=1.188
+[bmi2:0/4] 2024-07-08 23:25:45,850 (trainer:779) INFO: 43epoch:train:7848-8260batch: iter_time=2.135e-04, forward_time=0.159, loss_ctc=11.066, loss_att=4.845, acc=0.967, loss=6.711, backward_time=0.252, grad_norm=77.460, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.029, optim0_lr0=7.503e-04, train_time=1.175
+[bmi2:0/4] 2024-07-08 23:26:41,056 (trainer:365) INFO: 43epoch results: [train] iter_time=2.900e-04, forward_time=0.159, loss_ctc=11.127, loss_att=4.866, acc=0.970, loss=6.744, backward_time=0.254, grad_norm=80.547, clip=100.000, loss_scale=1.385e+31, optim_step_time=0.031, optim0_lr0=7.545e-04, train_time=1.187, time=1 hour, 21 minutes and 49.96 seconds, total_count=355481, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.678, cer_ctc=0.043, loss_att=5.874, acc=0.948, cer=0.034, wer=0.499, loss=7.015, time=16.93 seconds, total_count=1462, gpu_max_cached_mem_GB=22.529, [att_plot] time=33.46 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-08 23:26:45,212 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-08 23:26:45,213 (trainer:299) INFO: 44/100epoch started. Estimated time to finish: 3 days, 7 hours and 11 minutes
+[bmi2:0/4] 2024-07-08 23:30:56,618 (trainer:779) INFO: 44epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=11.071, loss_att=4.811, acc=0.972, loss=6.689, backward_time=0.252, grad_norm=83.472, clip=100.000, loss_scale=2.028e+31, optim_step_time=0.028, optim0_lr0=7.499e-04, train_time=1.218
+[bmi2:0/4] 2024-07-08 23:35:00,624 (trainer:779) INFO: 44epoch:train:414-826batch: iter_time=2.144e-04, forward_time=0.157, loss_ctc=11.004, loss_att=4.796, acc=0.968, loss=6.658, backward_time=0.253, grad_norm=80.753, clip=100.000, loss_scale=3.322e+31, optim_step_time=0.028, optim0_lr0=7.495e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 23:39:03,732 (trainer:779) INFO: 44epoch:train:827-1239batch: iter_time=2.049e-04, forward_time=0.157, loss_ctc=11.012, loss_att=4.796, acc=0.971, loss=6.660, backward_time=0.252, grad_norm=76.993, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.028, optim0_lr0=7.490e-04, train_time=1.178
+[bmi2:0/4] 2024-07-08 23:43:06,379 (trainer:779) INFO: 44epoch:train:1240-1652batch: iter_time=2.200e-04, forward_time=0.157, loss_ctc=10.966, loss_att=4.776, acc=0.968, loss=6.633, backward_time=0.252, grad_norm=76.294, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.029, optim0_lr0=7.486e-04, train_time=1.174
+[bmi2:0/4] 2024-07-08 23:47:11,461 (trainer:779) INFO: 44epoch:train:1653-2065batch: iter_time=2.658e-04, forward_time=0.160, loss_ctc=11.023, loss_att=4.768, acc=0.973, loss=6.645, backward_time=0.254, grad_norm=78.891, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.032, optim0_lr0=7.482e-04, train_time=1.187
+[bmi2:0/4] 2024-07-08 23:51:15,453 (trainer:779) INFO: 44epoch:train:2066-2478batch: iter_time=2.437e-04, forward_time=0.159, loss_ctc=11.036, loss_att=4.821, acc=0.969, loss=6.686, backward_time=0.252, grad_norm=78.807, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.032, optim0_lr0=7.477e-04, train_time=1.181
+[bmi2:0/4] 2024-07-08 23:55:20,198 (trainer:779) INFO: 44epoch:train:2479-2891batch: iter_time=2.244e-04, forward_time=0.160, loss_ctc=10.930, loss_att=4.791, acc=0.972, loss=6.633, backward_time=0.254, grad_norm=82.810, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.032, optim0_lr0=7.473e-04, train_time=1.185
+[bmi2:0/4] 2024-07-08 23:59:24,360 (trainer:779) INFO: 44epoch:train:2892-3304batch: iter_time=2.226e-04, forward_time=0.160, loss_ctc=11.039, loss_att=4.809, acc=0.971, loss=6.678, backward_time=0.252, grad_norm=76.802, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.031, optim0_lr0=7.469e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 00:03:28,387 (trainer:779) INFO: 44epoch:train:3305-3717batch: iter_time=2.208e-04, forward_time=0.157, loss_ctc=11.133, loss_att=4.856, acc=0.970, loss=6.739, backward_time=0.255, grad_norm=74.865, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.030, optim0_lr0=7.464e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 00:07:33,055 (trainer:779) INFO: 44epoch:train:3718-4130batch: iter_time=1.997e-04, forward_time=0.156, loss_ctc=11.066, loss_att=4.813, acc=0.973, loss=6.689, backward_time=0.254, grad_norm=72.964, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.029, optim0_lr0=7.460e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 00:11:37,264 (trainer:779) INFO: 44epoch:train:4131-4543batch: iter_time=1.908e-04, forward_time=0.157, loss_ctc=11.037, loss_att=4.808, acc=0.970, loss=6.677, backward_time=0.255, grad_norm=76.858, clip=100.000, loss_scale=4.056e+31, optim_step_time=0.030, optim0_lr0=7.456e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 00:15:42,425 (trainer:779) INFO: 44epoch:train:4544-4956batch: iter_time=1.960e-04, forward_time=0.157, loss_ctc=10.748, loss_att=4.718, acc=0.964, loss=6.527, backward_time=0.254, grad_norm=78.361, clip=100.000, loss_scale=7.917e+31, optim_step_time=0.030, optim0_lr0=7.452e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 00:19:45,947 (trainer:779) INFO: 44epoch:train:4957-5369batch: iter_time=1.840e-04, forward_time=0.157, loss_ctc=11.021, loss_att=4.822, acc=0.972, loss=6.682, backward_time=0.254, grad_norm=79.378, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.447e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 00:23:49,709 (trainer:779) INFO: 44epoch:train:5370-5782batch: iter_time=1.856e-04, forward_time=0.156, loss_ctc=11.025, loss_att=4.820, acc=0.973, loss=6.681, backward_time=0.253, grad_norm=77.266, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.443e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 00:27:52,546 (trainer:779) INFO: 44epoch:train:5783-6195batch: iter_time=1.881e-04, forward_time=0.155, loss_ctc=11.066, loss_att=4.834, acc=0.969, loss=6.704, backward_time=0.253, grad_norm=79.129, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.029, optim0_lr0=7.439e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 00:31:56,330 (trainer:779) INFO: 44epoch:train:6196-6608batch: iter_time=2.011e-04, forward_time=0.156, loss_ctc=11.057, loss_att=4.838, acc=0.970, loss=6.704, backward_time=0.253, grad_norm=85.538, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.029, optim0_lr0=7.435e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 00:36:00,233 (trainer:779) INFO: 44epoch:train:6609-7021batch: iter_time=2.099e-04, forward_time=0.158, loss_ctc=10.991, loss_att=4.797, acc=0.969, loss=6.655, backward_time=0.254, grad_norm=73.247, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.430e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 00:40:04,323 (trainer:779) INFO: 44epoch:train:7022-7434batch: iter_time=1.982e-04, forward_time=0.156, loss_ctc=11.111, loss_att=4.885, acc=0.971, loss=6.753, backward_time=0.254, grad_norm=83.752, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.426e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 00:44:08,388 (trainer:779) INFO: 44epoch:train:7435-7847batch: iter_time=2.064e-04, forward_time=0.157, loss_ctc=11.005, loss_att=4.819, acc=0.971, loss=6.675, backward_time=0.255, grad_norm=76.792, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.422e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 00:48:13,888 (trainer:779) INFO: 44epoch:train:7848-8260batch: iter_time=1.830e-04, forward_time=0.158, loss_ctc=11.110, loss_att=4.843, acc=0.973, loss=6.723, backward_time=0.255, grad_norm=78.943, clip=100.000, loss_scale=8.113e+31, optim_step_time=0.030, optim0_lr0=7.418e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 00:49:09,766 (trainer:365) INFO: 44epoch results: [train] iter_time=2.598e-04, forward_time=0.157, loss_ctc=11.021, loss_att=4.811, acc=0.970, loss=6.674, backward_time=0.253, grad_norm=78.591, clip=100.000, loss_scale=5.736e+31, optim_step_time=0.030, optim0_lr0=7.458e-04, train_time=1.184, time=1 hour, 21 minutes and 33.74 seconds, total_count=363748, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.813, cer_ctc=0.043, loss_att=6.106, acc=0.945, cer=0.034, wer=0.491, loss=7.218, time=17.09 seconds, total_count=1496, gpu_max_cached_mem_GB=22.529, [att_plot] time=33.71 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 00:49:13,963 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 00:49:14,039 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/43epoch.pth
+[bmi2:0/4] 2024-07-09 00:49:14,040 (trainer:299) INFO: 45/100epoch started. Estimated time to finish: 3 days, 5 hours and 47 minutes
+[bmi2:0/4] 2024-07-09 00:53:28,285 (trainer:779) INFO: 45epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=10.716, loss_att=4.663, acc=0.972, loss=6.479, backward_time=0.255, grad_norm=80.318, clip=100.000, loss_scale=1.040e+32, optim_step_time=0.030, optim0_lr0=7.413e-04, train_time=1.232
+[bmi2:0/4] 2024-07-09 00:57:31,637 (trainer:779) INFO: 45epoch:train:414-826batch: iter_time=1.953e-04, forward_time=0.155, loss_ctc=10.718, loss_att=4.669, acc=0.969, loss=6.484, backward_time=0.253, grad_norm=71.921, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.409e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 01:01:36,319 (trainer:779) INFO: 45epoch:train:827-1239batch: iter_time=1.877e-04, forward_time=0.159, loss_ctc=10.949, loss_att=4.822, acc=0.972, loss=6.660, backward_time=0.255, grad_norm=86.012, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.405e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 01:05:40,012 (trainer:779) INFO: 45epoch:train:1240-1652batch: iter_time=1.977e-04, forward_time=0.156, loss_ctc=10.672, loss_att=4.670, acc=0.967, loss=6.471, backward_time=0.254, grad_norm=77.906, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.401e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 01:09:43,444 (trainer:779) INFO: 45epoch:train:1653-2065batch: iter_time=1.874e-04, forward_time=0.157, loss_ctc=10.832, loss_att=4.755, acc=0.971, loss=6.578, backward_time=0.254, grad_norm=77.321, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.397e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 01:13:46,560 (trainer:779) INFO: 45epoch:train:2066-2478batch: iter_time=1.865e-04, forward_time=0.157, loss_ctc=10.886, loss_att=4.728, acc=0.970, loss=6.575, backward_time=0.253, grad_norm=78.056, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.392e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 01:17:49,747 (trainer:779) INFO: 45epoch:train:2479-2891batch: iter_time=1.855e-04, forward_time=0.157, loss_ctc=10.932, loss_att=4.773, acc=0.971, loss=6.621, backward_time=0.254, grad_norm=77.299, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.388e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 01:21:54,970 (trainer:779) INFO: 45epoch:train:2892-3304batch: iter_time=1.844e-04, forward_time=0.157, loss_ctc=10.875, loss_att=4.754, acc=0.968, loss=6.590, backward_time=0.254, grad_norm=77.380, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.384e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 01:25:58,415 (trainer:779) INFO: 45epoch:train:3305-3717batch: iter_time=1.954e-04, forward_time=0.157, loss_ctc=10.971, loss_att=4.759, acc=0.971, loss=6.622, backward_time=0.254, grad_norm=83.006, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.030, optim0_lr0=7.380e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 01:30:01,892 (trainer:779) INFO: 45epoch:train:3718-4130batch: iter_time=1.887e-04, forward_time=0.156, loss_ctc=10.982, loss_att=4.761, acc=0.974, loss=6.628, backward_time=0.253, grad_norm=75.312, clip=100.000, loss_scale=1.623e+32, optim_step_time=0.029, optim0_lr0=7.376e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 01:34:06,100 (trainer:779) INFO: 45epoch:train:4131-4543batch: iter_time=1.820e-04, forward_time=0.157, loss_ctc=11.043, loss_att=4.806, acc=0.971, loss=6.677, backward_time=0.254, grad_norm=75.865, clip=100.000, loss_scale=2.591e+32, optim_step_time=0.030, optim0_lr0=7.372e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 01:38:10,355 (trainer:779) INFO: 45epoch:train:4544-4956batch: iter_time=1.870e-04, forward_time=0.158, loss_ctc=10.857, loss_att=4.724, acc=0.970, loss=6.564, backward_time=0.254, grad_norm=73.696, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.031, optim0_lr0=7.368e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 01:42:14,281 (trainer:779) INFO: 45epoch:train:4957-5369batch: iter_time=1.958e-04, forward_time=0.157, loss_ctc=10.892, loss_att=4.747, acc=0.971, loss=6.591, backward_time=0.254, grad_norm=76.899, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.363e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 01:46:20,440 (trainer:779) INFO: 45epoch:train:5370-5782batch: iter_time=1.844e-04, forward_time=0.159, loss_ctc=10.997, loss_att=4.789, acc=0.973, loss=6.652, backward_time=0.255, grad_norm=73.542, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.359e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 01:50:24,266 (trainer:779) INFO: 45epoch:train:5783-6195batch: iter_time=1.908e-04, forward_time=0.157, loss_ctc=10.982, loss_att=4.776, acc=0.972, loss=6.638, backward_time=0.253, grad_norm=82.421, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.355e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 01:54:27,453 (trainer:779) INFO: 45epoch:train:6196-6608batch: iter_time=1.916e-04, forward_time=0.156, loss_ctc=10.932, loss_att=4.769, acc=0.970, loss=6.618, backward_time=0.253, grad_norm=77.274, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.351e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 01:58:30,636 (trainer:779) INFO: 45epoch:train:6609-7021batch: iter_time=1.790e-04, forward_time=0.156, loss_ctc=10.765, loss_att=4.701, acc=0.971, loss=6.520, backward_time=0.253, grad_norm=75.036, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.347e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 02:02:34,108 (trainer:779) INFO: 45epoch:train:7022-7434batch: iter_time=1.891e-04, forward_time=0.155, loss_ctc=10.952, loss_att=4.759, acc=0.970, loss=6.617, backward_time=0.253, grad_norm=78.459, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.343e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 02:06:37,472 (trainer:779) INFO: 45epoch:train:7435-7847batch: iter_time=1.806e-04, forward_time=0.157, loss_ctc=10.952, loss_att=4.763, acc=0.970, loss=6.619, backward_time=0.254, grad_norm=78.922, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.339e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 02:10:42,427 (trainer:779) INFO: 45epoch:train:7848-8260batch: iter_time=1.945e-04, forward_time=0.158, loss_ctc=10.979, loss_att=4.788, acc=0.974, loss=6.645, backward_time=0.254, grad_norm=80.598, clip=100.000, loss_scale=3.245e+32, optim_step_time=0.030, optim0_lr0=7.335e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 02:11:37,383 (trainer:365) INFO: 45epoch results: [train] iter_time=2.345e-04, forward_time=0.157, loss_ctc=10.892, loss_att=4.748, acc=0.971, loss=6.591, backward_time=0.254, grad_norm=77.859, clip=100.000, loss_scale=2.373e+32, optim_step_time=0.030, optim0_lr0=7.374e-04, train_time=1.184, time=1 hour, 21 minutes and 33.5 seconds, total_count=372015, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.784, cer_ctc=0.043, loss_att=5.858, acc=0.948, cer=0.033, wer=0.493, loss=7.036, time=17.07 seconds, total_count=1530, gpu_max_cached_mem_GB=22.529, [att_plot] time=32.77 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 02:11:42,314 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 02:11:42,358 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/44epoch.pth
+[bmi2:0/4] 2024-07-09 02:11:42,358 (trainer:299) INFO: 46/100epoch started. Estimated time to finish: 3 days, 4 hours and 22 minutes
+[bmi2:0/4] 2024-07-09 02:15:51,628 (trainer:779) INFO: 46epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=10.788, loss_att=4.686, acc=0.972, loss=6.516, backward_time=0.253, grad_norm=78.490, clip=100.000, loss_scale=6.254e+32, optim_step_time=0.030, optim0_lr0=7.331e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 02:19:55,338 (trainer:779) INFO: 46epoch:train:414-826batch: iter_time=1.878e-04, forward_time=0.157, loss_ctc=10.701, loss_att=4.663, acc=0.970, loss=6.475, backward_time=0.254, grad_norm=75.294, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.327e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 02:23:57,536 (trainer:779) INFO: 46epoch:train:827-1239batch: iter_time=1.836e-04, forward_time=0.155, loss_ctc=10.872, loss_att=4.710, acc=0.972, loss=6.559, backward_time=0.253, grad_norm=80.184, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.029, optim0_lr0=7.322e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 02:28:00,556 (trainer:779) INFO: 46epoch:train:1240-1652batch: iter_time=1.774e-04, forward_time=0.156, loss_ctc=10.660, loss_att=4.660, acc=0.966, loss=6.460, backward_time=0.253, grad_norm=75.855, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.318e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 02:32:03,022 (trainer:779) INFO: 46epoch:train:1653-2065batch: iter_time=1.899e-04, forward_time=0.156, loss_ctc=10.854, loss_att=4.733, acc=0.972, loss=6.569, backward_time=0.254, grad_norm=81.700, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.314e-04, train_time=1.175
+[bmi2:0/4] 2024-07-09 02:36:07,542 (trainer:779) INFO: 46epoch:train:2066-2478batch: iter_time=1.919e-04, forward_time=0.157, loss_ctc=10.706, loss_att=4.684, acc=0.972, loss=6.491, backward_time=0.254, grad_norm=76.185, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.310e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 02:40:10,625 (trainer:779) INFO: 46epoch:train:2479-2891batch: iter_time=1.916e-04, forward_time=0.156, loss_ctc=10.885, loss_att=4.776, acc=0.973, loss=6.609, backward_time=0.254, grad_norm=87.543, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.306e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 02:44:14,291 (trainer:779) INFO: 46epoch:train:2892-3304batch: iter_time=1.785e-04, forward_time=0.156, loss_ctc=10.802, loss_att=4.729, acc=0.974, loss=6.551, backward_time=0.253, grad_norm=74.696, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.302e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 02:48:16,509 (trainer:779) INFO: 46epoch:train:3305-3717batch: iter_time=1.778e-04, forward_time=0.155, loss_ctc=10.719, loss_att=4.678, acc=0.969, loss=6.490, backward_time=0.253, grad_norm=74.174, clip=100.000, loss_scale=6.490e+32, optim_step_time=0.030, optim0_lr0=7.298e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 02:52:21,586 (trainer:779) INFO: 46epoch:train:3718-4130batch: iter_time=1.900e-04, forward_time=0.159, loss_ctc=10.649, loss_att=4.646, acc=0.970, loss=6.447, backward_time=0.254, grad_norm=73.333, clip=100.000, loss_scale=8.058e+32, optim_step_time=0.030, optim0_lr0=7.294e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 02:56:25,209 (trainer:779) INFO: 46epoch:train:4131-4543batch: iter_time=2.011e-04, forward_time=0.158, loss_ctc=10.775, loss_att=4.693, acc=0.973, loss=6.518, backward_time=0.254, grad_norm=77.654, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.030, optim0_lr0=7.290e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 03:00:29,283 (trainer:779) INFO: 46epoch:train:4544-4956batch: iter_time=1.893e-04, forward_time=0.156, loss_ctc=10.847, loss_att=4.740, acc=0.973, loss=6.572, backward_time=0.254, grad_norm=82.517, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.030, optim0_lr0=7.286e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 03:04:33,029 (trainer:779) INFO: 46epoch:train:4957-5369batch: iter_time=1.817e-04, forward_time=0.157, loss_ctc=10.925, loss_att=4.751, acc=0.971, loss=6.603, backward_time=0.254, grad_norm=77.043, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.029, optim0_lr0=7.282e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 03:08:37,603 (trainer:779) INFO: 46epoch:train:5370-5782batch: iter_time=1.763e-04, forward_time=0.157, loss_ctc=10.839, loss_att=4.701, acc=0.970, loss=6.542, backward_time=0.253, grad_norm=77.178, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.030, optim0_lr0=7.278e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 03:12:41,503 (trainer:779) INFO: 46epoch:train:5783-6195batch: iter_time=1.848e-04, forward_time=0.157, loss_ctc=10.873, loss_att=4.758, acc=0.972, loss=6.593, backward_time=0.254, grad_norm=82.166, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.029, optim0_lr0=7.274e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 03:16:45,705 (trainer:779) INFO: 46epoch:train:6196-6608batch: iter_time=1.831e-04, forward_time=0.157, loss_ctc=10.603, loss_att=4.624, acc=0.968, loss=6.417, backward_time=0.254, grad_norm=75.909, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.029, optim0_lr0=7.270e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 03:20:50,268 (trainer:779) INFO: 46epoch:train:6609-7021batch: iter_time=1.803e-04, forward_time=0.158, loss_ctc=10.789, loss_att=4.694, acc=0.973, loss=6.522, backward_time=0.255, grad_norm=78.037, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.030, optim0_lr0=7.266e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 03:24:55,487 (trainer:779) INFO: 46epoch:train:7022-7434batch: iter_time=1.811e-04, forward_time=0.158, loss_ctc=10.888, loss_att=4.751, acc=0.973, loss=6.592, backward_time=0.254, grad_norm=75.553, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.030, optim0_lr0=7.262e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 03:28:57,850 (trainer:779) INFO: 46epoch:train:7435-7847batch: iter_time=1.802e-04, forward_time=0.155, loss_ctc=10.809, loss_att=4.692, acc=0.968, loss=6.527, backward_time=0.252, grad_norm=76.557, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.029, optim0_lr0=7.258e-04, train_time=1.174
+[bmi2:0/4] 2024-07-09 03:33:02,046 (trainer:779) INFO: 46epoch:train:7848-8260batch: iter_time=1.825e-04, forward_time=0.157, loss_ctc=10.919, loss_att=4.737, acc=0.972, loss=6.592, backward_time=0.254, grad_norm=77.604, clip=100.000, loss_scale=2.019e+33, optim_step_time=0.030, optim0_lr0=7.255e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 03:33:57,972 (trainer:365) INFO: 46epoch results: [train] iter_time=2.273e-04, forward_time=0.157, loss_ctc=10.795, loss_att=4.705, acc=0.971, loss=6.532, backward_time=0.254, grad_norm=77.904, clip=100.000, loss_scale=1.018e+33, optim_step_time=0.030, optim0_lr0=7.292e-04, train_time=1.181, time=1 hour, 21 minutes and 24.64 seconds, total_count=380282, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.647, cer_ctc=0.041, loss_att=5.670, acc=0.949, cer=0.033, wer=0.492, loss=6.863, time=17.12 seconds, total_count=1564, gpu_max_cached_mem_GB=22.529, [att_plot] time=33.85 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 03:34:02,464 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 03:34:02,605 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/42epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/45epoch.pth
+[bmi2:0/4] 2024-07-09 03:34:02,605 (trainer:299) INFO: 47/100epoch started. Estimated time to finish: 3 days, 2 hours and 58 minutes
+[bmi2:0/4] 2024-07-09 03:38:16,128 (trainer:779) INFO: 47epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=10.568, loss_att=4.598, acc=0.973, loss=6.389, backward_time=0.254, grad_norm=76.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.251e-04, train_time=1.228
+[bmi2:0/4] 2024-07-09 03:42:20,423 (trainer:779) INFO: 47epoch:train:414-826batch: iter_time=1.883e-04, forward_time=0.158, loss_ctc=10.449, loss_att=4.560, acc=0.969, loss=6.327, backward_time=0.255, grad_norm=72.949, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.247e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 03:46:24,556 (trainer:779) INFO: 47epoch:train:827-1239batch: iter_time=1.882e-04, forward_time=0.157, loss_ctc=10.678, loss_att=4.693, acc=0.972, loss=6.489, backward_time=0.254, grad_norm=84.049, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=7.243e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 03:50:26,509 (trainer:779) INFO: 47epoch:train:1240-1652batch: iter_time=1.889e-04, forward_time=0.154, loss_ctc=10.630, loss_att=4.626, acc=0.971, loss=6.427, backward_time=0.252, grad_norm=75.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.239e-04, train_time=1.171
+[bmi2:0/4] 2024-07-09 03:54:29,835 (trainer:779) INFO: 47epoch:train:1653-2065batch: iter_time=1.908e-04, forward_time=0.157, loss_ctc=10.644, loss_att=4.620, acc=0.974, loss=6.427, backward_time=0.254, grad_norm=77.681, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=7.235e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 03:58:34,367 (trainer:779) INFO: 47epoch:train:2066-2478batch: iter_time=2.002e-04, forward_time=0.157, loss_ctc=10.542, loss_att=4.609, acc=0.973, loss=6.389, backward_time=0.254, grad_norm=74.091, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.231e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 04:02:35,301 (trainer:779) INFO: 47epoch:train:2479-2891batch: iter_time=1.990e-04, forward_time=0.154, loss_ctc=10.760, loss_att=4.687, acc=0.972, loss=6.509, backward_time=0.252, grad_norm=73.304, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.227e-04, train_time=1.167
+[bmi2:0/4] 2024-07-09 04:06:39,997 (trainer:779) INFO: 47epoch:train:2892-3304batch: iter_time=2.023e-04, forward_time=0.158, loss_ctc=10.722, loss_att=4.667, acc=0.971, loss=6.483, backward_time=0.254, grad_norm=79.037, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.223e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 04:10:44,736 (trainer:779) INFO: 47epoch:train:3305-3717batch: iter_time=0.003, forward_time=0.157, loss_ctc=10.745, loss_att=4.702, acc=0.972, loss=6.515, backward_time=0.255, grad_norm=80.624, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.219e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 04:14:50,554 (trainer:779) INFO: 47epoch:train:3718-4130batch: iter_time=0.004, forward_time=0.158, loss_ctc=10.631, loss_att=4.635, acc=0.969, loss=6.434, backward_time=0.253, grad_norm=72.090, clip=100.000, loss_scale=4.891e+33, optim_step_time=0.030, optim0_lr0=7.215e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 04:18:57,182 (trainer:779) INFO: 47epoch:train:4131-4543batch: iter_time=0.006, forward_time=0.158, loss_ctc=10.641, loss_att=4.657, acc=0.967, loss=6.452, backward_time=0.254, grad_norm=76.563, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.211e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 04:23:03,092 (trainer:779) INFO: 47epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.159, loss_ctc=10.816, loss_att=4.695, acc=0.973, loss=6.531, backward_time=0.255, grad_norm=80.828, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.208e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 04:27:08,353 (trainer:779) INFO: 47epoch:train:4957-5369batch: iter_time=0.003, forward_time=0.158, loss_ctc=10.691, loss_att=4.661, acc=0.972, loss=6.470, backward_time=0.255, grad_norm=72.580, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.204e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 04:31:14,543 (trainer:779) INFO: 47epoch:train:5370-5782batch: iter_time=0.003, forward_time=0.157, loss_ctc=10.773, loss_att=4.695, acc=0.972, loss=6.518, backward_time=0.255, grad_norm=78.594, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.200e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 04:35:21,001 (trainer:779) INFO: 47epoch:train:5783-6195batch: iter_time=0.006, forward_time=0.159, loss_ctc=10.749, loss_att=4.687, acc=0.969, loss=6.506, backward_time=0.254, grad_norm=79.267, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.196e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 04:39:24,379 (trainer:779) INFO: 47epoch:train:6196-6608batch: iter_time=1.968e-04, forward_time=0.156, loss_ctc=10.835, loss_att=4.724, acc=0.975, loss=6.557, backward_time=0.253, grad_norm=80.366, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.192e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 04:43:32,081 (trainer:779) INFO: 47epoch:train:6609-7021batch: iter_time=0.005, forward_time=0.160, loss_ctc=10.662, loss_att=4.636, acc=0.969, loss=6.444, backward_time=0.256, grad_norm=75.296, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.188e-04, train_time=1.200
+[bmi2:0/4] 2024-07-09 04:46:00,668 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 04:47:37,266 (trainer:779) INFO: 47epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.158, loss_ctc=10.886, loss_att=4.730, acc=0.973, loss=6.577, backward_time=0.255, grad_norm=84.565, clip=100.000, loss_scale=4.171e+33, optim_step_time=0.030, optim0_lr0=7.185e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 04:51:43,069 (trainer:779) INFO: 47epoch:train:7435-7847batch: iter_time=0.004, forward_time=0.158, loss_ctc=10.639, loss_att=4.609, acc=0.971, loss=6.418, backward_time=0.254, grad_norm=76.865, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.181e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 04:55:48,772 (trainer:779) INFO: 47epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.158, loss_ctc=10.628, loss_att=4.634, acc=0.970, loss=6.432, backward_time=0.254, grad_norm=74.574, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.177e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 04:57:16,978 (trainer:365) INFO: 47epoch results: [train] iter_time=0.002, forward_time=0.157, loss_ctc=10.684, loss_att=4.656, acc=0.971, loss=6.464, backward_time=0.254, grad_norm=77.228, clip=100.000, loss_scale=3.698e+33, optim_step_time=0.030, optim0_lr0=7.213e-04, train_time=1.188, time=1 hour, 21 minutes and 51.14 seconds, total_count=388549, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.725, cer_ctc=0.042, loss_att=5.767, acc=0.949, cer=0.033, wer=0.488, loss=6.954, time=45.94 seconds, total_count=1598, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.29 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 04:57:20,987 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 04:57:21,066 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/26epoch.pth
+[bmi2:0/4] 2024-07-09 04:57:21,066 (trainer:299) INFO: 48/100epoch started. Estimated time to finish: 3 days, 1 hour and 35 minutes
+[bmi2:0/4] 2024-07-09 05:01:32,365 (trainer:779) INFO: 48epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=10.662, loss_att=4.626, acc=0.974, loss=6.437, backward_time=0.254, grad_norm=79.415, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.173e-04, train_time=1.217
+[bmi2:0/4] 2024-07-09 05:05:35,521 (trainer:779) INFO: 48epoch:train:414-826batch: iter_time=0.001, forward_time=0.155, loss_ctc=10.661, loss_att=4.665, acc=0.972, loss=6.464, backward_time=0.253, grad_norm=89.514, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=7.169e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 05:09:39,613 (trainer:779) INFO: 48epoch:train:827-1239batch: iter_time=9.770e-04, forward_time=0.158, loss_ctc=10.563, loss_att=4.614, acc=0.971, loss=6.399, backward_time=0.253, grad_norm=74.761, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.165e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 05:13:42,989 (trainer:779) INFO: 48epoch:train:1240-1652batch: iter_time=2.090e-04, forward_time=0.157, loss_ctc=10.660, loss_att=4.626, acc=0.972, loss=6.436, backward_time=0.254, grad_norm=79.346, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.162e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 05:17:47,184 (trainer:779) INFO: 48epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.158, loss_ctc=10.604, loss_att=4.609, acc=0.970, loss=6.408, backward_time=0.253, grad_norm=82.418, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.158e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 05:21:52,317 (trainer:779) INFO: 48epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.158, loss_ctc=10.687, loss_att=4.648, acc=0.973, loss=6.459, backward_time=0.255, grad_norm=78.071, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.154e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 05:25:56,216 (trainer:779) INFO: 48epoch:train:2479-2891batch: iter_time=7.346e-04, forward_time=0.157, loss_ctc=10.538, loss_att=4.585, acc=0.970, loss=6.371, backward_time=0.254, grad_norm=77.927, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=7.150e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 05:30:01,017 (trainer:779) INFO: 48epoch:train:2892-3304batch: iter_time=0.002, forward_time=0.157, loss_ctc=10.671, loss_att=4.623, acc=0.972, loss=6.437, backward_time=0.254, grad_norm=72.362, clip=100.000, loss_scale=4.465e+33, optim_step_time=0.030, optim0_lr0=7.147e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 05:34:05,920 (trainer:779) INFO: 48epoch:train:3305-3717batch: iter_time=9.472e-04, forward_time=0.159, loss_ctc=10.426, loss_att=4.552, acc=0.970, loss=6.314, backward_time=0.255, grad_norm=73.390, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.143e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 05:38:11,049 (trainer:779) INFO: 48epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.158, loss_ctc=10.494, loss_att=4.588, acc=0.971, loss=6.360, backward_time=0.254, grad_norm=77.902, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.139e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 05:42:14,724 (trainer:779) INFO: 48epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.156, loss_ctc=10.615, loss_att=4.598, acc=0.972, loss=6.403, backward_time=0.254, grad_norm=80.129, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.135e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 05:46:20,167 (trainer:779) INFO: 48epoch:train:4544-4956batch: iter_time=0.003, forward_time=0.158, loss_ctc=10.632, loss_att=4.626, acc=0.969, loss=6.428, backward_time=0.253, grad_norm=78.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.131e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 05:50:24,772 (trainer:779) INFO: 48epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.159, loss_ctc=10.764, loss_att=4.645, acc=0.974, loss=6.481, backward_time=0.254, grad_norm=74.019, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.128e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 05:54:29,133 (trainer:779) INFO: 48epoch:train:5370-5782batch: iter_time=2.194e-04, forward_time=0.157, loss_ctc=10.610, loss_att=4.605, acc=0.971, loss=6.407, backward_time=0.255, grad_norm=77.477, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.124e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 05:58:33,157 (trainer:779) INFO: 48epoch:train:5783-6195batch: iter_time=1.872e-04, forward_time=0.159, loss_ctc=10.632, loss_att=4.598, acc=0.970, loss=6.408, backward_time=0.255, grad_norm=78.847, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.120e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 06:02:37,513 (trainer:779) INFO: 48epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.157, loss_ctc=10.641, loss_att=4.594, acc=0.969, loss=6.408, backward_time=0.254, grad_norm=73.751, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.117e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 06:06:41,048 (trainer:779) INFO: 48epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.156, loss_ctc=10.572, loss_att=4.584, acc=0.972, loss=6.380, backward_time=0.254, grad_norm=76.107, clip=100.000, loss_scale=5.369e+33, optim_step_time=0.030, optim0_lr0=7.113e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 06:10:45,624 (trainer:779) INFO: 48epoch:train:7022-7434batch: iter_time=4.807e-04, forward_time=0.158, loss_ctc=10.896, loss_att=4.736, acc=0.974, loss=6.584, backward_time=0.254, grad_norm=86.241, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=7.109e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 06:13:28,516 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 06:14:51,129 (trainer:779) INFO: 48epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.158, loss_ctc=10.614, loss_att=4.605, acc=0.974, loss=6.408, backward_time=0.255, grad_norm=79.927, clip=100.000, loss_scale=8.637e+33, optim_step_time=0.030, optim0_lr0=7.105e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 06:18:55,134 (trainer:779) INFO: 48epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.156, loss_ctc=10.908, loss_att=4.736, acc=0.972, loss=6.587, backward_time=0.254, grad_norm=78.711, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.102e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 06:19:49,664 (trainer:365) INFO: 48epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=10.641, loss_att=4.623, acc=0.972, loss=6.428, backward_time=0.254, grad_norm=78.442, clip=100.000, loss_scale=4.688e+33, optim_step_time=0.030, optim0_lr0=7.137e-04, train_time=1.185, time=1 hour, 21 minutes and 39.19 seconds, total_count=396816, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.647, cer_ctc=0.041, loss_att=5.759, acc=0.948, cer=0.033, wer=0.487, loss=6.925, time=17.23 seconds, total_count=1632, gpu_max_cached_mem_GB=22.529, [att_plot] time=32.17 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 06:19:55,131 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 06:19:55,132 (trainer:299) INFO: 49/100epoch started. Estimated time to finish: 3 days, 10 minutes and 57.78 seconds
+[bmi2:0/4] 2024-07-09 06:24:08,666 (trainer:779) INFO: 49epoch:train:1-413batch: iter_time=9.796e-04, forward_time=0.158, loss_ctc=10.630, loss_att=4.611, acc=0.975, loss=6.417, backward_time=0.254, grad_norm=74.481, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.098e-04, train_time=1.228
+[bmi2:0/4] 2024-07-09 06:28:13,693 (trainer:779) INFO: 49epoch:train:414-826batch: iter_time=1.918e-04, forward_time=0.159, loss_ctc=10.435, loss_att=4.516, acc=0.973, loss=6.292, backward_time=0.255, grad_norm=73.288, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.094e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 06:32:16,228 (trainer:779) INFO: 49epoch:train:827-1239batch: iter_time=1.798e-04, forward_time=0.156, loss_ctc=10.339, loss_att=4.490, acc=0.972, loss=6.245, backward_time=0.253, grad_norm=75.159, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.091e-04, train_time=1.175
+[bmi2:0/4] 2024-07-09 06:36:18,801 (trainer:779) INFO: 49epoch:train:1240-1652batch: iter_time=1.774e-04, forward_time=0.155, loss_ctc=10.495, loss_att=4.539, acc=0.975, loss=6.326, backward_time=0.253, grad_norm=77.642, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=7.087e-04, train_time=1.174
+[bmi2:0/4] 2024-07-09 06:40:21,034 (trainer:779) INFO: 49epoch:train:1653-2065batch: iter_time=1.844e-04, forward_time=0.156, loss_ctc=10.581, loss_att=4.571, acc=0.972, loss=6.374, backward_time=0.252, grad_norm=76.591, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.083e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 06:44:24,292 (trainer:779) INFO: 49epoch:train:2066-2478batch: iter_time=1.891e-04, forward_time=0.157, loss_ctc=10.500, loss_att=4.540, acc=0.974, loss=6.328, backward_time=0.253, grad_norm=76.172, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.080e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 06:48:28,056 (trainer:779) INFO: 49epoch:train:2479-2891batch: iter_time=1.827e-04, forward_time=0.158, loss_ctc=10.443, loss_att=4.517, acc=0.970, loss=6.295, backward_time=0.254, grad_norm=79.636, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.076e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 06:52:30,441 (trainer:779) INFO: 49epoch:train:2892-3304batch: iter_time=1.847e-04, forward_time=0.155, loss_ctc=10.639, loss_att=4.588, acc=0.972, loss=6.403, backward_time=0.252, grad_norm=78.902, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.072e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 06:56:34,085 (trainer:779) INFO: 49epoch:train:3305-3717batch: iter_time=1.908e-04, forward_time=0.157, loss_ctc=10.567, loss_att=4.573, acc=0.972, loss=6.371, backward_time=0.254, grad_norm=74.485, clip=100.000, loss_scale=8.645e+33, optim_step_time=0.030, optim0_lr0=7.069e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 07:00:26,083 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 07:00:37,987 (trainer:779) INFO: 49epoch:train:3718-4130batch: iter_time=1.926e-04, forward_time=0.156, loss_ctc=10.502, loss_att=4.562, acc=0.971, loss=6.344, backward_time=0.253, grad_norm=76.561, clip=100.000, loss_scale=1.013e+34, optim_step_time=0.030, optim0_lr0=7.065e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 07:04:41,651 (trainer:779) INFO: 49epoch:train:4131-4543batch: iter_time=1.876e-04, forward_time=0.157, loss_ctc=10.504, loss_att=4.549, acc=0.972, loss=6.336, backward_time=0.254, grad_norm=74.605, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=7.061e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 07:08:47,299 (trainer:779) INFO: 49epoch:train:4544-4956batch: iter_time=1.891e-04, forward_time=0.159, loss_ctc=10.609, loss_att=4.600, acc=0.971, loss=6.403, backward_time=0.255, grad_norm=79.990, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.058e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 07:12:50,718 (trainer:779) INFO: 49epoch:train:4957-5369batch: iter_time=1.957e-04, forward_time=0.157, loss_ctc=10.577, loss_att=4.599, acc=0.973, loss=6.392, backward_time=0.254, grad_norm=83.703, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.054e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 07:16:55,206 (trainer:779) INFO: 49epoch:train:5370-5782batch: iter_time=1.862e-04, forward_time=0.156, loss_ctc=10.493, loss_att=4.546, acc=0.968, loss=6.330, backward_time=0.254, grad_norm=74.683, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.050e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 07:20:58,940 (trainer:779) INFO: 49epoch:train:5783-6195batch: iter_time=1.841e-04, forward_time=0.158, loss_ctc=10.645, loss_att=4.615, acc=0.970, loss=6.424, backward_time=0.253, grad_norm=81.758, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.047e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 07:25:03,370 (trainer:779) INFO: 49epoch:train:6196-6608batch: iter_time=1.829e-04, forward_time=0.157, loss_ctc=10.608, loss_att=4.599, acc=0.971, loss=6.402, backward_time=0.254, grad_norm=76.215, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=7.043e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 07:29:05,895 (trainer:779) INFO: 49epoch:train:6609-7021batch: iter_time=2.098e-04, forward_time=0.155, loss_ctc=10.637, loss_att=4.596, acc=0.971, loss=6.409, backward_time=0.253, grad_norm=76.835, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=7.040e-04, train_time=1.175
+[bmi2:0/4] 2024-07-09 07:33:10,807 (trainer:779) INFO: 49epoch:train:7022-7434batch: iter_time=2.032e-04, forward_time=0.158, loss_ctc=10.538, loss_att=4.582, acc=0.972, loss=6.369, backward_time=0.255, grad_norm=79.021, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.036e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 07:37:14,763 (trainer:779) INFO: 49epoch:train:7435-7847batch: iter_time=1.891e-04, forward_time=0.158, loss_ctc=10.633, loss_att=4.583, acc=0.972, loss=6.398, backward_time=0.253, grad_norm=78.876, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.032e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 07:41:20,262 (trainer:779) INFO: 49epoch:train:7848-8260batch: iter_time=1.753e-04, forward_time=0.158, loss_ctc=10.787, loss_att=4.687, acc=0.974, loss=6.517, backward_time=0.254, grad_norm=90.252, clip=100.000, loss_scale=7.074e+33, optim_step_time=0.030, optim0_lr0=7.029e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 07:42:18,509 (trainer:365) INFO: 49epoch results: [train] iter_time=2.277e-04, forward_time=0.157, loss_ctc=10.558, loss_att=4.573, acc=0.972, loss=6.368, backward_time=0.254, grad_norm=77.947, clip=100.000, loss_scale=5.709e+33, optim_step_time=0.030, optim0_lr0=7.063e-04, train_time=1.183, time=1 hour, 21 minutes and 29.86 seconds, total_count=405083, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.737, cer_ctc=0.041, loss_att=5.632, acc=0.950, cer=0.032, wer=0.491, loss=6.864, time=16.91 seconds, total_count=1666, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.61 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 07:42:23,468 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-09 07:42:23,574 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/37epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/48epoch.pth
+[bmi2:0/4] 2024-07-09 07:42:23,575 (trainer:299) INFO: 50/100epoch started. Estimated time to finish: 2 days, 22 hours and 46 minutes
+[bmi2:0/4] 2024-07-09 07:46:34,915 (trainer:779) INFO: 50epoch:train:1-413batch: iter_time=9.568e-04, forward_time=0.155, loss_ctc=10.492, loss_att=4.544, acc=0.973, loss=6.328, backward_time=0.252, grad_norm=83.803, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=7.025e-04, train_time=1.218
+[bmi2:0/4] 2024-07-09 07:50:38,622 (trainer:779) INFO: 50epoch:train:414-826batch: iter_time=1.984e-04, forward_time=0.158, loss_ctc=10.518, loss_att=4.561, acc=0.971, loss=6.348, backward_time=0.253, grad_norm=82.839, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=7.022e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 07:54:04,729 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 07:54:42,571 (trainer:779) INFO: 50epoch:train:827-1239batch: iter_time=2.310e-04, forward_time=0.158, loss_ctc=10.591, loss_att=4.568, acc=0.971, loss=6.375, backward_time=0.252, grad_norm=77.934, clip=100.000, loss_scale=9.574e+33, optim_step_time=0.031, optim0_lr0=7.018e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 07:58:45,737 (trainer:779) INFO: 50epoch:train:1240-1652batch: iter_time=2.239e-04, forward_time=0.157, loss_ctc=10.571, loss_att=4.579, acc=0.975, loss=6.377, backward_time=0.252, grad_norm=79.218, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.015e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 08:02:50,342 (trainer:779) INFO: 50epoch:train:1653-2065batch: iter_time=2.524e-04, forward_time=0.159, loss_ctc=10.447, loss_att=4.521, acc=0.974, loss=6.299, backward_time=0.254, grad_norm=77.499, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.011e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 08:06:53,407 (trainer:779) INFO: 50epoch:train:2066-2478batch: iter_time=2.625e-04, forward_time=0.158, loss_ctc=10.418, loss_att=4.520, acc=0.969, loss=6.289, backward_time=0.252, grad_norm=74.108, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=7.007e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 08:10:57,231 (trainer:779) INFO: 50epoch:train:2479-2891batch: iter_time=2.526e-04, forward_time=0.158, loss_ctc=10.214, loss_att=4.440, acc=0.971, loss=6.172, backward_time=0.253, grad_norm=75.106, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=7.004e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 08:15:00,492 (trainer:779) INFO: 50epoch:train:2892-3304batch: iter_time=2.210e-04, forward_time=0.158, loss_ctc=10.414, loss_att=4.539, acc=0.972, loss=6.301, backward_time=0.252, grad_norm=78.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=7.000e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 08:19:04,395 (trainer:779) INFO: 50epoch:train:3305-3717batch: iter_time=2.283e-04, forward_time=0.158, loss_ctc=10.396, loss_att=4.545, acc=0.969, loss=6.301, backward_time=0.253, grad_norm=76.856, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.997e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 08:23:09,602 (trainer:779) INFO: 50epoch:train:3718-4130batch: iter_time=2.447e-04, forward_time=0.160, loss_ctc=10.399, loss_att=4.515, acc=0.973, loss=6.280, backward_time=0.255, grad_norm=72.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.993e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 08:27:13,321 (trainer:779) INFO: 50epoch:train:4131-4543batch: iter_time=2.677e-04, forward_time=0.160, loss_ctc=10.405, loss_att=4.515, acc=0.973, loss=6.282, backward_time=0.253, grad_norm=77.562, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.990e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 08:31:17,146 (trainer:779) INFO: 50epoch:train:4544-4956batch: iter_time=2.133e-04, forward_time=0.158, loss_ctc=10.512, loss_att=4.547, acc=0.976, loss=6.336, backward_time=0.253, grad_norm=76.843, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.986e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 08:35:20,708 (trainer:779) INFO: 50epoch:train:4957-5369batch: iter_time=1.931e-04, forward_time=0.159, loss_ctc=10.548, loss_att=4.598, acc=0.975, loss=6.383, backward_time=0.254, grad_norm=74.165, clip=100.000, loss_scale=7.637e+33, optim_step_time=0.029, optim0_lr0=6.983e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 08:39:24,538 (trainer:779) INFO: 50epoch:train:5370-5782batch: iter_time=2.099e-04, forward_time=0.160, loss_ctc=10.368, loss_att=4.488, acc=0.972, loss=6.252, backward_time=0.253, grad_norm=73.133, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.029, optim0_lr0=6.979e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 08:41:01,922 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 08:43:29,122 (trainer:779) INFO: 50epoch:train:5783-6195batch: iter_time=2.078e-04, forward_time=0.160, loss_ctc=10.415, loss_att=4.517, acc=0.973, loss=6.286, backward_time=0.254, grad_norm=76.846, clip=100.000, loss_scale=7.244e+33, optim_step_time=0.031, optim0_lr0=6.976e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 08:47:34,114 (trainer:779) INFO: 50epoch:train:6196-6608batch: iter_time=2.261e-04, forward_time=0.159, loss_ctc=10.380, loss_att=4.522, acc=0.968, loss=6.279, backward_time=0.255, grad_norm=74.875, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.972e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 08:51:38,321 (trainer:779) INFO: 50epoch:train:6609-7021batch: iter_time=2.006e-04, forward_time=0.158, loss_ctc=10.411, loss_att=4.502, acc=0.970, loss=6.274, backward_time=0.256, grad_norm=74.403, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=6.969e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 08:53:42,817 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 08:55:43,706 (trainer:779) INFO: 50epoch:train:7022-7434batch: iter_time=1.934e-04, forward_time=0.159, loss_ctc=10.479, loss_att=4.520, acc=0.974, loss=6.308, backward_time=0.256, grad_norm=74.503, clip=100.000, loss_scale=3.907e+33, optim_step_time=0.030, optim0_lr0=6.965e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 08:59:49,166 (trainer:779) INFO: 50epoch:train:7435-7847batch: iter_time=2.073e-04, forward_time=0.159, loss_ctc=10.571, loss_att=4.564, acc=0.974, loss=6.366, backward_time=0.255, grad_norm=80.537, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.962e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 09:03:54,711 (trainer:779) INFO: 50epoch:train:7848-8260batch: iter_time=1.969e-04, forward_time=0.159, loss_ctc=10.337, loss_att=4.476, acc=0.972, loss=6.234, backward_time=0.256, grad_norm=74.165, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=6.958e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 09:04:53,387 (trainer:365) INFO: 50epoch results: [train] iter_time=2.593e-04, forward_time=0.158, loss_ctc=10.442, loss_att=4.528, acc=0.972, loss=6.302, backward_time=0.254, grad_norm=76.749, clip=100.000, loss_scale=6.087e+33, optim_step_time=0.030, optim0_lr0=6.992e-04, train_time=1.184, time=1 hour, 21 minutes and 36.11 seconds, total_count=413350, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.537, cer_ctc=0.041, loss_att=5.955, acc=0.950, cer=0.033, wer=0.485, loss=7.030, time=17.54 seconds, total_count=1700, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.16 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 09:04:57,536 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-09 09:04:57,591 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/38epoch.pth
+[bmi2:0/4] 2024-07-09 09:04:57,591 (trainer:299) INFO: 51/100epoch started. Estimated time to finish: 2 days, 21 hours and 22 minutes
+[bmi2:0/4] 2024-07-09 09:09:12,348 (trainer:779) INFO: 51epoch:train:1-413batch: iter_time=0.001, forward_time=0.158, loss_ctc=10.247, loss_att=4.424, acc=0.975, loss=6.171, backward_time=0.255, grad_norm=72.722, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.955e-04, train_time=1.234
+[bmi2:0/4] 2024-07-09 09:13:17,790 (trainer:779) INFO: 51epoch:train:414-826batch: iter_time=2.649e-04, forward_time=0.160, loss_ctc=10.205, loss_att=4.448, acc=0.967, loss=6.175, backward_time=0.255, grad_norm=73.188, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=6.951e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 09:17:22,944 (trainer:779) INFO: 51epoch:train:827-1239batch: iter_time=2.030e-04, forward_time=0.159, loss_ctc=10.340, loss_att=4.462, acc=0.971, loss=6.225, backward_time=0.254, grad_norm=72.793, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.948e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 09:21:28,008 (trainer:779) INFO: 51epoch:train:1240-1652batch: iter_time=2.490e-04, forward_time=0.160, loss_ctc=10.378, loss_att=4.510, acc=0.974, loss=6.271, backward_time=0.254, grad_norm=75.736, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.944e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 09:25:32,969 (trainer:779) INFO: 51epoch:train:1653-2065batch: iter_time=2.557e-04, forward_time=0.161, loss_ctc=10.325, loss_att=4.506, acc=0.971, loss=6.252, backward_time=0.254, grad_norm=76.504, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.941e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 09:29:39,771 (trainer:779) INFO: 51epoch:train:2066-2478batch: iter_time=2.376e-04, forward_time=0.161, loss_ctc=10.407, loss_att=4.521, acc=0.973, loss=6.287, backward_time=0.255, grad_norm=75.992, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=6.937e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 09:33:44,799 (trainer:779) INFO: 51epoch:train:2479-2891batch: iter_time=2.673e-04, forward_time=0.159, loss_ctc=10.231, loss_att=4.425, acc=0.971, loss=6.167, backward_time=0.255, grad_norm=75.878, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=6.934e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 09:37:49,669 (trainer:779) INFO: 51epoch:train:2892-3304batch: iter_time=2.819e-04, forward_time=0.160, loss_ctc=10.423, loss_att=4.531, acc=0.972, loss=6.298, backward_time=0.254, grad_norm=79.664, clip=100.000, loss_scale=4.728e+33, optim_step_time=0.030, optim0_lr0=6.931e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 09:41:53,200 (trainer:779) INFO: 51epoch:train:3305-3717batch: iter_time=3.034e-04, forward_time=0.158, loss_ctc=10.326, loss_att=4.462, acc=0.974, loss=6.221, backward_time=0.253, grad_norm=75.342, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.927e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 09:45:58,524 (trainer:779) INFO: 51epoch:train:3718-4130batch: iter_time=2.090e-04, forward_time=0.160, loss_ctc=10.346, loss_att=4.474, acc=0.973, loss=6.235, backward_time=0.254, grad_norm=75.088, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.924e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 09:50:03,630 (trainer:779) INFO: 51epoch:train:4131-4543batch: iter_time=2.510e-04, forward_time=0.160, loss_ctc=10.299, loss_att=4.447, acc=0.975, loss=6.203, backward_time=0.255, grad_norm=74.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.920e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 09:54:08,605 (trainer:779) INFO: 51epoch:train:4544-4956batch: iter_time=2.001e-04, forward_time=0.160, loss_ctc=10.348, loss_att=4.504, acc=0.971, loss=6.258, backward_time=0.255, grad_norm=71.317, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=6.917e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 09:58:12,695 (trainer:779) INFO: 51epoch:train:4957-5369batch: iter_time=2.057e-04, forward_time=0.159, loss_ctc=10.419, loss_att=4.523, acc=0.972, loss=6.292, backward_time=0.254, grad_norm=71.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.913e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 10:02:16,357 (trainer:779) INFO: 51epoch:train:5370-5782batch: iter_time=1.836e-04, forward_time=0.157, loss_ctc=10.355, loss_att=4.495, acc=0.971, loss=6.253, backward_time=0.254, grad_norm=75.912, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=6.910e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 10:06:18,981 (trainer:779) INFO: 51epoch:train:5783-6195batch: iter_time=1.810e-04, forward_time=0.156, loss_ctc=10.316, loss_att=4.479, acc=0.974, loss=6.230, backward_time=0.253, grad_norm=74.166, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=6.907e-04, train_time=1.175
+[bmi2:0/4] 2024-07-09 10:10:20,870 (trainer:779) INFO: 51epoch:train:6196-6608batch: iter_time=1.971e-04, forward_time=0.157, loss_ctc=10.236, loss_att=4.452, acc=0.973, loss=6.187, backward_time=0.253, grad_norm=75.922, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=6.903e-04, train_time=1.171
+[bmi2:0/4] 2024-07-09 10:14:23,713 (trainer:779) INFO: 51epoch:train:6609-7021batch: iter_time=2.069e-04, forward_time=0.159, loss_ctc=10.286, loss_att=4.473, acc=0.974, loss=6.217, backward_time=0.254, grad_norm=73.719, clip=100.000, loss_scale=5.898e+33, optim_step_time=0.030, optim0_lr0=6.900e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 10:18:25,999 (trainer:779) INFO: 51epoch:train:7022-7434batch: iter_time=2.116e-04, forward_time=0.159, loss_ctc=10.353, loss_att=4.498, acc=0.972, loss=6.254, backward_time=0.254, grad_norm=75.908, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=6.896e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 10:22:30,158 (trainer:779) INFO: 51epoch:train:7435-7847batch: iter_time=2.364e-04, forward_time=0.158, loss_ctc=10.395, loss_att=4.502, acc=0.973, loss=6.270, backward_time=0.254, grad_norm=76.254, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.893e-04, train_time=1.183
+[bmi2:0/4] 2024-07-09 10:22:30,800 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 10:26:34,487 (trainer:779) INFO: 51epoch:train:7848-8260batch: iter_time=2.285e-04, forward_time=0.159, loss_ctc=10.466, loss_att=4.551, acc=0.972, loss=6.325, backward_time=0.255, grad_norm=79.549, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.890e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 10:27:39,650 (trainer:365) INFO: 51epoch results: [train] iter_time=2.703e-04, forward_time=0.159, loss_ctc=10.334, loss_att=4.484, acc=0.972, loss=6.239, backward_time=0.254, grad_norm=75.060, clip=100.000, loss_scale=4.815e+33, optim_step_time=0.030, optim0_lr0=6.922e-04, train_time=1.186, time=1 hour, 21 minutes and 42.41 seconds, total_count=421617, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.961, cer_ctc=0.043, loss_att=5.863, acc=0.948, cer=0.033, wer=0.487, loss=7.092, time=18.7 seconds, total_count=1734, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.95 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 10:27:44,380 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 10:27:44,381 (trainer:299) INFO: 52/100epoch started. Estimated time to finish: 2 days, 19 hours and 59 minutes
+[bmi2:0/4] 2024-07-09 10:31:59,722 (trainer:779) INFO: 52epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=10.075, loss_att=4.391, acc=0.972, loss=6.096, backward_time=0.255, grad_norm=80.308, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.886e-04, train_time=1.237
+[bmi2:0/4] 2024-07-09 10:36:04,533 (trainer:779) INFO: 52epoch:train:414-826batch: iter_time=2.086e-04, forward_time=0.159, loss_ctc=10.325, loss_att=4.434, acc=0.975, loss=6.202, backward_time=0.255, grad_norm=72.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.883e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 10:40:09,542 (trainer:779) INFO: 52epoch:train:827-1239batch: iter_time=2.169e-04, forward_time=0.160, loss_ctc=10.199, loss_att=4.424, acc=0.972, loss=6.156, backward_time=0.255, grad_norm=74.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.880e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 10:44:14,247 (trainer:779) INFO: 52epoch:train:1240-1652batch: iter_time=2.118e-04, forward_time=0.158, loss_ctc=10.313, loss_att=4.491, acc=0.973, loss=6.237, backward_time=0.255, grad_norm=77.572, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.876e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 10:48:17,946 (trainer:779) INFO: 52epoch:train:1653-2065batch: iter_time=2.076e-04, forward_time=0.159, loss_ctc=10.268, loss_att=4.444, acc=0.975, loss=6.192, backward_time=0.255, grad_norm=76.301, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.873e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 10:52:24,519 (trainer:779) INFO: 52epoch:train:2066-2478batch: iter_time=2.033e-04, forward_time=0.161, loss_ctc=10.325, loss_att=4.474, acc=0.976, loss=6.229, backward_time=0.257, grad_norm=76.577, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.869e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 10:56:30,959 (trainer:779) INFO: 52epoch:train:2479-2891batch: iter_time=2.126e-04, forward_time=0.160, loss_ctc=10.364, loss_att=4.473, acc=0.974, loss=6.240, backward_time=0.257, grad_norm=77.359, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.866e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 11:00:38,132 (trainer:779) INFO: 52epoch:train:2892-3304batch: iter_time=2.120e-04, forward_time=0.159, loss_ctc=10.149, loss_att=4.395, acc=0.976, loss=6.121, backward_time=0.257, grad_norm=77.636, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.863e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 11:04:44,621 (trainer:779) INFO: 52epoch:train:3305-3717batch: iter_time=2.175e-04, forward_time=0.159, loss_ctc=10.256, loss_att=4.439, acc=0.975, loss=6.184, backward_time=0.257, grad_norm=76.538, clip=100.000, loss_scale=6.881e+33, optim_step_time=0.032, optim0_lr0=6.859e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 11:08:51,538 (trainer:779) INFO: 52epoch:train:3718-4130batch: iter_time=2.109e-04, forward_time=0.159, loss_ctc=10.281, loss_att=4.465, acc=0.970, loss=6.210, backward_time=0.256, grad_norm=77.575, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.856e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 11:12:56,371 (trainer:779) INFO: 52epoch:train:4131-4543batch: iter_time=2.114e-04, forward_time=0.158, loss_ctc=10.240, loss_att=4.443, acc=0.972, loss=6.182, backward_time=0.255, grad_norm=72.220, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.853e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 11:17:03,275 (trainer:779) INFO: 52epoch:train:4544-4956batch: iter_time=2.213e-04, forward_time=0.158, loss_ctc=10.484, loss_att=4.523, acc=0.977, loss=6.311, backward_time=0.257, grad_norm=75.258, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.849e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 11:21:06,941 (trainer:779) INFO: 52epoch:train:4957-5369batch: iter_time=2.119e-04, forward_time=0.157, loss_ctc=10.340, loss_att=4.465, acc=0.974, loss=6.227, backward_time=0.255, grad_norm=81.106, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.846e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 11:22:26,684 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 11:25:11,747 (trainer:779) INFO: 52epoch:train:5370-5782batch: iter_time=2.076e-04, forward_time=0.158, loss_ctc=10.178, loss_att=4.384, acc=0.971, loss=6.122, backward_time=0.256, grad_norm=76.415, clip=100.000, loss_scale=6.856e+33, optim_step_time=0.031, optim0_lr0=6.843e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 11:29:17,241 (trainer:779) INFO: 52epoch:train:5783-6195batch: iter_time=2.006e-04, forward_time=0.158, loss_ctc=10.157, loss_att=4.401, acc=0.970, loss=6.128, backward_time=0.255, grad_norm=71.882, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.840e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 11:33:23,332 (trainer:779) INFO: 52epoch:train:6196-6608batch: iter_time=1.991e-04, forward_time=0.158, loss_ctc=10.321, loss_att=4.448, acc=0.972, loss=6.210, backward_time=0.256, grad_norm=74.136, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.836e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 11:37:29,131 (trainer:779) INFO: 52epoch:train:6609-7021batch: iter_time=2.035e-04, forward_time=0.159, loss_ctc=10.054, loss_att=4.368, acc=0.967, loss=6.074, backward_time=0.255, grad_norm=76.522, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.833e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 11:41:34,539 (trainer:779) INFO: 52epoch:train:7022-7434batch: iter_time=1.983e-04, forward_time=0.158, loss_ctc=10.463, loss_att=4.512, acc=0.975, loss=6.297, backward_time=0.255, grad_norm=76.990, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.830e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 11:45:39,598 (trainer:779) INFO: 52epoch:train:7435-7847batch: iter_time=1.965e-04, forward_time=0.158, loss_ctc=10.315, loss_att=4.448, acc=0.972, loss=6.208, backward_time=0.255, grad_norm=73.492, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.826e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 11:49:45,613 (trainer:779) INFO: 52epoch:train:7848-8260batch: iter_time=1.834e-04, forward_time=0.159, loss_ctc=10.125, loss_att=4.393, acc=0.969, loss=6.113, backward_time=0.255, grad_norm=69.299, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.823e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 11:50:45,474 (trainer:365) INFO: 52epoch results: [train] iter_time=2.483e-04, forward_time=0.159, loss_ctc=10.260, loss_att=4.440, acc=0.973, loss=6.186, backward_time=0.256, grad_norm=75.720, clip=100.000, loss_scale=6.397e+33, optim_step_time=0.031, optim0_lr0=6.854e-04, train_time=1.191, time=1 hour, 22 minutes and 6.44 seconds, total_count=429884, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.587, cer_ctc=0.040, loss_att=5.849, acc=0.950, cer=0.033, wer=0.482, loss=6.970, time=18.5 seconds, total_count=1768, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.15 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 11:50:49,878 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-09 11:50:49,981 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/46epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/51epoch.pth
+[bmi2:0/4] 2024-07-09 11:50:49,981 (trainer:299) INFO: 53/100epoch started. Estimated time to finish: 2 days, 18 hours and 35 minutes
+[bmi2:0/4] 2024-07-09 11:55:01,465 (trainer:779) INFO: 53epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=10.053, loss_att=4.369, acc=0.971, loss=6.074, backward_time=0.253, grad_norm=70.967, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.820e-04, train_time=1.218
+[bmi2:0/4] 2024-07-09 11:59:04,658 (trainer:779) INFO: 53epoch:train:414-826batch: iter_time=2.006e-04, forward_time=0.157, loss_ctc=10.068, loss_att=4.362, acc=0.971, loss=6.074, backward_time=0.254, grad_norm=71.995, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.816e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 12:03:07,188 (trainer:779) INFO: 53epoch:train:827-1239batch: iter_time=2.023e-04, forward_time=0.155, loss_ctc=10.131, loss_att=4.373, acc=0.973, loss=6.101, backward_time=0.253, grad_norm=73.808, clip=100.000, loss_scale=5.218e+33, optim_step_time=0.031, optim0_lr0=6.813e-04, train_time=1.175
+[bmi2:0/4] 2024-07-09 12:07:10,115 (trainer:779) INFO: 53epoch:train:1240-1652batch: iter_time=1.893e-04, forward_time=0.155, loss_ctc=10.030, loss_att=4.366, acc=0.970, loss=6.066, backward_time=0.254, grad_norm=71.843, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=6.810e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 12:11:15,168 (trainer:779) INFO: 53epoch:train:1653-2065batch: iter_time=2.270e-04, forward_time=0.158, loss_ctc=9.991, loss_att=4.343, acc=0.974, loss=6.038, backward_time=0.257, grad_norm=75.062, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=6.807e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 12:15:21,403 (trainer:779) INFO: 53epoch:train:2066-2478batch: iter_time=2.021e-04, forward_time=0.159, loss_ctc=10.173, loss_att=4.382, acc=0.977, loss=6.119, backward_time=0.256, grad_norm=79.226, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=6.803e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 12:19:27,777 (trainer:779) INFO: 53epoch:train:2479-2891batch: iter_time=2.058e-04, forward_time=0.159, loss_ctc=10.294, loss_att=4.454, acc=0.975, loss=6.206, backward_time=0.256, grad_norm=72.714, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.800e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 12:23:33,489 (trainer:779) INFO: 53epoch:train:2892-3304batch: iter_time=2.062e-04, forward_time=0.157, loss_ctc=10.287, loss_att=4.412, acc=0.975, loss=6.174, backward_time=0.255, grad_norm=74.677, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.797e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 12:27:39,105 (trainer:779) INFO: 53epoch:train:3305-3717batch: iter_time=1.921e-04, forward_time=0.158, loss_ctc=10.199, loss_att=4.401, acc=0.973, loss=6.140, backward_time=0.255, grad_norm=74.668, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.794e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 12:31:45,304 (trainer:779) INFO: 53epoch:train:3718-4130batch: iter_time=1.984e-04, forward_time=0.158, loss_ctc=10.235, loss_att=4.439, acc=0.972, loss=6.177, backward_time=0.255, grad_norm=70.452, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.790e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 12:34:08,068 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 12:35:51,738 (trainer:779) INFO: 53epoch:train:4131-4543batch: iter_time=2.031e-04, forward_time=0.160, loss_ctc=10.058, loss_att=4.396, acc=0.973, loss=6.095, backward_time=0.257, grad_norm=75.613, clip=100.000, loss_scale=8.206e+33, optim_step_time=0.031, optim0_lr0=6.787e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 12:39:57,491 (trainer:779) INFO: 53epoch:train:4544-4956batch: iter_time=2.102e-04, forward_time=0.159, loss_ctc=10.243, loss_att=4.419, acc=0.974, loss=6.166, backward_time=0.255, grad_norm=79.455, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.784e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 12:44:02,890 (trainer:779) INFO: 53epoch:train:4957-5369batch: iter_time=2.064e-04, forward_time=0.159, loss_ctc=10.135, loss_att=4.391, acc=0.973, loss=6.114, backward_time=0.256, grad_norm=82.485, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.781e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 12:48:08,530 (trainer:779) INFO: 53epoch:train:5370-5782batch: iter_time=2.067e-04, forward_time=0.158, loss_ctc=10.127, loss_att=4.388, acc=0.970, loss=6.109, backward_time=0.255, grad_norm=76.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.778e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 12:52:14,723 (trainer:779) INFO: 53epoch:train:5783-6195batch: iter_time=1.986e-04, forward_time=0.159, loss_ctc=10.205, loss_att=4.406, acc=0.973, loss=6.146, backward_time=0.256, grad_norm=72.557, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.774e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 12:56:22,422 (trainer:779) INFO: 53epoch:train:6196-6608batch: iter_time=2.109e-04, forward_time=0.160, loss_ctc=10.338, loss_att=4.493, acc=0.974, loss=6.246, backward_time=0.257, grad_norm=81.180, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.771e-04, train_time=1.199
+[bmi2:0/4] 2024-07-09 13:00:27,424 (trainer:779) INFO: 53epoch:train:6609-7021batch: iter_time=2.174e-04, forward_time=0.156, loss_ctc=10.164, loss_att=4.399, acc=0.975, loss=6.128, backward_time=0.256, grad_norm=74.443, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.768e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 13:04:30,968 (trainer:779) INFO: 53epoch:train:7022-7434batch: iter_time=2.138e-04, forward_time=0.156, loss_ctc=10.295, loss_att=4.434, acc=0.973, loss=6.192, backward_time=0.254, grad_norm=71.239, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.765e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 13:08:37,697 (trainer:779) INFO: 53epoch:train:7435-7847batch: iter_time=2.014e-04, forward_time=0.160, loss_ctc=10.202, loss_att=4.410, acc=0.974, loss=6.148, backward_time=0.256, grad_norm=79.750, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.762e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 13:12:41,912 (trainer:779) INFO: 53epoch:train:7848-8260batch: iter_time=2.213e-04, forward_time=0.157, loss_ctc=9.966, loss_att=4.332, acc=0.969, loss=6.022, backward_time=0.255, grad_norm=77.076, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.758e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 13:13:41,705 (trainer:365) INFO: 53epoch results: [train] iter_time=2.550e-04, forward_time=0.158, loss_ctc=10.159, loss_att=4.398, acc=0.973, loss=6.126, backward_time=0.255, grad_norm=75.297, clip=100.000, loss_scale=7.160e+33, optim_step_time=0.031, optim0_lr0=6.789e-04, train_time=1.189, time=1 hour, 21 minutes and 56.65 seconds, total_count=438151, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.783, cer_ctc=0.042, loss_att=5.878, acc=0.950, cer=0.033, wer=0.486, loss=7.049, time=17.94 seconds, total_count=1802, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.13 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 13:13:46,209 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 13:13:46,267 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/29epoch.pth
+[bmi2:0/4] 2024-07-09 13:13:46,268 (trainer:299) INFO: 54/100epoch started. Estimated time to finish: 2 days, 17 hours and 12 minutes
+[bmi2:0/4] 2024-07-09 13:17:59,949 (trainer:779) INFO: 54epoch:train:1-413batch: iter_time=9.818e-04, forward_time=0.156, loss_ctc=10.068, loss_att=4.359, acc=0.974, loss=6.072, backward_time=0.254, grad_norm=75.813, clip=100.000, loss_scale=9.074e+33, optim_step_time=0.031, optim0_lr0=6.755e-04, train_time=1.229
+[bmi2:0/4] 2024-07-09 13:22:03,634 (trainer:779) INFO: 54epoch:train:414-826batch: iter_time=2.056e-04, forward_time=0.157, loss_ctc=9.975, loss_att=4.313, acc=0.972, loss=6.012, backward_time=0.254, grad_norm=71.737, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.752e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 13:26:08,748 (trainer:779) INFO: 54epoch:train:827-1239batch: iter_time=2.203e-04, forward_time=0.158, loss_ctc=10.008, loss_att=4.329, acc=0.973, loss=6.033, backward_time=0.256, grad_norm=74.916, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.749e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 13:30:12,392 (trainer:779) INFO: 54epoch:train:1240-1652batch: iter_time=2.292e-04, forward_time=0.157, loss_ctc=9.987, loss_att=4.319, acc=0.974, loss=6.020, backward_time=0.254, grad_norm=79.362, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.746e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 13:34:17,857 (trainer:779) INFO: 54epoch:train:1653-2065batch: iter_time=2.089e-04, forward_time=0.159, loss_ctc=10.169, loss_att=4.427, acc=0.974, loss=6.150, backward_time=0.256, grad_norm=77.693, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.742e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 13:38:25,579 (trainer:779) INFO: 54epoch:train:2066-2478batch: iter_time=2.119e-04, forward_time=0.160, loss_ctc=10.059, loss_att=4.347, acc=0.978, loss=6.061, backward_time=0.258, grad_norm=71.677, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.739e-04, train_time=1.199
+[bmi2:0/4] 2024-07-09 13:42:31,133 (trainer:779) INFO: 54epoch:train:2479-2891batch: iter_time=2.033e-04, forward_time=0.159, loss_ctc=10.141, loss_att=4.398, acc=0.972, loss=6.121, backward_time=0.256, grad_norm=72.936, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.736e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 13:46:38,274 (trainer:779) INFO: 54epoch:train:2892-3304batch: iter_time=1.976e-04, forward_time=0.160, loss_ctc=9.990, loss_att=4.320, acc=0.972, loss=6.021, backward_time=0.256, grad_norm=74.429, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.733e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 13:50:40,488 (trainer:779) INFO: 54epoch:train:3305-3717batch: iter_time=2.376e-04, forward_time=0.155, loss_ctc=10.099, loss_att=4.366, acc=0.972, loss=6.086, backward_time=0.254, grad_norm=70.386, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.730e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 13:54:43,663 (trainer:779) INFO: 54epoch:train:3718-4130batch: iter_time=2.365e-04, forward_time=0.156, loss_ctc=10.171, loss_att=4.365, acc=0.973, loss=6.107, backward_time=0.254, grad_norm=78.210, clip=100.000, loss_scale=1.104e+34, optim_step_time=0.030, optim0_lr0=6.727e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 13:56:39,001 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 13:58:48,503 (trainer:779) INFO: 54epoch:train:4131-4543batch: iter_time=1.947e-04, forward_time=0.158, loss_ctc=10.209, loss_att=4.389, acc=0.975, loss=6.135, backward_time=0.255, grad_norm=78.832, clip=100.000, loss_scale=1.525e+34, optim_step_time=0.029, optim0_lr0=6.724e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 14:02:53,862 (trainer:779) INFO: 54epoch:train:4544-4956batch: iter_time=2.013e-04, forward_time=0.158, loss_ctc=10.189, loss_att=4.397, acc=0.976, loss=6.135, backward_time=0.255, grad_norm=82.940, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.029, optim0_lr0=6.720e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 14:06:58,261 (trainer:779) INFO: 54epoch:train:4957-5369batch: iter_time=2.181e-04, forward_time=0.159, loss_ctc=10.212, loss_att=4.389, acc=0.974, loss=6.136, backward_time=0.255, grad_norm=80.284, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.030, optim0_lr0=6.717e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 14:08:24,919 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 14:11:01,725 (trainer:779) INFO: 54epoch:train:5370-5782batch: iter_time=2.503e-04, forward_time=0.156, loss_ctc=10.169, loss_att=4.365, acc=0.972, loss=6.107, backward_time=0.253, grad_norm=82.300, clip=100.000, loss_scale=7.032e+33, optim_step_time=0.031, optim0_lr0=6.714e-04, train_time=1.178
+[bmi2:0/4] 2024-07-09 14:15:08,051 (trainer:779) INFO: 54epoch:train:5783-6195batch: iter_time=2.283e-04, forward_time=0.158, loss_ctc=10.074, loss_att=4.352, acc=0.974, loss=6.068, backward_time=0.256, grad_norm=82.547, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.711e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 14:19:13,906 (trainer:779) INFO: 54epoch:train:6196-6608batch: iter_time=2.050e-04, forward_time=0.158, loss_ctc=10.252, loss_att=4.397, acc=0.973, loss=6.153, backward_time=0.255, grad_norm=77.659, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.708e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 14:23:21,012 (trainer:779) INFO: 54epoch:train:6609-7021batch: iter_time=2.161e-04, forward_time=0.160, loss_ctc=10.083, loss_att=4.338, acc=0.973, loss=6.062, backward_time=0.257, grad_norm=77.685, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.705e-04, train_time=1.197
+[bmi2:0/4] 2024-07-09 14:27:27,368 (trainer:779) INFO: 54epoch:train:7022-7434batch: iter_time=2.054e-04, forward_time=0.160, loss_ctc=10.073, loss_att=4.376, acc=0.970, loss=6.085, backward_time=0.256, grad_norm=76.978, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.702e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 14:31:34,029 (trainer:779) INFO: 54epoch:train:7435-7847batch: iter_time=2.159e-04, forward_time=0.160, loss_ctc=10.094, loss_att=4.368, acc=0.972, loss=6.086, backward_time=0.256, grad_norm=78.542, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.699e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 14:32:39,143 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 14:35:40,414 (trainer:779) INFO: 54epoch:train:7848-8260batch: iter_time=2.147e-04, forward_time=0.159, loss_ctc=10.124, loss_att=4.395, acc=0.973, loss=6.114, backward_time=0.257, grad_norm=79.499, clip=100.000, loss_scale=3.277e+33, optim_step_time=0.031, optim0_lr0=6.696e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 14:36:37,789 (trainer:365) INFO: 54epoch results: [train] iter_time=2.541e-04, forward_time=0.158, loss_ctc=10.107, loss_att=4.365, acc=0.973, loss=6.088, backward_time=0.255, grad_norm=77.215, clip=100.000, loss_scale=8.768e+33, optim_step_time=0.031, optim0_lr0=6.725e-04, train_time=1.190, time=1 hour, 21 minutes and 59.18 seconds, total_count=446418, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.696, cer_ctc=0.041, loss_att=5.783, acc=0.950, cer=0.032, wer=0.488, loss=6.957, time=17.44 seconds, total_count=1836, gpu_max_cached_mem_GB=22.529, [att_plot] time=34.9 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 14:36:42,550 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 14:36:42,603 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/47epoch.pth
+[bmi2:0/4] 2024-07-09 14:36:42,603 (trainer:299) INFO: 55/100epoch started. Estimated time to finish: 2 days, 15 hours and 48 minutes
+[bmi2:0/4] 2024-07-09 14:40:56,078 (trainer:779) INFO: 55epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=10.277, loss_att=4.416, acc=0.974, loss=6.174, backward_time=0.256, grad_norm=80.222, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.692e-04, train_time=1.228
+[bmi2:0/4] 2024-07-09 14:45:01,208 (trainer:779) INFO: 55epoch:train:414-826batch: iter_time=2.111e-04, forward_time=0.157, loss_ctc=10.113, loss_att=4.360, acc=0.969, loss=6.086, backward_time=0.256, grad_norm=78.139, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.689e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 14:49:06,792 (trainer:779) INFO: 55epoch:train:827-1239batch: iter_time=1.981e-04, forward_time=0.158, loss_ctc=10.073, loss_att=4.331, acc=0.974, loss=6.054, backward_time=0.255, grad_norm=80.318, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.686e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 14:53:12,545 (trainer:779) INFO: 55epoch:train:1240-1652batch: iter_time=1.995e-04, forward_time=0.159, loss_ctc=9.995, loss_att=4.328, acc=0.975, loss=6.028, backward_time=0.256, grad_norm=73.983, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.683e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 14:57:18,509 (trainer:779) INFO: 55epoch:train:1653-2065batch: iter_time=1.961e-04, forward_time=0.159, loss_ctc=10.072, loss_att=4.329, acc=0.974, loss=6.052, backward_time=0.256, grad_norm=76.210, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.680e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 15:01:24,594 (trainer:779) INFO: 55epoch:train:2066-2478batch: iter_time=1.937e-04, forward_time=0.159, loss_ctc=10.073, loss_att=4.350, acc=0.974, loss=6.067, backward_time=0.256, grad_norm=79.118, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.677e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 15:05:31,095 (trainer:779) INFO: 55epoch:train:2479-2891batch: iter_time=2.014e-04, forward_time=0.159, loss_ctc=9.968, loss_att=4.293, acc=0.973, loss=5.995, backward_time=0.257, grad_norm=72.999, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.674e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 15:09:36,905 (trainer:779) INFO: 55epoch:train:2892-3304batch: iter_time=2.025e-04, forward_time=0.159, loss_ctc=10.062, loss_att=4.331, acc=0.975, loss=6.050, backward_time=0.256, grad_norm=71.070, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.671e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 15:13:42,350 (trainer:779) INFO: 55epoch:train:3305-3717batch: iter_time=2.011e-04, forward_time=0.158, loss_ctc=10.102, loss_att=4.343, acc=0.975, loss=6.071, backward_time=0.256, grad_norm=77.588, clip=100.000, loss_scale=2.760e+33, optim_step_time=0.031, optim0_lr0=6.668e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 15:17:47,020 (trainer:779) INFO: 55epoch:train:3718-4130batch: iter_time=1.986e-04, forward_time=0.157, loss_ctc=10.104, loss_att=4.364, acc=0.973, loss=6.086, backward_time=0.254, grad_norm=76.437, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.665e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 15:21:51,489 (trainer:779) INFO: 55epoch:train:4131-4543batch: iter_time=2.000e-04, forward_time=0.157, loss_ctc=10.021, loss_att=4.328, acc=0.972, loss=6.036, backward_time=0.255, grad_norm=78.285, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.662e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 15:25:58,257 (trainer:779) INFO: 55epoch:train:4544-4956batch: iter_time=1.933e-04, forward_time=0.159, loss_ctc=10.147, loss_att=4.370, acc=0.970, loss=6.103, backward_time=0.256, grad_norm=77.278, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.659e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 15:30:03,023 (trainer:779) INFO: 55epoch:train:4957-5369batch: iter_time=2.070e-04, forward_time=0.157, loss_ctc=10.113, loss_att=4.347, acc=0.973, loss=6.077, backward_time=0.254, grad_norm=74.975, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.656e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 15:34:06,628 (trainer:779) INFO: 55epoch:train:5370-5782batch: iter_time=1.969e-04, forward_time=0.156, loss_ctc=10.076, loss_att=4.333, acc=0.972, loss=6.056, backward_time=0.254, grad_norm=79.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.653e-04, train_time=1.179
+[bmi2:0/4] 2024-07-09 15:38:12,008 (trainer:779) INFO: 55epoch:train:5783-6195batch: iter_time=2.102e-04, forward_time=0.157, loss_ctc=10.238, loss_att=4.389, acc=0.977, loss=6.144, backward_time=0.256, grad_norm=78.408, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.649e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 15:42:17,885 (trainer:779) INFO: 55epoch:train:6196-6608batch: iter_time=1.993e-04, forward_time=0.159, loss_ctc=10.151, loss_att=4.357, acc=0.973, loss=6.095, backward_time=0.256, grad_norm=79.968, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.646e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 15:42:30,954 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 15:46:24,872 (trainer:779) INFO: 55epoch:train:6609-7021batch: iter_time=1.990e-04, forward_time=0.160, loss_ctc=10.064, loss_att=4.394, acc=0.973, loss=6.095, backward_time=0.257, grad_norm=83.349, clip=100.000, loss_scale=2.723e+33, optim_step_time=0.031, optim0_lr0=6.643e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 15:50:30,776 (trainer:779) INFO: 55epoch:train:7022-7434batch: iter_time=2.003e-04, forward_time=0.159, loss_ctc=10.125, loss_att=4.387, acc=0.973, loss=6.109, backward_time=0.256, grad_norm=74.891, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.640e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 15:54:35,253 (trainer:779) INFO: 55epoch:train:7435-7847batch: iter_time=1.997e-04, forward_time=0.157, loss_ctc=10.356, loss_att=4.440, acc=0.972, loss=6.214, backward_time=0.255, grad_norm=79.288, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.637e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 15:58:42,390 (trainer:779) INFO: 55epoch:train:7848-8260batch: iter_time=2.082e-04, forward_time=0.161, loss_ctc=10.139, loss_att=4.336, acc=0.974, loss=6.077, backward_time=0.257, grad_norm=73.071, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.634e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 15:59:42,140 (trainer:365) INFO: 55epoch results: [train] iter_time=2.514e-04, forward_time=0.158, loss_ctc=10.113, loss_att=4.356, acc=0.973, loss=6.084, backward_time=0.256, grad_norm=77.256, clip=100.000, loss_scale=3.519e+33, optim_step_time=0.031, optim0_lr0=6.663e-04, train_time=1.191, time=1 hour, 22 minutes and 4.6 seconds, total_count=454685, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.704, cer_ctc=0.041, loss_att=5.790, acc=0.949, cer=0.032, wer=0.481, loss=6.964, time=18 seconds, total_count=1870, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.93 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 15:59:46,772 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 15:59:46,825 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/36epoch.pth
+[bmi2:0/4] 2024-07-09 15:59:46,825 (trainer:299) INFO: 56/100epoch started. Estimated time to finish: 2 days, 14 hours and 25 minutes
+[bmi2:0/4] 2024-07-09 16:04:01,133 (trainer:779) INFO: 56epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=10.052, loss_att=4.318, acc=0.976, loss=6.038, backward_time=0.255, grad_norm=77.966, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.631e-04, train_time=1.232
+[bmi2:0/4] 2024-07-09 16:08:05,797 (trainer:779) INFO: 56epoch:train:414-826batch: iter_time=2.158e-04, forward_time=0.158, loss_ctc=9.985, loss_att=4.289, acc=0.974, loss=5.998, backward_time=0.255, grad_norm=75.960, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.628e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 16:12:12,487 (trainer:779) INFO: 56epoch:train:827-1239batch: iter_time=2.008e-04, forward_time=0.159, loss_ctc=9.934, loss_att=4.295, acc=0.975, loss=5.987, backward_time=0.257, grad_norm=75.575, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.625e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 16:16:18,286 (trainer:779) INFO: 56epoch:train:1240-1652batch: iter_time=1.994e-04, forward_time=0.159, loss_ctc=9.835, loss_att=4.278, acc=0.970, loss=5.945, backward_time=0.256, grad_norm=74.677, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.622e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 16:20:23,276 (trainer:779) INFO: 56epoch:train:1653-2065batch: iter_time=2.057e-04, forward_time=0.158, loss_ctc=9.887, loss_att=4.256, acc=0.972, loss=5.945, backward_time=0.255, grad_norm=76.313, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.619e-04, train_time=1.187
+[bmi2:0/4] 2024-07-09 16:24:27,237 (trainer:779) INFO: 56epoch:train:2066-2478batch: iter_time=1.945e-04, forward_time=0.157, loss_ctc=9.994, loss_att=4.317, acc=0.975, loss=6.020, backward_time=0.254, grad_norm=76.657, clip=100.000, loss_scale=3.311e+33, optim_step_time=0.031, optim0_lr0=6.616e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 16:28:30,875 (trainer:779) INFO: 56epoch:train:2479-2891batch: iter_time=2.061e-04, forward_time=0.156, loss_ctc=9.934, loss_att=4.309, acc=0.970, loss=5.997, backward_time=0.255, grad_norm=73.062, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.613e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 16:32:37,848 (trainer:779) INFO: 56epoch:train:2892-3304batch: iter_time=2.021e-04, forward_time=0.160, loss_ctc=9.877, loss_att=4.286, acc=0.975, loss=5.963, backward_time=0.257, grad_norm=71.860, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.610e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 16:36:43,159 (trainer:779) INFO: 56epoch:train:3305-3717batch: iter_time=1.985e-04, forward_time=0.158, loss_ctc=9.833, loss_att=4.276, acc=0.973, loss=5.943, backward_time=0.256, grad_norm=70.598, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.607e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 16:37:08,986 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 16:40:50,102 (trainer:779) INFO: 56epoch:train:3718-4130batch: iter_time=2.085e-04, forward_time=0.160, loss_ctc=9.859, loss_att=4.272, acc=0.974, loss=5.948, backward_time=0.257, grad_norm=73.313, clip=100.000, loss_scale=2.861e+33, optim_step_time=0.031, optim0_lr0=6.604e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 16:44:56,289 (trainer:779) INFO: 56epoch:train:4131-4543batch: iter_time=1.984e-04, forward_time=0.159, loss_ctc=10.027, loss_att=4.333, acc=0.974, loss=6.041, backward_time=0.257, grad_norm=74.883, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.601e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 16:49:02,808 (trainer:779) INFO: 56epoch:train:4544-4956batch: iter_time=2.003e-04, forward_time=0.159, loss_ctc=10.024, loss_att=4.310, acc=0.974, loss=6.024, backward_time=0.256, grad_norm=74.784, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.598e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 16:53:08,972 (trainer:779) INFO: 56epoch:train:4957-5369batch: iter_time=2.068e-04, forward_time=0.159, loss_ctc=10.068, loss_att=4.353, acc=0.975, loss=6.067, backward_time=0.256, grad_norm=73.576, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.595e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 16:57:15,562 (trainer:779) INFO: 56epoch:train:5370-5782batch: iter_time=2.001e-04, forward_time=0.159, loss_ctc=10.134, loss_att=4.354, acc=0.976, loss=6.088, backward_time=0.256, grad_norm=74.187, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.593e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 17:01:21,998 (trainer:779) INFO: 56epoch:train:5783-6195batch: iter_time=2.006e-04, forward_time=0.160, loss_ctc=10.120, loss_att=4.344, acc=0.975, loss=6.077, backward_time=0.256, grad_norm=76.883, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.590e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 17:05:28,304 (trainer:779) INFO: 56epoch:train:6196-6608batch: iter_time=2.028e-04, forward_time=0.159, loss_ctc=10.172, loss_att=4.339, acc=0.973, loss=6.089, backward_time=0.256, grad_norm=79.940, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.587e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 17:09:34,048 (trainer:779) INFO: 56epoch:train:6609-7021batch: iter_time=2.088e-04, forward_time=0.159, loss_ctc=10.164, loss_att=4.352, acc=0.973, loss=6.095, backward_time=0.255, grad_norm=76.185, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.584e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 17:13:40,269 (trainer:779) INFO: 56epoch:train:7022-7434batch: iter_time=2.014e-04, forward_time=0.159, loss_ctc=9.897, loss_att=4.266, acc=0.973, loss=5.955, backward_time=0.256, grad_norm=75.153, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.581e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 17:17:45,713 (trainer:779) INFO: 56epoch:train:7435-7847batch: iter_time=2.065e-04, forward_time=0.158, loss_ctc=10.030, loss_att=4.323, acc=0.972, loss=6.035, backward_time=0.256, grad_norm=78.068, clip=100.000, loss_scale=3.138e+33, optim_step_time=0.031, optim0_lr0=6.578e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 17:21:51,791 (trainer:779) INFO: 56epoch:train:7848-8260batch: iter_time=2.031e-04, forward_time=0.160, loss_ctc=10.304, loss_att=4.420, acc=0.973, loss=6.186, backward_time=0.256, grad_norm=79.558, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.575e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 17:22:49,296 (trainer:365) INFO: 56epoch results: [train] iter_time=2.504e-04, forward_time=0.159, loss_ctc=10.006, loss_att=4.314, acc=0.974, loss=6.022, backward_time=0.256, grad_norm=75.457, clip=100.000, loss_scale=3.193e+33, optim_step_time=0.031, optim0_lr0=6.603e-04, train_time=1.192, time=1 hour, 22 minutes and 10.07 seconds, total_count=462952, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.712, cer_ctc=0.041, loss_att=6.567, acc=0.937, cer=0.032, wer=0.483, loss=7.510, time=17.66 seconds, total_count=1904, gpu_max_cached_mem_GB=22.529, [att_plot] time=34.74 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 17:22:54,032 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 17:22:54,033 (trainer:299) INFO: 57/100epoch started. Estimated time to finish: 2 days, 13 hours and 2 minutes
+[bmi2:0/4] 2024-07-09 17:27:08,594 (trainer:779) INFO: 57epoch:train:1-413batch: iter_time=0.001, forward_time=0.158, loss_ctc=9.863, loss_att=4.234, acc=0.975, loss=5.923, backward_time=0.256, grad_norm=77.584, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.572e-04, train_time=1.233
+[bmi2:0/4] 2024-07-09 17:31:15,014 (trainer:779) INFO: 57epoch:train:414-826batch: iter_time=2.160e-04, forward_time=0.160, loss_ctc=9.951, loss_att=4.289, acc=0.972, loss=5.987, backward_time=0.257, grad_norm=75.984, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.569e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 17:35:20,969 (trainer:779) INFO: 57epoch:train:827-1239batch: iter_time=2.050e-04, forward_time=0.159, loss_ctc=9.987, loss_att=4.283, acc=0.975, loss=5.994, backward_time=0.256, grad_norm=80.958, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.566e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 17:39:27,162 (trainer:779) INFO: 57epoch:train:1240-1652batch: iter_time=2.040e-04, forward_time=0.158, loss_ctc=9.990, loss_att=4.296, acc=0.973, loss=6.004, backward_time=0.256, grad_norm=77.777, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.563e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 17:43:33,006 (trainer:779) INFO: 57epoch:train:1653-2065batch: iter_time=2.084e-04, forward_time=0.159, loss_ctc=9.941, loss_att=4.266, acc=0.973, loss=5.968, backward_time=0.256, grad_norm=73.399, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.560e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 17:47:40,016 (trainer:779) INFO: 57epoch:train:2066-2478batch: iter_time=2.040e-04, forward_time=0.159, loss_ctc=10.141, loss_att=4.345, acc=0.975, loss=6.084, backward_time=0.257, grad_norm=78.953, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.557e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 17:51:45,277 (trainer:779) INFO: 57epoch:train:2479-2891batch: iter_time=1.949e-04, forward_time=0.159, loss_ctc=9.956, loss_att=4.244, acc=0.973, loss=5.957, backward_time=0.256, grad_norm=72.253, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.554e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 17:55:50,919 (trainer:779) INFO: 57epoch:train:2892-3304batch: iter_time=2.060e-04, forward_time=0.159, loss_ctc=9.885, loss_att=4.247, acc=0.975, loss=5.939, backward_time=0.256, grad_norm=79.725, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.551e-04, train_time=1.189
+[bmi2:0/4] 2024-07-09 17:59:57,377 (trainer:779) INFO: 57epoch:train:3305-3717batch: iter_time=2.014e-04, forward_time=0.160, loss_ctc=9.958, loss_att=4.269, acc=0.975, loss=5.976, backward_time=0.257, grad_norm=73.057, clip=100.000, loss_scale=7.990e+33, optim_step_time=0.032, optim0_lr0=6.549e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 18:04:03,573 (trainer:779) INFO: 57epoch:train:3718-4130batch: iter_time=2.100e-04, forward_time=0.159, loss_ctc=9.844, loss_att=4.242, acc=0.972, loss=5.922, backward_time=0.256, grad_norm=75.736, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.546e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 18:08:07,390 (trainer:779) INFO: 57epoch:train:4131-4543batch: iter_time=2.003e-04, forward_time=0.156, loss_ctc=10.063, loss_att=4.313, acc=0.972, loss=6.038, backward_time=0.255, grad_norm=77.514, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.543e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 18:09:02,329 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 18:12:13,252 (trainer:779) INFO: 57epoch:train:4544-4956batch: iter_time=2.155e-04, forward_time=0.158, loss_ctc=10.071, loss_att=4.332, acc=0.973, loss=6.054, backward_time=0.257, grad_norm=78.434, clip=100.000, loss_scale=6.327e+33, optim_step_time=0.031, optim0_lr0=6.540e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 18:16:18,557 (trainer:779) INFO: 57epoch:train:4957-5369batch: iter_time=2.085e-04, forward_time=0.159, loss_ctc=9.921, loss_att=4.245, acc=0.974, loss=5.948, backward_time=0.255, grad_norm=78.929, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.537e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 18:20:24,495 (trainer:779) INFO: 57epoch:train:5370-5782batch: iter_time=2.255e-04, forward_time=0.159, loss_ctc=10.040, loss_att=4.279, acc=0.973, loss=6.008, backward_time=0.256, grad_norm=74.667, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.534e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 18:24:30,362 (trainer:779) INFO: 57epoch:train:5783-6195batch: iter_time=2.026e-04, forward_time=0.159, loss_ctc=10.111, loss_att=4.332, acc=0.974, loss=6.066, backward_time=0.256, grad_norm=73.606, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.531e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 18:28:35,803 (trainer:779) INFO: 57epoch:train:6196-6608batch: iter_time=1.972e-04, forward_time=0.158, loss_ctc=10.077, loss_att=4.325, acc=0.974, loss=6.051, backward_time=0.255, grad_norm=74.466, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.528e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 18:32:42,814 (trainer:779) INFO: 57epoch:train:6609-7021batch: iter_time=2.011e-04, forward_time=0.161, loss_ctc=9.988, loss_att=4.307, acc=0.977, loss=6.011, backward_time=0.257, grad_norm=72.303, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.525e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 18:36:48,882 (trainer:779) INFO: 57epoch:train:7022-7434batch: iter_time=2.033e-04, forward_time=0.158, loss_ctc=10.017, loss_att=4.406, acc=0.976, loss=6.089, backward_time=0.256, grad_norm=97.260, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.523e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 18:40:55,043 (trainer:779) INFO: 57epoch:train:7435-7847batch: iter_time=2.163e-04, forward_time=0.159, loss_ctc=9.907, loss_att=4.293, acc=0.972, loss=5.977, backward_time=0.256, grad_norm=72.485, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.520e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 18:45:01,224 (trainer:779) INFO: 57epoch:train:7848-8260batch: iter_time=1.996e-04, forward_time=0.159, loss_ctc=9.933, loss_att=4.247, acc=0.972, loss=5.952, backward_time=0.256, grad_norm=69.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.517e-04, train_time=1.191
+[bmi2:0/4] 2024-07-09 18:46:02,009 (trainer:365) INFO: 57epoch results: [train] iter_time=2.530e-04, forward_time=0.159, loss_ctc=9.982, loss_att=4.290, acc=0.974, loss=5.997, backward_time=0.256, grad_norm=76.715, clip=100.000, loss_scale=5.907e+33, optim_step_time=0.031, optim0_lr0=6.544e-04, train_time=1.193, time=1 hour, 22 minutes and 12.33 seconds, total_count=471219, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.750, cer_ctc=0.041, loss_att=5.745, acc=0.949, cer=0.032, wer=0.490, loss=6.946, time=17.37 seconds, total_count=1938, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.28 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 18:46:07,432 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 18:46:07,502 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/56epoch.pth
+[bmi2:0/4] 2024-07-09 18:46:07,503 (trainer:299) INFO: 58/100epoch started. Estimated time to finish: 2 days, 11 hours and 38 minutes
+[bmi2:0/4] 2024-07-09 18:50:24,745 (trainer:779) INFO: 58epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.727, loss_att=4.182, acc=0.976, loss=5.846, backward_time=0.257, grad_norm=66.427, clip=100.000, loss_scale=5.747e+33, optim_step_time=0.032, optim0_lr0=6.514e-04, train_time=1.246
+[bmi2:0/4] 2024-07-09 18:54:31,423 (trainer:779) INFO: 58epoch:train:414-826batch: iter_time=2.057e-04, forward_time=0.159, loss_ctc=9.850, loss_att=4.238, acc=0.974, loss=5.922, backward_time=0.257, grad_norm=72.662, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.511e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 18:58:37,453 (trainer:779) INFO: 58epoch:train:827-1239batch: iter_time=1.979e-04, forward_time=0.160, loss_ctc=9.790, loss_att=4.205, acc=0.975, loss=5.881, backward_time=0.257, grad_norm=74.602, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.508e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 19:02:41,559 (trainer:779) INFO: 58epoch:train:1240-1652batch: iter_time=2.257e-04, forward_time=0.156, loss_ctc=9.729, loss_att=4.190, acc=0.975, loss=5.852, backward_time=0.256, grad_norm=72.443, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.505e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 19:03:14,760 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 19:06:46,370 (trainer:779) INFO: 58epoch:train:1653-2065batch: iter_time=2.001e-04, forward_time=0.158, loss_ctc=9.916, loss_att=4.241, acc=0.976, loss=5.943, backward_time=0.256, grad_norm=76.753, clip=100.000, loss_scale=5.876e+33, optim_step_time=0.031, optim0_lr0=6.503e-04, train_time=1.186
+[bmi2:0/4] 2024-07-09 19:10:51,874 (trainer:779) INFO: 58epoch:train:2066-2478batch: iter_time=2.122e-04, forward_time=0.158, loss_ctc=9.772, loss_att=4.201, acc=0.973, loss=5.873, backward_time=0.256, grad_norm=74.037, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.500e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 19:14:57,049 (trainer:779) INFO: 58epoch:train:2479-2891batch: iter_time=1.931e-04, forward_time=0.159, loss_ctc=9.776, loss_att=4.208, acc=0.975, loss=5.878, backward_time=0.256, grad_norm=70.352, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.497e-04, train_time=1.188
+[bmi2:0/4] 2024-07-09 19:19:03,603 (trainer:779) INFO: 58epoch:train:2892-3304batch: iter_time=2.016e-04, forward_time=0.159, loss_ctc=9.896, loss_att=4.248, acc=0.975, loss=5.942, backward_time=0.256, grad_norm=72.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.494e-04, train_time=1.193
+[bmi2:0/4] 2024-07-09 19:23:10,653 (trainer:779) INFO: 58epoch:train:3305-3717batch: iter_time=2.011e-04, forward_time=0.160, loss_ctc=9.711, loss_att=4.182, acc=0.976, loss=5.841, backward_time=0.257, grad_norm=73.910, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.491e-04, train_time=1.197
+[bmi2:0/4] 2024-07-09 19:27:17,451 (trainer:779) INFO: 58epoch:train:3718-4130batch: iter_time=2.194e-04, forward_time=0.159, loss_ctc=9.747, loss_att=4.209, acc=0.976, loss=5.871, backward_time=0.257, grad_norm=73.229, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.488e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 19:31:26,258 (trainer:779) INFO: 58epoch:train:4131-4543batch: iter_time=2.465e-04, forward_time=0.158, loss_ctc=9.815, loss_att=4.231, acc=0.974, loss=5.906, backward_time=0.260, grad_norm=75.316, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.486e-04, train_time=1.205
+[bmi2:0/4] 2024-07-09 19:35:35,062 (trainer:779) INFO: 58epoch:train:4544-4956batch: iter_time=2.365e-04, forward_time=0.158, loss_ctc=9.690, loss_att=4.192, acc=0.970, loss=5.841, backward_time=0.259, grad_norm=73.634, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.483e-04, train_time=1.204
+[bmi2:0/4] 2024-07-09 19:39:42,618 (trainer:779) INFO: 58epoch:train:4957-5369batch: iter_time=2.475e-04, forward_time=0.158, loss_ctc=9.908, loss_att=4.260, acc=0.974, loss=5.955, backward_time=0.259, grad_norm=70.117, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.480e-04, train_time=1.199
+[bmi2:0/4] 2024-07-09 19:43:50,636 (trainer:779) INFO: 58epoch:train:5370-5782batch: iter_time=2.391e-04, forward_time=0.157, loss_ctc=9.927, loss_att=4.268, acc=0.976, loss=5.966, backward_time=0.259, grad_norm=79.879, clip=100.000, loss_scale=6.120e+33, optim_step_time=0.033, optim0_lr0=6.477e-04, train_time=1.200
+[bmi2:0/4] 2024-07-09 19:44:13,724 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 19:48:00,160 (trainer:779) INFO: 58epoch:train:5783-6195batch: iter_time=2.386e-04, forward_time=0.159, loss_ctc=9.830, loss_att=4.246, acc=0.975, loss=5.921, backward_time=0.260, grad_norm=77.431, clip=100.000, loss_scale=5.648e+33, optim_step_time=0.034, optim0_lr0=6.474e-04, train_time=1.209
+[bmi2:0/4] 2024-07-09 19:52:09,380 (trainer:779) INFO: 58epoch:train:6196-6608batch: iter_time=2.419e-04, forward_time=0.159, loss_ctc=9.803, loss_att=4.222, acc=0.974, loss=5.896, backward_time=0.259, grad_norm=77.865, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.472e-04, train_time=1.206
+[bmi2:0/4] 2024-07-09 19:56:15,997 (trainer:779) INFO: 58epoch:train:6609-7021batch: iter_time=2.310e-04, forward_time=0.157, loss_ctc=9.847, loss_att=4.234, acc=0.970, loss=5.918, backward_time=0.257, grad_norm=71.713, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.469e-04, train_time=1.195
+[bmi2:0/4] 2024-07-09 20:00:26,961 (trainer:779) INFO: 58epoch:train:7022-7434batch: iter_time=2.474e-04, forward_time=0.160, loss_ctc=9.811, loss_att=4.225, acc=0.974, loss=5.900, backward_time=0.262, grad_norm=73.688, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.466e-04, train_time=1.214
+[bmi2:0/4] 2024-07-09 20:04:35,962 (trainer:779) INFO: 58epoch:train:7435-7847batch: iter_time=2.326e-04, forward_time=0.159, loss_ctc=9.822, loss_att=4.253, acc=0.974, loss=5.924, backward_time=0.259, grad_norm=71.366, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.463e-04, train_time=1.206
+[bmi2:0/4] 2024-07-09 20:07:24,988 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 20:08:45,556 (trainer:779) INFO: 58epoch:train:7848-8260batch: iter_time=2.398e-04, forward_time=0.160, loss_ctc=9.804, loss_att=4.234, acc=0.971, loss=5.905, backward_time=0.260, grad_norm=76.003, clip=100.000, loss_scale=4.348e+33, optim_step_time=0.034, optim0_lr0=6.460e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 20:09:52,346 (trainer:365) INFO: 58epoch results: [train] iter_time=2.715e-04, forward_time=0.159, loss_ctc=9.809, loss_att=4.224, acc=0.974, loss=5.899, backward_time=0.258, grad_norm=73.732, clip=100.000, loss_scale=6.059e+33, optim_step_time=0.033, optim0_lr0=6.487e-04, train_time=1.200, time=1 hour, 22 minutes and 43.36 seconds, total_count=479486, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.747, cer_ctc=0.040, loss_att=5.903, acc=0.949, cer=0.032, wer=0.484, loss=7.056, time=20.57 seconds, total_count=1972, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.91 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 20:09:57,599 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 20:09:57,709 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/41epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/57epoch.pth
+[bmi2:0/4] 2024-07-09 20:09:57,709 (trainer:299) INFO: 59/100epoch started. Estimated time to finish: 2 days, 10 hours and 16 minutes
+[bmi2:0/4] 2024-07-09 20:14:17,989 (trainer:779) INFO: 59epoch:train:1-413batch: iter_time=0.001, forward_time=0.158, loss_ctc=9.750, loss_att=4.178, acc=0.975, loss=5.850, backward_time=0.260, grad_norm=71.991, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.458e-04, train_time=1.261
+[bmi2:0/4] 2024-07-09 20:18:26,600 (trainer:779) INFO: 59epoch:train:414-826batch: iter_time=2.433e-04, forward_time=0.158, loss_ctc=9.678, loss_att=4.152, acc=0.976, loss=5.810, backward_time=0.259, grad_norm=76.122, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.455e-04, train_time=1.203
+[bmi2:0/4] 2024-07-09 20:22:35,339 (trainer:779) INFO: 59epoch:train:827-1239batch: iter_time=2.462e-04, forward_time=0.158, loss_ctc=9.780, loss_att=4.191, acc=0.976, loss=5.868, backward_time=0.260, grad_norm=70.627, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.452e-04, train_time=1.205
+[bmi2:0/4] 2024-07-09 20:26:43,829 (trainer:779) INFO: 59epoch:train:1240-1652batch: iter_time=2.411e-04, forward_time=0.158, loss_ctc=9.771, loss_att=4.203, acc=0.975, loss=5.873, backward_time=0.259, grad_norm=74.102, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.449e-04, train_time=1.202
+[bmi2:0/4] 2024-07-09 20:30:53,350 (trainer:779) INFO: 59epoch:train:1653-2065batch: iter_time=2.413e-04, forward_time=0.159, loss_ctc=9.828, loss_att=4.222, acc=0.971, loss=5.904, backward_time=0.259, grad_norm=68.485, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.447e-04, train_time=1.209
+[bmi2:0/4] 2024-07-09 20:35:04,700 (trainer:779) INFO: 59epoch:train:2066-2478batch: iter_time=2.454e-04, forward_time=0.160, loss_ctc=9.677, loss_att=4.172, acc=0.978, loss=5.823, backward_time=0.262, grad_norm=72.079, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.444e-04, train_time=1.216
+[bmi2:0/4] 2024-07-09 20:39:15,767 (trainer:779) INFO: 59epoch:train:2479-2891batch: iter_time=2.393e-04, forward_time=0.160, loss_ctc=9.716, loss_att=4.178, acc=0.975, loss=5.839, backward_time=0.261, grad_norm=71.670, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.441e-04, train_time=1.216
+[bmi2:0/4] 2024-07-09 20:43:25,050 (trainer:779) INFO: 59epoch:train:2892-3304batch: iter_time=2.581e-04, forward_time=0.161, loss_ctc=9.651, loss_att=4.159, acc=0.975, loss=5.807, backward_time=0.260, grad_norm=73.006, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.438e-04, train_time=1.207
+[bmi2:0/4] 2024-07-09 20:47:35,058 (trainer:779) INFO: 59epoch:train:3305-3717batch: iter_time=2.571e-04, forward_time=0.161, loss_ctc=9.594, loss_att=4.120, acc=0.973, loss=5.763, backward_time=0.261, grad_norm=68.847, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.436e-04, train_time=1.211
+[bmi2:0/4] 2024-07-09 20:51:43,357 (trainer:779) INFO: 59epoch:train:3718-4130batch: iter_time=2.553e-04, forward_time=0.158, loss_ctc=9.916, loss_att=4.272, acc=0.973, loss=5.966, backward_time=0.259, grad_norm=76.374, clip=100.000, loss_scale=4.289e+33, optim_step_time=0.033, optim0_lr0=6.433e-04, train_time=1.202
+[bmi2:0/4] 2024-07-09 20:55:52,730 (trainer:779) INFO: 59epoch:train:4131-4543batch: iter_time=2.538e-04, forward_time=0.162, loss_ctc=9.760, loss_att=4.203, acc=0.973, loss=5.870, backward_time=0.261, grad_norm=76.436, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.430e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 21:00:03,355 (trainer:779) INFO: 59epoch:train:4544-4956batch: iter_time=2.161e-04, forward_time=0.161, loss_ctc=9.756, loss_att=4.212, acc=0.973, loss=5.875, backward_time=0.261, grad_norm=74.483, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.427e-04, train_time=1.213
+[bmi2:0/4] 2024-07-09 21:04:13,272 (trainer:779) INFO: 59epoch:train:4957-5369batch: iter_time=2.502e-04, forward_time=0.160, loss_ctc=9.859, loss_att=4.215, acc=0.976, loss=5.908, backward_time=0.262, grad_norm=77.133, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.425e-04, train_time=1.211
+[bmi2:0/4] 2024-07-09 21:08:24,920 (trainer:779) INFO: 59epoch:train:5370-5782batch: iter_time=2.576e-04, forward_time=0.162, loss_ctc=9.733, loss_att=4.186, acc=0.975, loss=5.850, backward_time=0.262, grad_norm=74.082, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.422e-04, train_time=1.218
+[bmi2:0/4] 2024-07-09 21:12:33,878 (trainer:779) INFO: 59epoch:train:5783-6195batch: iter_time=2.510e-04, forward_time=0.158, loss_ctc=9.893, loss_att=4.231, acc=0.976, loss=5.930, backward_time=0.258, grad_norm=75.851, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.419e-04, train_time=1.206
+[bmi2:0/4] 2024-07-09 21:16:43,711 (trainer:779) INFO: 59epoch:train:6196-6608batch: iter_time=2.378e-04, forward_time=0.160, loss_ctc=9.775, loss_att=4.186, acc=0.973, loss=5.863, backward_time=0.258, grad_norm=74.287, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.416e-04, train_time=1.209
+[bmi2:0/4] 2024-07-09 21:20:53,953 (trainer:779) INFO: 59epoch:train:6609-7021batch: iter_time=2.323e-04, forward_time=0.159, loss_ctc=9.850, loss_att=4.231, acc=0.974, loss=5.917, backward_time=0.257, grad_norm=72.018, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.414e-04, train_time=1.212
+[bmi2:0/4] 2024-07-09 21:25:05,058 (trainer:779) INFO: 59epoch:train:7022-7434batch: iter_time=2.501e-04, forward_time=0.160, loss_ctc=9.900, loss_att=4.226, acc=0.974, loss=5.928, backward_time=0.257, grad_norm=73.990, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.411e-04, train_time=1.215
+[bmi2:0/4] 2024-07-09 21:29:15,373 (trainer:779) INFO: 59epoch:train:7435-7847batch: iter_time=2.301e-04, forward_time=0.159, loss_ctc=9.734, loss_att=4.180, acc=0.974, loss=5.846, backward_time=0.257, grad_norm=73.661, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.408e-04, train_time=1.212
+[bmi2:0/4] 2024-07-09 21:33:22,091 (trainer:779) INFO: 59epoch:train:7848-8260batch: iter_time=2.269e-04, forward_time=0.160, loss_ctc=9.822, loss_att=4.216, acc=0.971, loss=5.897, backward_time=0.257, grad_norm=70.573, clip=100.000, loss_scale=1.021e+34, optim_step_time=0.032, optim0_lr0=6.405e-04, train_time=1.194
+[bmi2:0/4] 2024-07-09 21:34:23,640 (trainer:365) INFO: 59epoch results: [train] iter_time=2.956e-04, forward_time=0.160, loss_ctc=9.772, loss_att=4.197, acc=0.974, loss=5.869, backward_time=0.260, grad_norm=73.283, clip=100.000, loss_scale=4.235e+33, optim_step_time=0.033, optim0_lr0=6.431e-04, train_time=1.211, time=1 hour, 23 minutes and 29.76 seconds, total_count=487753, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.867, cer_ctc=0.041, loss_att=5.945, acc=0.948, cer=0.034, wer=0.491, loss=7.122, time=19.89 seconds, total_count=2006, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.28 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 21:34:28,223 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 21:34:28,224 (trainer:299) INFO: 60/100epoch started. Estimated time to finish: 2 days, 8 hours and 53 minutes
+[bmi2:0/4] 2024-07-09 21:38:48,242 (trainer:779) INFO: 60epoch:train:1-413batch: iter_time=0.001, forward_time=0.164, loss_ctc=9.597, loss_att=4.114, acc=0.977, loss=5.759, backward_time=0.258, grad_norm=65.364, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.034, optim0_lr0=6.403e-04, train_time=1.259
+[bmi2:0/4] 2024-07-09 21:41:37,966 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 21:42:57,758 (trainer:779) INFO: 60epoch:train:414-826batch: iter_time=2.340e-04, forward_time=0.163, loss_ctc=9.652, loss_att=4.145, acc=0.973, loss=5.797, backward_time=0.258, grad_norm=73.398, clip=100.000, loss_scale=8.721e+33, optim_step_time=0.032, optim0_lr0=6.400e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 21:47:07,132 (trainer:779) INFO: 60epoch:train:827-1239batch: iter_time=2.508e-04, forward_time=0.164, loss_ctc=9.656, loss_att=4.159, acc=0.973, loss=5.808, backward_time=0.258, grad_norm=70.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.397e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 21:51:15,508 (trainer:779) INFO: 60epoch:train:1240-1652batch: iter_time=2.692e-04, forward_time=0.164, loss_ctc=9.650, loss_att=4.150, acc=0.975, loss=5.800, backward_time=0.257, grad_norm=72.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.395e-04, train_time=1.202
+[bmi2:0/4] 2024-07-09 21:55:23,090 (trainer:779) INFO: 60epoch:train:1653-2065batch: iter_time=2.558e-04, forward_time=0.161, loss_ctc=9.701, loss_att=4.178, acc=0.972, loss=5.835, backward_time=0.257, grad_norm=72.738, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.392e-04, train_time=1.199
+[bmi2:0/4] 2024-07-09 21:59:32,427 (trainer:779) INFO: 60epoch:train:2066-2478batch: iter_time=2.819e-04, forward_time=0.161, loss_ctc=9.712, loss_att=4.194, acc=0.971, loss=5.849, backward_time=0.261, grad_norm=71.793, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.389e-04, train_time=1.207
+[bmi2:0/4] 2024-07-09 22:03:42,184 (trainer:779) INFO: 60epoch:train:2479-2891batch: iter_time=2.501e-04, forward_time=0.160, loss_ctc=9.683, loss_att=4.144, acc=0.976, loss=5.806, backward_time=0.261, grad_norm=71.680, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.387e-04, train_time=1.210
+[bmi2:0/4] 2024-07-09 22:07:51,321 (trainer:779) INFO: 60epoch:train:2892-3304batch: iter_time=2.525e-04, forward_time=0.159, loss_ctc=9.792, loss_att=4.209, acc=0.976, loss=5.884, backward_time=0.258, grad_norm=73.127, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.384e-04, train_time=1.205
+[bmi2:0/4] 2024-07-09 22:12:00,423 (trainer:779) INFO: 60epoch:train:3305-3717batch: iter_time=2.405e-04, forward_time=0.161, loss_ctc=9.651, loss_att=4.148, acc=0.974, loss=5.799, backward_time=0.259, grad_norm=72.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.381e-04, train_time=1.207
+[bmi2:0/4] 2024-07-09 22:16:10,097 (trainer:779) INFO: 60epoch:train:3718-4130batch: iter_time=2.674e-04, forward_time=0.160, loss_ctc=9.690, loss_att=4.159, acc=0.978, loss=5.819, backward_time=0.260, grad_norm=70.390, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.378e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 22:18:35,891 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 22:20:19,308 (trainer:779) INFO: 60epoch:train:4131-4543batch: iter_time=2.386e-04, forward_time=0.160, loss_ctc=9.629, loss_att=4.155, acc=0.974, loss=5.797, backward_time=0.260, grad_norm=72.215, clip=100.000, loss_scale=4.116e+33, optim_step_time=0.034, optim0_lr0=6.376e-04, train_time=1.207
+[bmi2:0/4] 2024-07-09 22:24:28,875 (trainer:779) INFO: 60epoch:train:4544-4956batch: iter_time=2.443e-04, forward_time=0.160, loss_ctc=9.732, loss_att=4.168, acc=0.974, loss=5.837, backward_time=0.260, grad_norm=69.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.373e-04, train_time=1.208
+[bmi2:0/4] 2024-07-09 22:28:37,938 (trainer:779) INFO: 60epoch:train:4957-5369batch: iter_time=2.361e-04, forward_time=0.160, loss_ctc=9.797, loss_att=4.176, acc=0.975, loss=5.862, backward_time=0.261, grad_norm=75.923, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.370e-04, train_time=1.207
+[bmi2:0/4] 2024-07-09 22:32:49,111 (trainer:779) INFO: 60epoch:train:5370-5782batch: iter_time=2.335e-04, forward_time=0.160, loss_ctc=9.649, loss_att=4.149, acc=0.977, loss=5.799, backward_time=0.264, grad_norm=75.921, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.368e-04, train_time=1.215
+[bmi2:0/4] 2024-07-09 22:37:01,327 (trainer:779) INFO: 60epoch:train:5783-6195batch: iter_time=2.388e-04, forward_time=0.160, loss_ctc=9.698, loss_att=4.168, acc=0.974, loss=5.827, backward_time=0.267, grad_norm=70.885, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.365e-04, train_time=1.222
+[bmi2:0/4] 2024-07-09 22:41:05,447 (trainer:779) INFO: 60epoch:train:6196-6608batch: iter_time=1.905e-04, forward_time=0.158, loss_ctc=9.739, loss_att=4.207, acc=0.976, loss=5.866, backward_time=0.256, grad_norm=79.496, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.362e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 22:45:09,942 (trainer:779) INFO: 60epoch:train:6609-7021batch: iter_time=2.051e-04, forward_time=0.158, loss_ctc=9.744, loss_att=4.185, acc=0.973, loss=5.853, backward_time=0.256, grad_norm=70.972, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.360e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 22:49:12,934 (trainer:779) INFO: 60epoch:train:7022-7434batch: iter_time=1.951e-04, forward_time=0.157, loss_ctc=9.785, loss_att=4.190, acc=0.975, loss=5.869, backward_time=0.254, grad_norm=76.612, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.357e-04, train_time=1.176
+[bmi2:0/4] 2024-07-09 22:53:15,297 (trainer:779) INFO: 60epoch:train:7435-7847batch: iter_time=1.957e-04, forward_time=0.156, loss_ctc=9.745, loss_att=4.205, acc=0.971, loss=5.867, backward_time=0.254, grad_norm=75.314, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=6.355e-04, train_time=1.174
+[bmi2:0/4] 2024-07-09 22:57:19,878 (trainer:779) INFO: 60epoch:train:7848-8260batch: iter_time=2.195e-04, forward_time=0.157, loss_ctc=9.726, loss_att=4.171, acc=0.975, loss=5.837, backward_time=0.255, grad_norm=72.133, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.352e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 22:58:23,530 (trainer:365) INFO: 60epoch results: [train] iter_time=2.911e-04, forward_time=0.160, loss_ctc=9.701, loss_att=4.169, acc=0.974, loss=5.829, backward_time=0.259, grad_norm=72.601, clip=100.000, loss_scale=4.404e+33, optim_step_time=0.033, optim0_lr0=6.377e-04, train_time=1.204, time=1 hour, 22 minutes and 56.58 seconds, total_count=496020, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.778, cer_ctc=0.041, loss_att=5.918, acc=0.949, cer=0.033, wer=0.490, loss=7.076, time=18.19 seconds, total_count=2040, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.54 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-09 22:58:28,893 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-09 22:58:28,929 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/59epoch.pth
+[bmi2:0/4] 2024-07-09 22:58:28,929 (trainer:299) INFO: 61/100epoch started. Estimated time to finish: 2 days, 7 hours and 30 minutes
+[bmi2:0/4] 2024-07-09 23:02:44,882 (trainer:779) INFO: 61epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.524, loss_att=4.071, acc=0.978, loss=5.707, backward_time=0.256, grad_norm=73.064, clip=100.000, loss_scale=4.524e+33, optim_step_time=0.031, optim0_lr0=6.349e-04, train_time=1.240
+[bmi2:0/4] 2024-07-09 23:06:52,071 (trainer:779) INFO: 61epoch:train:414-826batch: iter_time=1.964e-04, forward_time=0.161, loss_ctc=9.619, loss_att=4.100, acc=0.977, loss=5.756, backward_time=0.257, grad_norm=75.498, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.347e-04, train_time=1.196
+[bmi2:0/4] 2024-07-09 23:10:57,722 (trainer:779) INFO: 61epoch:train:827-1239batch: iter_time=2.028e-04, forward_time=0.160, loss_ctc=9.615, loss_att=4.124, acc=0.975, loss=5.771, backward_time=0.256, grad_norm=67.952, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.344e-04, train_time=1.190
+[bmi2:0/4] 2024-07-09 23:15:03,960 (trainer:779) INFO: 61epoch:train:1240-1652batch: iter_time=1.946e-04, forward_time=0.161, loss_ctc=9.551, loss_att=4.081, acc=0.975, loss=5.722, backward_time=0.256, grad_norm=72.865, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.341e-04, train_time=1.192
+[bmi2:0/4] 2024-07-09 23:19:07,829 (trainer:779) INFO: 61epoch:train:1653-2065batch: iter_time=1.959e-04, forward_time=0.158, loss_ctc=9.443, loss_att=4.090, acc=0.970, loss=5.696, backward_time=0.254, grad_norm=70.963, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.339e-04, train_time=1.181
+[bmi2:0/4] 2024-07-09 23:23:12,727 (trainer:779) INFO: 61epoch:train:2066-2478batch: iter_time=1.995e-04, forward_time=0.159, loss_ctc=9.589, loss_att=4.129, acc=0.977, loss=5.767, backward_time=0.255, grad_norm=75.683, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.336e-04, train_time=1.185
+[bmi2:0/4] 2024-07-09 23:27:16,346 (trainer:779) INFO: 61epoch:train:2479-2891batch: iter_time=2.159e-04, forward_time=0.158, loss_ctc=9.622, loss_att=4.121, acc=0.971, loss=5.772, backward_time=0.254, grad_norm=71.002, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.333e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 23:31:20,907 (trainer:779) INFO: 61epoch:train:2892-3304batch: iter_time=2.069e-04, forward_time=0.158, loss_ctc=9.829, loss_att=4.238, acc=0.976, loss=5.915, backward_time=0.255, grad_norm=79.507, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.331e-04, train_time=1.184
+[bmi2:0/4] 2024-07-09 23:35:25,029 (trainer:779) INFO: 61epoch:train:3305-3717batch: iter_time=2.209e-04, forward_time=0.157, loss_ctc=9.667, loss_att=4.135, acc=0.977, loss=5.795, backward_time=0.255, grad_norm=72.318, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.328e-04, train_time=1.182
+[bmi2:0/4] 2024-07-09 23:39:28,811 (trainer:779) INFO: 61epoch:train:3718-4130batch: iter_time=2.195e-04, forward_time=0.157, loss_ctc=9.576, loss_att=4.094, acc=0.974, loss=5.739, backward_time=0.255, grad_norm=72.476, clip=100.000, loss_scale=5.493e+33, optim_step_time=0.031, optim0_lr0=6.326e-04, train_time=1.180
+[bmi2:0/4] 2024-07-09 23:40:59,511 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-09 23:43:31,024 (trainer:779) INFO: 61epoch:train:4131-4543batch: iter_time=2.060e-04, forward_time=0.155, loss_ctc=9.605, loss_att=4.136, acc=0.974, loss=5.776, backward_time=0.253, grad_norm=72.928, clip=100.000, loss_scale=7.117e+33, optim_step_time=0.031, optim0_lr0=6.323e-04, train_time=1.173
+[bmi2:0/4] 2024-07-09 23:47:34,299 (trainer:779) INFO: 61epoch:train:4544-4956batch: iter_time=2.315e-04, forward_time=0.156, loss_ctc=9.687, loss_att=4.168, acc=0.972, loss=5.824, backward_time=0.254, grad_norm=74.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.320e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 23:51:36,666 (trainer:779) INFO: 61epoch:train:4957-5369batch: iter_time=2.362e-04, forward_time=0.156, loss_ctc=9.594, loss_att=4.146, acc=0.974, loss=5.780, backward_time=0.253, grad_norm=68.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.318e-04, train_time=1.174
+[bmi2:0/4] 2024-07-09 23:55:39,873 (trainer:779) INFO: 61epoch:train:5370-5782batch: iter_time=2.393e-04, forward_time=0.156, loss_ctc=9.567, loss_att=4.069, acc=0.977, loss=5.718, backward_time=0.253, grad_norm=79.174, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.315e-04, train_time=1.177
+[bmi2:0/4] 2024-07-09 23:59:43,489 (trainer:779) INFO: 61epoch:train:5783-6195batch: iter_time=2.309e-04, forward_time=0.157, loss_ctc=9.603, loss_att=4.111, acc=0.975, loss=5.759, backward_time=0.254, grad_norm=75.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.313e-04, train_time=1.180
+[bmi2:0/4] 2024-07-10 00:03:46,050 (trainer:779) INFO: 61epoch:train:6196-6608batch: iter_time=2.126e-04, forward_time=0.156, loss_ctc=9.803, loss_att=4.188, acc=0.977, loss=5.873, backward_time=0.254, grad_norm=71.882, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.310e-04, train_time=1.174
+[bmi2:0/4] 2024-07-10 00:07:49,540 (trainer:779) INFO: 61epoch:train:6609-7021batch: iter_time=1.969e-04, forward_time=0.158, loss_ctc=9.664, loss_att=4.155, acc=0.973, loss=5.808, backward_time=0.255, grad_norm=72.118, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.307e-04, train_time=1.180
+[bmi2:0/4] 2024-07-10 00:11:52,404 (trainer:779) INFO: 61epoch:train:7022-7434batch: iter_time=1.951e-04, forward_time=0.158, loss_ctc=9.718, loss_att=4.164, acc=0.975, loss=5.831, backward_time=0.254, grad_norm=71.580, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.305e-04, train_time=1.175
+[bmi2:0/4] 2024-07-10 00:15:54,329 (trainer:779) INFO: 61epoch:train:7435-7847batch: iter_time=2.047e-04, forward_time=0.156, loss_ctc=9.719, loss_att=4.165, acc=0.976, loss=5.832, backward_time=0.254, grad_norm=71.885, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.302e-04, train_time=1.172
+[bmi2:0/4] 2024-07-10 00:19:57,822 (trainer:779) INFO: 61epoch:train:7848-8260batch: iter_time=2.001e-04, forward_time=0.159, loss_ctc=9.590, loss_att=4.132, acc=0.972, loss=5.769, backward_time=0.255, grad_norm=75.283, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.300e-04, train_time=1.178
+[bmi2:0/4] 2024-07-10 00:20:59,831 (trainer:365) INFO: 61epoch results: [train] iter_time=2.512e-04, forward_time=0.158, loss_ctc=9.629, loss_att=4.131, acc=0.975, loss=5.780, backward_time=0.255, grad_norm=73.237, clip=100.000, loss_scale=5.270e+33, optim_step_time=0.031, optim0_lr0=6.324e-04, train_time=1.184, time=1 hour, 21 minutes and 33.71 seconds, total_count=504287, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.796, cer_ctc=0.040, loss_att=5.859, acc=0.950, cer=0.032, wer=0.487, loss=7.040, time=17.21 seconds, total_count=2074, gpu_max_cached_mem_GB=22.529, [att_plot] time=39.98 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 00:21:05,422 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 00:21:05,505 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/32epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/60epoch.pth
+[bmi2:0/4] 2024-07-10 00:21:05,506 (trainer:299) INFO: 62/100epoch started. Estimated time to finish: 2 days, 6 hours and 7 minutes
+[bmi2:0/4] 2024-07-10 00:22:21,045 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 00:25:19,171 (trainer:779) INFO: 62epoch:train:1-413batch: iter_time=0.001, forward_time=0.157, loss_ctc=9.372, loss_att=4.039, acc=0.974, loss=5.639, backward_time=0.255, grad_norm=73.588, clip=100.000, loss_scale=6.332e+33, optim_step_time=0.031, optim0_lr0=6.297e-04, train_time=1.229
+[bmi2:0/4] 2024-07-10 00:29:23,920 (trainer:779) INFO: 62epoch:train:414-826batch: iter_time=2.068e-04, forward_time=0.159, loss_ctc=9.565, loss_att=4.071, acc=0.977, loss=5.719, backward_time=0.256, grad_norm=72.074, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.294e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 00:33:29,345 (trainer:779) INFO: 62epoch:train:827-1239batch: iter_time=2.077e-04, forward_time=0.160, loss_ctc=9.480, loss_att=4.064, acc=0.977, loss=5.689, backward_time=0.257, grad_norm=74.975, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.292e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 00:37:31,662 (trainer:779) INFO: 62epoch:train:1240-1652batch: iter_time=2.037e-04, forward_time=0.157, loss_ctc=9.494, loss_att=4.086, acc=0.973, loss=5.708, backward_time=0.254, grad_norm=66.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.289e-04, train_time=1.173
+[bmi2:0/4] 2024-07-10 00:41:34,825 (trainer:779) INFO: 62epoch:train:1653-2065batch: iter_time=2.248e-04, forward_time=0.158, loss_ctc=9.411, loss_att=4.052, acc=0.972, loss=5.660, backward_time=0.255, grad_norm=73.519, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.287e-04, train_time=1.178
+[bmi2:0/4] 2024-07-10 00:45:37,713 (trainer:779) INFO: 62epoch:train:2066-2478batch: iter_time=2.029e-04, forward_time=0.158, loss_ctc=9.510, loss_att=4.093, acc=0.976, loss=5.718, backward_time=0.254, grad_norm=70.018, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.284e-04, train_time=1.175
+[bmi2:0/4] 2024-07-10 00:49:41,616 (trainer:779) INFO: 62epoch:train:2479-2891batch: iter_time=2.057e-04, forward_time=0.159, loss_ctc=9.587, loss_att=4.104, acc=0.973, loss=5.749, backward_time=0.255, grad_norm=73.613, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.282e-04, train_time=1.181
+[bmi2:0/4] 2024-07-10 00:53:49,293 (trainer:779) INFO: 62epoch:train:2892-3304batch: iter_time=2.031e-04, forward_time=0.162, loss_ctc=9.529, loss_att=4.107, acc=0.975, loss=5.734, backward_time=0.258, grad_norm=74.090, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.279e-04, train_time=1.199
+[bmi2:0/4] 2024-07-10 00:57:53,931 (trainer:779) INFO: 62epoch:train:3305-3717batch: iter_time=2.034e-04, forward_time=0.160, loss_ctc=9.608, loss_att=4.103, acc=0.976, loss=5.755, backward_time=0.256, grad_norm=73.376, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.276e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 01:01:58,144 (trainer:779) INFO: 62epoch:train:3718-4130batch: iter_time=2.043e-04, forward_time=0.159, loss_ctc=9.484, loss_att=4.078, acc=0.972, loss=5.700, backward_time=0.255, grad_norm=73.691, clip=100.000, loss_scale=5.443e+33, optim_step_time=0.031, optim0_lr0=6.274e-04, train_time=1.182
+[bmi2:0/4] 2024-07-10 01:06:02,114 (trainer:779) INFO: 62epoch:train:4131-4543batch: iter_time=2.015e-04, forward_time=0.158, loss_ctc=9.621, loss_att=4.144, acc=0.977, loss=5.787, backward_time=0.255, grad_norm=75.796, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.271e-04, train_time=1.182
+[bmi2:0/4] 2024-07-10 01:07:25,321 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 01:10:06,980 (trainer:779) INFO: 62epoch:train:4544-4956batch: iter_time=2.079e-04, forward_time=0.160, loss_ctc=9.582, loss_att=4.144, acc=0.974, loss=5.775, backward_time=0.256, grad_norm=76.222, clip=100.000, loss_scale=6.957e+33, optim_step_time=0.031, optim0_lr0=6.269e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 01:14:10,561 (trainer:779) INFO: 62epoch:train:4957-5369batch: iter_time=2.021e-04, forward_time=0.158, loss_ctc=9.656, loss_att=4.143, acc=0.978, loss=5.797, backward_time=0.255, grad_norm=72.996, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.266e-04, train_time=1.180
+[bmi2:0/4] 2024-07-10 01:18:15,142 (trainer:779) INFO: 62epoch:train:5370-5782batch: iter_time=2.124e-04, forward_time=0.159, loss_ctc=9.577, loss_att=4.133, acc=0.975, loss=5.766, backward_time=0.256, grad_norm=72.052, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.264e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 01:22:19,794 (trainer:779) INFO: 62epoch:train:5783-6195batch: iter_time=2.066e-04, forward_time=0.160, loss_ctc=9.624, loss_att=4.170, acc=0.975, loss=5.806, backward_time=0.256, grad_norm=75.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.261e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 01:26:24,097 (trainer:779) INFO: 62epoch:train:6196-6608batch: iter_time=2.129e-04, forward_time=0.159, loss_ctc=9.555, loss_att=4.106, acc=0.974, loss=5.740, backward_time=0.255, grad_norm=72.257, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.259e-04, train_time=1.182
+[bmi2:0/4] 2024-07-10 01:30:28,897 (trainer:779) INFO: 62epoch:train:6609-7021batch: iter_time=2.044e-04, forward_time=0.160, loss_ctc=9.621, loss_att=4.142, acc=0.972, loss=5.786, backward_time=0.256, grad_norm=71.414, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.256e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 01:34:31,966 (trainer:779) INFO: 62epoch:train:7022-7434batch: iter_time=2.127e-04, forward_time=0.157, loss_ctc=9.596, loss_att=4.118, acc=0.976, loss=5.761, backward_time=0.255, grad_norm=73.292, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.254e-04, train_time=1.176
+[bmi2:0/4] 2024-07-10 01:38:35,092 (trainer:779) INFO: 62epoch:train:7435-7847batch: iter_time=2.066e-04, forward_time=0.158, loss_ctc=9.611, loss_att=4.101, acc=0.975, loss=5.754, backward_time=0.254, grad_norm=72.951, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.251e-04, train_time=1.178
+[bmi2:0/4] 2024-07-10 01:42:39,401 (trainer:779) INFO: 62epoch:train:7848-8260batch: iter_time=2.245e-04, forward_time=0.159, loss_ctc=9.559, loss_att=4.089, acc=0.975, loss=5.730, backward_time=0.255, grad_norm=81.930, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.249e-04, train_time=1.183
+[bmi2:0/4] 2024-07-10 01:43:39,229 (trainer:365) INFO: 62epoch results: [train] iter_time=2.594e-04, forward_time=0.159, loss_ctc=9.551, loss_att=4.104, acc=0.975, loss=5.738, backward_time=0.255, grad_norm=73.501, clip=100.000, loss_scale=5.608e+33, optim_step_time=0.031, optim0_lr0=6.273e-04, train_time=1.185, time=1 hour, 21 minutes and 38.98 seconds, total_count=512554, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.802, cer_ctc=0.040, loss_att=5.804, acc=0.950, cer=0.032, wer=0.483, loss=7.004, time=17.67 seconds, total_count=2108, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.08 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 01:43:44,142 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 01:43:44,149 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/35epoch.pth
+[bmi2:0/4] 2024-07-10 01:43:44,150 (trainer:299) INFO: 63/100epoch started. Estimated time to finish: 2 days, 4 hours and 43 minutes
+[bmi2:0/4] 2024-07-10 01:47:58,333 (trainer:779) INFO: 63epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.452, loss_att=4.072, acc=0.976, loss=5.686, backward_time=0.256, grad_norm=76.321, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.246e-04, train_time=1.231
+[bmi2:0/4] 2024-07-10 01:52:03,913 (trainer:779) INFO: 63epoch:train:414-826batch: iter_time=2.694e-04, forward_time=0.160, loss_ctc=9.402, loss_att=4.040, acc=0.974, loss=5.649, backward_time=0.253, grad_norm=75.422, clip=100.000, loss_scale=1.031e+34, optim_step_time=0.033, optim0_lr0=6.243e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 01:56:09,503 (trainer:779) INFO: 63epoch:train:827-1239batch: iter_time=2.562e-04, forward_time=0.160, loss_ctc=9.432, loss_att=4.069, acc=0.977, loss=5.678, backward_time=0.255, grad_norm=77.834, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.241e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 01:58:34,082 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 02:00:12,577 (trainer:779) INFO: 63epoch:train:1240-1652batch: iter_time=2.488e-04, forward_time=0.158, loss_ctc=9.439, loss_att=4.059, acc=0.973, loss=5.673, backward_time=0.253, grad_norm=71.978, clip=100.000, loss_scale=8.267e+33, optim_step_time=0.032, optim0_lr0=6.238e-04, train_time=1.177
+[bmi2:0/4] 2024-07-10 02:04:16,758 (trainer:779) INFO: 63epoch:train:1653-2065batch: iter_time=2.509e-04, forward_time=0.160, loss_ctc=9.494, loss_att=4.094, acc=0.977, loss=5.714, backward_time=0.254, grad_norm=76.090, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.236e-04, train_time=1.183
+[bmi2:0/4] 2024-07-10 02:06:47,485 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 02:08:22,787 (trainer:779) INFO: 63epoch:train:2066-2478batch: iter_time=2.254e-04, forward_time=0.160, loss_ctc=9.522, loss_att=4.085, acc=0.974, loss=5.716, backward_time=0.254, grad_norm=71.927, clip=100.000, loss_scale=4.184e+33, optim_step_time=0.031, optim0_lr0=6.233e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 02:12:27,487 (trainer:779) INFO: 63epoch:train:2479-2891batch: iter_time=2.411e-04, forward_time=0.159, loss_ctc=9.440, loss_att=4.081, acc=0.977, loss=5.689, backward_time=0.253, grad_norm=69.090, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.231e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 02:16:34,519 (trainer:779) INFO: 63epoch:train:2892-3304batch: iter_time=2.378e-04, forward_time=0.162, loss_ctc=9.433, loss_att=4.047, acc=0.973, loss=5.663, backward_time=0.253, grad_norm=72.619, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.228e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 02:20:40,022 (trainer:779) INFO: 63epoch:train:3305-3717batch: iter_time=2.824e-04, forward_time=0.160, loss_ctc=9.522, loss_att=4.090, acc=0.975, loss=5.720, backward_time=0.253, grad_norm=73.628, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.226e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 02:24:48,109 (trainer:779) INFO: 63epoch:train:3718-4130batch: iter_time=3.417e-04, forward_time=0.163, loss_ctc=9.363, loss_att=4.026, acc=0.976, loss=5.627, backward_time=0.255, grad_norm=68.579, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.224e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 02:28:54,382 (trainer:779) INFO: 63epoch:train:4131-4543batch: iter_time=2.349e-04, forward_time=0.162, loss_ctc=9.509, loss_att=4.070, acc=0.978, loss=5.702, backward_time=0.254, grad_norm=71.648, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.221e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 02:33:01,325 (trainer:779) INFO: 63epoch:train:4544-4956batch: iter_time=2.308e-04, forward_time=0.161, loss_ctc=9.436, loss_att=4.078, acc=0.973, loss=5.686, backward_time=0.254, grad_norm=70.473, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.219e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 02:37:06,788 (trainer:779) INFO: 63epoch:train:4957-5369batch: iter_time=2.277e-04, forward_time=0.160, loss_ctc=9.499, loss_att=4.094, acc=0.973, loss=5.715, backward_time=0.253, grad_norm=70.608, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.216e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 02:41:15,143 (trainer:779) INFO: 63epoch:train:5370-5782batch: iter_time=2.303e-04, forward_time=0.161, loss_ctc=9.444, loss_att=4.066, acc=0.976, loss=5.680, backward_time=0.255, grad_norm=77.439, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.214e-04, train_time=1.202
+[bmi2:0/4] 2024-07-10 02:45:21,482 (trainer:779) INFO: 63epoch:train:5783-6195batch: iter_time=2.374e-04, forward_time=0.161, loss_ctc=9.439, loss_att=4.090, acc=0.974, loss=5.694, backward_time=0.254, grad_norm=74.066, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.211e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 02:49:30,444 (trainer:779) INFO: 63epoch:train:6196-6608batch: iter_time=2.468e-04, forward_time=0.163, loss_ctc=9.487, loss_att=4.062, acc=0.977, loss=5.690, backward_time=0.256, grad_norm=76.161, clip=100.000, loss_scale=4.415e+33, optim_step_time=0.036, optim0_lr0=6.209e-04, train_time=1.205
+[bmi2:0/4] 2024-07-10 02:53:37,221 (trainer:779) INFO: 63epoch:train:6609-7021batch: iter_time=2.291e-04, forward_time=0.161, loss_ctc=9.543, loss_att=4.119, acc=0.975, loss=5.746, backward_time=0.255, grad_norm=77.365, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.206e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 02:57:43,490 (trainer:779) INFO: 63epoch:train:7022-7434batch: iter_time=2.748e-04, forward_time=0.161, loss_ctc=9.327, loss_att=4.018, acc=0.976, loss=5.610, backward_time=0.254, grad_norm=73.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.204e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 03:01:50,270 (trainer:779) INFO: 63epoch:train:7435-7847batch: iter_time=2.361e-04, forward_time=0.161, loss_ctc=9.490, loss_att=4.071, acc=0.971, loss=5.696, backward_time=0.254, grad_norm=75.319, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.201e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 03:05:57,279 (trainer:779) INFO: 63epoch:train:7848-8260batch: iter_time=2.316e-04, forward_time=0.161, loss_ctc=9.549, loss_att=4.111, acc=0.976, loss=5.743, backward_time=0.254, grad_norm=75.961, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.199e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 03:07:01,693 (trainer:365) INFO: 63epoch results: [train] iter_time=2.912e-04, forward_time=0.161, loss_ctc=9.461, loss_att=4.072, acc=0.975, loss=5.689, backward_time=0.254, grad_norm=73.775, clip=100.000, loss_scale=4.604e+33, optim_step_time=0.033, optim0_lr0=6.222e-04, train_time=1.194, time=1 hour, 22 minutes and 18.72 seconds, total_count=520821, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.814, cer_ctc=0.041, loss_att=5.947, acc=0.945, cer=0.032, wer=0.487, loss=7.107, time=18.64 seconds, total_count=2142, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.18 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 03:07:06,466 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 03:07:06,467 (trainer:299) INFO: 64/100epoch started. Estimated time to finish: 2 days, 3 hours and 20 minutes
+[bmi2:0/4] 2024-07-10 03:09:50,330 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 03:11:23,860 (trainer:779) INFO: 64epoch:train:1-413batch: iter_time=0.001, forward_time=0.161, loss_ctc=9.358, loss_att=4.020, acc=0.974, loss=5.621, backward_time=0.253, grad_norm=74.527, clip=100.000, loss_scale=4.204e+33, optim_step_time=0.034, optim0_lr0=6.196e-04, train_time=1.247
+[bmi2:0/4] 2024-07-10 03:15:31,512 (trainer:779) INFO: 64epoch:train:414-826batch: iter_time=2.447e-04, forward_time=0.163, loss_ctc=9.331, loss_att=4.036, acc=0.975, loss=5.624, backward_time=0.255, grad_norm=73.240, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.194e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 03:19:40,492 (trainer:779) INFO: 64epoch:train:827-1239batch: iter_time=2.716e-04, forward_time=0.163, loss_ctc=9.468, loss_att=4.060, acc=0.978, loss=5.683, backward_time=0.256, grad_norm=75.996, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.191e-04, train_time=1.206
+[bmi2:0/4] 2024-07-10 03:23:47,522 (trainer:779) INFO: 64epoch:train:1240-1652batch: iter_time=2.839e-04, forward_time=0.161, loss_ctc=9.404, loss_att=4.025, acc=0.976, loss=5.639, backward_time=0.253, grad_norm=70.447, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.189e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 03:27:54,125 (trainer:779) INFO: 64epoch:train:1653-2065batch: iter_time=2.359e-04, forward_time=0.163, loss_ctc=9.463, loss_att=4.063, acc=0.976, loss=5.683, backward_time=0.253, grad_norm=73.838, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.186e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 03:32:01,575 (trainer:779) INFO: 64epoch:train:2066-2478batch: iter_time=2.380e-04, forward_time=0.162, loss_ctc=9.378, loss_att=4.029, acc=0.973, loss=5.634, backward_time=0.253, grad_norm=75.205, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.184e-04, train_time=1.197
+[bmi2:0/4] 2024-07-10 03:36:07,722 (trainer:779) INFO: 64epoch:train:2479-2891batch: iter_time=2.359e-04, forward_time=0.161, loss_ctc=9.464, loss_att=4.041, acc=0.977, loss=5.668, backward_time=0.253, grad_norm=76.549, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.182e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 03:40:14,854 (trainer:779) INFO: 64epoch:train:2892-3304batch: iter_time=2.475e-04, forward_time=0.162, loss_ctc=9.429, loss_att=4.037, acc=0.973, loss=5.655, backward_time=0.253, grad_norm=75.310, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.179e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 03:44:22,154 (trainer:779) INFO: 64epoch:train:3305-3717batch: iter_time=2.420e-04, forward_time=0.162, loss_ctc=9.414, loss_att=4.045, acc=0.977, loss=5.655, backward_time=0.254, grad_norm=75.595, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.177e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 03:48:28,936 (trainer:779) INFO: 64epoch:train:3718-4130batch: iter_time=2.493e-04, forward_time=0.162, loss_ctc=9.463, loss_att=4.050, acc=0.979, loss=5.674, backward_time=0.253, grad_norm=79.165, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.174e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 03:52:35,787 (trainer:779) INFO: 64epoch:train:4131-4543batch: iter_time=2.399e-04, forward_time=0.161, loss_ctc=9.587, loss_att=4.126, acc=0.977, loss=5.765, backward_time=0.254, grad_norm=79.129, clip=100.000, loss_scale=4.398e+33, optim_step_time=0.034, optim0_lr0=6.172e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 03:56:42,357 (trainer:779) INFO: 64epoch:train:4544-4956batch: iter_time=2.350e-04, forward_time=0.161, loss_ctc=9.457, loss_att=4.050, acc=0.974, loss=5.672, backward_time=0.253, grad_norm=74.812, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.169e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 04:00:48,008 (trainer:779) INFO: 64epoch:train:4957-5369batch: iter_time=2.362e-04, forward_time=0.161, loss_ctc=9.355, loss_att=4.017, acc=0.972, loss=5.619, backward_time=0.253, grad_norm=75.559, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.167e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 04:04:54,252 (trainer:779) INFO: 64epoch:train:5370-5782batch: iter_time=2.362e-04, forward_time=0.160, loss_ctc=9.447, loss_att=4.038, acc=0.973, loss=5.661, backward_time=0.253, grad_norm=75.363, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=6.165e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 04:09:00,354 (trainer:779) INFO: 64epoch:train:5783-6195batch: iter_time=2.372e-04, forward_time=0.161, loss_ctc=9.467, loss_att=4.049, acc=0.975, loss=5.675, backward_time=0.254, grad_norm=79.722, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.162e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 04:13:07,574 (trainer:779) INFO: 64epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.160, loss_ctc=9.529, loss_att=4.074, acc=0.976, loss=5.710, backward_time=0.255, grad_norm=74.410, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.160e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 04:17:15,420 (trainer:779) INFO: 64epoch:train:6609-7021batch: iter_time=2.661e-04, forward_time=0.163, loss_ctc=9.573, loss_att=4.127, acc=0.974, loss=5.761, backward_time=0.255, grad_norm=81.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.157e-04, train_time=1.201
+[bmi2:0/4] 2024-07-10 04:19:00,935 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 04:21:19,628 (trainer:779) INFO: 64epoch:train:7022-7434batch: iter_time=2.556e-04, forward_time=0.160, loss_ctc=9.523, loss_att=4.112, acc=0.972, loss=5.735, backward_time=0.253, grad_norm=77.837, clip=100.000, loss_scale=3.718e+33, optim_step_time=0.031, optim0_lr0=6.155e-04, train_time=1.182
+[bmi2:0/4] 2024-07-10 04:25:25,605 (trainer:779) INFO: 64epoch:train:7435-7847batch: iter_time=2.211e-04, forward_time=0.161, loss_ctc=9.549, loss_att=4.084, acc=0.974, loss=5.724, backward_time=0.254, grad_norm=77.443, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.153e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 04:29:29,840 (trainer:779) INFO: 64epoch:train:7848-8260batch: iter_time=2.075e-04, forward_time=0.160, loss_ctc=9.486, loss_att=4.065, acc=0.977, loss=5.691, backward_time=0.252, grad_norm=81.804, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.150e-04, train_time=1.182
+[bmi2:0/4] 2024-07-10 04:30:52,517 (trainer:365) INFO: 64epoch results: [train] iter_time=3.658e-04, forward_time=0.161, loss_ctc=9.457, loss_att=4.057, acc=0.975, loss=5.677, backward_time=0.254, grad_norm=76.361, clip=100.000, loss_scale=3.600e+33, optim_step_time=0.033, optim0_lr0=6.173e-04, train_time=1.197, time=1 hour, 22 minutes and 28.47 seconds, total_count=529088, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.657, cer_ctc=0.041, loss_att=5.711, acc=0.951, cer=0.032, wer=0.480, loss=6.895, time=39.7 seconds, total_count=2176, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.88 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 04:30:57,517 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-10 04:30:57,551 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/55epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/63epoch.pth
+[bmi2:0/4] 2024-07-10 04:30:57,552 (trainer:299) INFO: 65/100epoch started. Estimated time to finish: 2 days, 1 hour and 57 minutes
+[bmi2:0/4] 2024-07-10 04:35:14,086 (trainer:779) INFO: 65epoch:train:1-413batch: iter_time=0.002, forward_time=0.163, loss_ctc=9.346, loss_att=4.002, acc=0.975, loss=5.605, backward_time=0.254, grad_norm=79.214, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.148e-04, train_time=1.242
+[bmi2:0/4] 2024-07-10 04:39:19,894 (trainer:779) INFO: 65epoch:train:414-826batch: iter_time=2.195e-04, forward_time=0.162, loss_ctc=9.340, loss_att=4.007, acc=0.975, loss=5.607, backward_time=0.253, grad_norm=80.242, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.145e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 04:43:24,539 (trainer:779) INFO: 65epoch:train:827-1239batch: iter_time=6.925e-04, forward_time=0.160, loss_ctc=9.425, loss_att=4.073, acc=0.972, loss=5.679, backward_time=0.252, grad_norm=71.278, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.143e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 04:47:29,444 (trainer:779) INFO: 65epoch:train:1240-1652batch: iter_time=4.157e-04, forward_time=0.160, loss_ctc=9.393, loss_att=4.005, acc=0.975, loss=5.622, backward_time=0.253, grad_norm=73.103, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=6.141e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 04:51:34,770 (trainer:779) INFO: 65epoch:train:1653-2065batch: iter_time=2.493e-04, forward_time=0.159, loss_ctc=9.344, loss_att=3.995, acc=0.975, loss=5.600, backward_time=0.254, grad_norm=75.353, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.138e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 04:55:41,059 (trainer:779) INFO: 65epoch:train:2066-2478batch: iter_time=2.752e-04, forward_time=0.161, loss_ctc=9.384, loss_att=4.011, acc=0.977, loss=5.623, backward_time=0.255, grad_norm=77.327, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.136e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 04:59:46,172 (trainer:779) INFO: 65epoch:train:2479-2891batch: iter_time=2.222e-04, forward_time=0.161, loss_ctc=9.382, loss_att=4.012, acc=0.977, loss=5.623, backward_time=0.255, grad_norm=72.764, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=6.133e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 05:03:52,169 (trainer:779) INFO: 65epoch:train:2892-3304batch: iter_time=2.110e-04, forward_time=0.160, loss_ctc=9.466, loss_att=4.063, acc=0.978, loss=5.684, backward_time=0.255, grad_norm=75.210, clip=100.000, loss_scale=4.916e+33, optim_step_time=0.030, optim0_lr0=6.131e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 05:07:57,025 (trainer:779) INFO: 65epoch:train:3305-3717batch: iter_time=2.148e-04, forward_time=0.160, loss_ctc=9.337, loss_att=4.010, acc=0.976, loss=5.608, backward_time=0.254, grad_norm=77.764, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.129e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 05:12:04,963 (trainer:779) INFO: 65epoch:train:3718-4130batch: iter_time=2.186e-04, forward_time=0.163, loss_ctc=9.443, loss_att=4.028, acc=0.978, loss=5.653, backward_time=0.256, grad_norm=74.344, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.126e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 05:16:10,659 (trainer:779) INFO: 65epoch:train:4131-4543batch: iter_time=2.127e-04, forward_time=0.160, loss_ctc=9.564, loss_att=4.104, acc=0.977, loss=5.742, backward_time=0.255, grad_norm=78.067, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.124e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 05:20:15,389 (trainer:779) INFO: 65epoch:train:4544-4956batch: iter_time=2.117e-04, forward_time=0.160, loss_ctc=9.287, loss_att=3.990, acc=0.975, loss=5.579, backward_time=0.253, grad_norm=76.386, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.121e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 05:24:20,396 (trainer:779) INFO: 65epoch:train:4957-5369batch: iter_time=1.946e-04, forward_time=0.160, loss_ctc=9.308, loss_att=3.999, acc=0.974, loss=5.592, backward_time=0.254, grad_norm=76.355, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.119e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 05:28:26,776 (trainer:779) INFO: 65epoch:train:5370-5782batch: iter_time=2.089e-04, forward_time=0.161, loss_ctc=9.560, loss_att=4.083, acc=0.975, loss=5.726, backward_time=0.254, grad_norm=76.774, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.117e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 05:32:32,287 (trainer:779) INFO: 65epoch:train:5783-6195batch: iter_time=2.095e-04, forward_time=0.162, loss_ctc=9.489, loss_att=4.062, acc=0.974, loss=5.690, backward_time=0.253, grad_norm=74.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.114e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 05:36:38,165 (trainer:779) INFO: 65epoch:train:6196-6608batch: iter_time=2.041e-04, forward_time=0.161, loss_ctc=9.348, loss_att=4.009, acc=0.974, loss=5.611, backward_time=0.255, grad_norm=81.775, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.112e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 05:40:41,885 (trainer:779) INFO: 65epoch:train:6609-7021batch: iter_time=2.600e-04, forward_time=0.159, loss_ctc=9.569, loss_att=4.064, acc=0.976, loss=5.716, backward_time=0.254, grad_norm=73.942, clip=100.000, loss_scale=6.276e+33, optim_step_time=0.029, optim0_lr0=6.110e-04, train_time=1.180
+[bmi2:0/4] 2024-07-10 05:44:46,457 (trainer:779) INFO: 65epoch:train:7022-7434batch: iter_time=2.185e-04, forward_time=0.159, loss_ctc=9.538, loss_att=4.082, acc=0.978, loss=5.719, backward_time=0.255, grad_norm=71.933, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.029, optim0_lr0=6.107e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 05:47:14,641 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 05:48:51,315 (trainer:779) INFO: 65epoch:train:7435-7847batch: iter_time=4.498e-04, forward_time=0.159, loss_ctc=9.351, loss_att=4.041, acc=0.972, loss=5.634, backward_time=0.254, grad_norm=74.600, clip=100.000, loss_scale=8.333e+33, optim_step_time=0.029, optim0_lr0=6.105e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 05:52:55,947 (trainer:779) INFO: 65epoch:train:7848-8260batch: iter_time=2.368e-04, forward_time=0.160, loss_ctc=9.541, loss_att=4.069, acc=0.974, loss=5.711, backward_time=0.252, grad_norm=73.937, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.103e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 05:53:55,032 (trainer:365) INFO: 65epoch results: [train] iter_time=3.404e-04, forward_time=0.160, loss_ctc=9.421, loss_att=4.035, acc=0.975, loss=5.651, backward_time=0.254, grad_norm=75.754, clip=100.000, loss_scale=4.741e+33, optim_step_time=0.031, optim0_lr0=6.125e-04, train_time=1.191, time=1 hour, 22 minutes and 3.41 seconds, total_count=537355, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.883, cer_ctc=0.041, loss_att=5.698, acc=0.950, cer=0.032, wer=0.486, loss=6.953, time=17.87 seconds, total_count=2210, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.2 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 05:53:59,638 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 05:53:59,645 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/58epoch.pth
+[bmi2:0/4] 2024-07-10 05:53:59,645 (trainer:299) INFO: 66/100epoch started. Estimated time to finish: 2 days, 34 minutes and 5.59 seconds
+[bmi2:0/4] 2024-07-10 05:58:14,661 (trainer:779) INFO: 66epoch:train:1-413batch: iter_time=0.002, forward_time=0.162, loss_ctc=9.358, loss_att=4.012, acc=0.975, loss=5.616, backward_time=0.253, grad_norm=79.004, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=6.100e-04, train_time=1.236
+[bmi2:0/4] 2024-07-10 06:02:21,303 (trainer:779) INFO: 66epoch:train:414-826batch: iter_time=0.003, forward_time=0.160, loss_ctc=9.519, loss_att=4.056, acc=0.975, loss=5.695, backward_time=0.255, grad_norm=77.130, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=6.098e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 06:06:28,671 (trainer:779) INFO: 66epoch:train:827-1239batch: iter_time=0.003, forward_time=0.160, loss_ctc=9.561, loss_att=4.080, acc=0.972, loss=5.725, backward_time=0.253, grad_norm=84.358, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.096e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 06:10:35,521 (trainer:779) INFO: 66epoch:train:1240-1652batch: iter_time=2.563e-04, forward_time=0.161, loss_ctc=9.647, loss_att=4.098, acc=0.974, loss=5.763, backward_time=0.254, grad_norm=83.790, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.093e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 06:14:42,813 (trainer:779) INFO: 66epoch:train:1653-2065batch: iter_time=3.450e-04, forward_time=0.162, loss_ctc=9.635, loss_att=4.101, acc=0.974, loss=5.761, backward_time=0.256, grad_norm=80.971, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.091e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 06:18:49,388 (trainer:779) INFO: 66epoch:train:2066-2478batch: iter_time=2.415e-04, forward_time=0.160, loss_ctc=9.619, loss_att=4.080, acc=0.977, loss=5.742, backward_time=0.254, grad_norm=83.171, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.089e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 06:22:55,830 (trainer:779) INFO: 66epoch:train:2479-2891batch: iter_time=2.422e-04, forward_time=0.161, loss_ctc=9.459, loss_att=4.048, acc=0.975, loss=5.671, backward_time=0.254, grad_norm=79.402, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.086e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 06:27:01,825 (trainer:779) INFO: 66epoch:train:2892-3304batch: iter_time=2.277e-04, forward_time=0.160, loss_ctc=9.593, loss_att=4.087, acc=0.975, loss=5.739, backward_time=0.254, grad_norm=76.412, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.084e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 06:31:06,988 (trainer:779) INFO: 66epoch:train:3305-3717batch: iter_time=3.094e-04, forward_time=0.159, loss_ctc=9.472, loss_att=4.035, acc=0.977, loss=5.666, backward_time=0.254, grad_norm=83.097, clip=100.000, loss_scale=8.948e+33, optim_step_time=0.032, optim0_lr0=6.082e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 06:35:11,548 (trainer:779) INFO: 66epoch:train:3718-4130batch: iter_time=2.453e-04, forward_time=0.159, loss_ctc=9.569, loss_att=4.093, acc=0.973, loss=5.736, backward_time=0.252, grad_norm=79.015, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.031, optim0_lr0=6.079e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 06:39:16,090 (trainer:779) INFO: 66epoch:train:4131-4543batch: iter_time=2.388e-04, forward_time=0.159, loss_ctc=9.651, loss_att=4.119, acc=0.975, loss=5.779, backward_time=0.253, grad_norm=83.938, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.033, optim0_lr0=6.077e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 06:43:21,354 (trainer:779) INFO: 66epoch:train:4544-4956batch: iter_time=2.444e-04, forward_time=0.158, loss_ctc=9.647, loss_att=4.093, acc=0.974, loss=5.759, backward_time=0.255, grad_norm=79.957, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.032, optim0_lr0=6.075e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 06:47:28,784 (trainer:779) INFO: 66epoch:train:4957-5369batch: iter_time=2.396e-04, forward_time=0.162, loss_ctc=9.550, loss_att=4.046, acc=0.975, loss=5.697, backward_time=0.254, grad_norm=77.155, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.035, optim0_lr0=6.072e-04, train_time=1.199
+[bmi2:0/4] 2024-07-10 06:51:36,405 (trainer:779) INFO: 66epoch:train:5370-5782batch: iter_time=2.341e-04, forward_time=0.161, loss_ctc=9.477, loss_att=4.065, acc=0.974, loss=5.689, backward_time=0.254, grad_norm=78.990, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.033, optim0_lr0=6.070e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 06:52:15,503 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 06:55:42,694 (trainer:779) INFO: 66epoch:train:5783-6195batch: iter_time=2.262e-04, forward_time=0.160, loss_ctc=9.512, loss_att=4.052, acc=0.976, loss=5.690, backward_time=0.255, grad_norm=75.004, clip=100.000, loss_scale=6.003e+33, optim_step_time=0.032, optim0_lr0=6.068e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 06:59:49,395 (trainer:779) INFO: 66epoch:train:6196-6608batch: iter_time=2.322e-04, forward_time=0.161, loss_ctc=9.586, loss_att=4.117, acc=0.972, loss=5.758, backward_time=0.253, grad_norm=75.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.065e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 07:03:54,797 (trainer:779) INFO: 66epoch:train:6609-7021batch: iter_time=2.187e-04, forward_time=0.160, loss_ctc=9.539, loss_att=4.078, acc=0.976, loss=5.716, backward_time=0.254, grad_norm=75.842, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.063e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 07:08:01,544 (trainer:779) INFO: 66epoch:train:7022-7434batch: iter_time=2.415e-04, forward_time=0.161, loss_ctc=9.508, loss_att=4.040, acc=0.977, loss=5.680, backward_time=0.254, grad_norm=71.119, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.061e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 07:12:08,012 (trainer:779) INFO: 66epoch:train:7435-7847batch: iter_time=2.262e-04, forward_time=0.161, loss_ctc=9.489, loss_att=4.038, acc=0.976, loss=5.674, backward_time=0.254, grad_norm=75.384, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.059e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 07:16:13,523 (trainer:779) INFO: 66epoch:train:7848-8260batch: iter_time=2.276e-04, forward_time=0.160, loss_ctc=9.525, loss_att=4.038, acc=0.979, loss=5.684, backward_time=0.254, grad_norm=77.382, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.056e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 07:17:12,877 (trainer:365) INFO: 66epoch results: [train] iter_time=5.575e-04, forward_time=0.160, loss_ctc=9.545, loss_att=4.069, acc=0.975, loss=5.712, backward_time=0.254, grad_norm=78.848, clip=100.000, loss_scale=6.718e+33, optim_step_time=0.032, optim0_lr0=6.078e-04, train_time=1.194, time=1 hour, 22 minutes and 19.12 seconds, total_count=545622, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.707, cer_ctc=0.040, loss_att=5.811, acc=0.949, cer=0.032, wer=0.487, loss=6.980, time=17.83 seconds, total_count=2244, gpu_max_cached_mem_GB=22.529, [att_plot] time=36.28 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 07:17:17,958 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 07:17:17,960 (trainer:299) INFO: 67/100epoch started. Estimated time to finish: 1 day, 23 hours and 10 minutes
+[bmi2:0/4] 2024-07-10 07:21:34,176 (trainer:779) INFO: 67epoch:train:1-413batch: iter_time=0.001, forward_time=0.161, loss_ctc=9.322, loss_att=3.995, acc=0.975, loss=5.593, backward_time=0.254, grad_norm=79.192, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.054e-04, train_time=1.241
+[bmi2:0/4] 2024-07-10 07:25:40,925 (trainer:779) INFO: 67epoch:train:414-826batch: iter_time=2.323e-04, forward_time=0.160, loss_ctc=9.312, loss_att=3.998, acc=0.977, loss=5.592, backward_time=0.254, grad_norm=75.312, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.052e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 07:29:48,164 (trainer:779) INFO: 67epoch:train:827-1239batch: iter_time=0.001, forward_time=0.162, loss_ctc=9.333, loss_att=3.990, acc=0.975, loss=5.593, backward_time=0.255, grad_norm=76.579, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.049e-04, train_time=1.197
+[bmi2:0/4] 2024-07-10 07:33:54,851 (trainer:779) INFO: 67epoch:train:1240-1652batch: iter_time=2.362e-04, forward_time=0.162, loss_ctc=9.346, loss_att=3.982, acc=0.978, loss=5.591, backward_time=0.253, grad_norm=75.259, clip=100.000, loss_scale=6.070e+33, optim_step_time=0.034, optim0_lr0=6.047e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 07:37:08,429 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 07:38:01,006 (trainer:779) INFO: 67epoch:train:1653-2065batch: iter_time=2.404e-04, forward_time=0.160, loss_ctc=9.458, loss_att=4.034, acc=0.977, loss=5.661, backward_time=0.254, grad_norm=76.599, clip=100.000, loss_scale=9.270e+33, optim_step_time=0.034, optim0_lr0=6.045e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 07:42:08,193 (trainer:779) INFO: 67epoch:train:2066-2478batch: iter_time=2.440e-04, forward_time=0.162, loss_ctc=9.368, loss_att=3.992, acc=0.977, loss=5.605, backward_time=0.254, grad_norm=73.513, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.042e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 07:46:15,032 (trainer:779) INFO: 67epoch:train:2479-2891batch: iter_time=2.526e-04, forward_time=0.161, loss_ctc=9.365, loss_att=3.994, acc=0.974, loss=5.605, backward_time=0.255, grad_norm=74.071, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.040e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 07:50:21,927 (trainer:779) INFO: 67epoch:train:2892-3304batch: iter_time=2.685e-04, forward_time=0.160, loss_ctc=9.438, loss_att=4.003, acc=0.977, loss=5.634, backward_time=0.255, grad_norm=71.313, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=6.038e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 07:54:26,101 (trainer:779) INFO: 67epoch:train:3305-3717batch: iter_time=2.417e-04, forward_time=0.159, loss_ctc=9.367, loss_att=4.000, acc=0.975, loss=5.610, backward_time=0.252, grad_norm=78.743, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=6.036e-04, train_time=1.183
+[bmi2:0/4] 2024-07-10 07:55:13,983 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 07:58:33,991 (trainer:779) INFO: 67epoch:train:3718-4130batch: iter_time=2.374e-04, forward_time=0.162, loss_ctc=9.353, loss_att=3.995, acc=0.977, loss=5.603, backward_time=0.255, grad_norm=72.251, clip=100.000, loss_scale=3.088e+33, optim_step_time=0.033, optim0_lr0=6.033e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 08:02:41,231 (trainer:779) INFO: 67epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.161, loss_ctc=9.329, loss_att=3.966, acc=0.974, loss=5.575, backward_time=0.254, grad_norm=69.701, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.031e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 08:06:47,958 (trainer:779) INFO: 67epoch:train:4544-4956batch: iter_time=2.459e-04, forward_time=0.161, loss_ctc=9.343, loss_att=4.002, acc=0.973, loss=5.604, backward_time=0.254, grad_norm=88.140, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.033, optim0_lr0=6.029e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 08:10:53,265 (trainer:779) INFO: 67epoch:train:4957-5369batch: iter_time=2.222e-04, forward_time=0.160, loss_ctc=9.390, loss_att=4.018, acc=0.976, loss=5.630, backward_time=0.253, grad_norm=71.959, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.027e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 08:14:59,562 (trainer:779) INFO: 67epoch:train:5370-5782batch: iter_time=2.315e-04, forward_time=0.159, loss_ctc=9.398, loss_att=4.009, acc=0.974, loss=5.626, backward_time=0.254, grad_norm=69.012, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.024e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 08:19:04,606 (trainer:779) INFO: 67epoch:train:5783-6195batch: iter_time=2.413e-04, forward_time=0.159, loss_ctc=9.424, loss_att=4.031, acc=0.974, loss=5.649, backward_time=0.253, grad_norm=74.911, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.022e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 08:23:10,487 (trainer:779) INFO: 67epoch:train:6196-6608batch: iter_time=2.232e-04, forward_time=0.159, loss_ctc=9.531, loss_att=4.064, acc=0.975, loss=5.704, backward_time=0.253, grad_norm=76.989, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=6.020e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 08:27:18,105 (trainer:779) INFO: 67epoch:train:6609-7021batch: iter_time=2.387e-04, forward_time=0.161, loss_ctc=9.582, loss_att=4.086, acc=0.973, loss=5.735, backward_time=0.255, grad_norm=81.003, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=6.018e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 08:29:37,935 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 08:31:23,018 (trainer:779) INFO: 67epoch:train:7022-7434batch: iter_time=2.149e-04, forward_time=0.159, loss_ctc=9.631, loss_att=4.061, acc=0.974, loss=5.732, backward_time=0.252, grad_norm=78.374, clip=100.000, loss_scale=2.042e+33, optim_step_time=0.032, optim0_lr0=6.015e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 08:35:27,963 (trainer:779) INFO: 67epoch:train:7435-7847batch: iter_time=2.360e-04, forward_time=0.159, loss_ctc=9.456, loss_att=3.998, acc=0.978, loss=5.635, backward_time=0.253, grad_norm=77.312, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.033, optim0_lr0=6.013e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 08:39:33,988 (trainer:779) INFO: 67epoch:train:7848-8260batch: iter_time=2.727e-04, forward_time=0.160, loss_ctc=9.329, loss_att=3.978, acc=0.977, loss=5.583, backward_time=0.254, grad_norm=71.293, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.032, optim0_lr0=6.011e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 08:40:37,901 (trainer:365) INFO: 67epoch results: [train] iter_time=4.167e-04, forward_time=0.160, loss_ctc=9.404, loss_att=4.010, acc=0.975, loss=5.628, backward_time=0.254, grad_norm=75.554, clip=100.000, loss_scale=3.876e+33, optim_step_time=0.033, optim0_lr0=6.032e-04, train_time=1.195, time=1 hour, 22 minutes and 21.1 seconds, total_count=553889, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.891, cer_ctc=0.040, loss_att=5.754, acc=0.949, cer=0.032, wer=0.486, loss=6.995, time=19.05 seconds, total_count=2278, gpu_max_cached_mem_GB=22.529, [att_plot] time=39.78 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 08:40:42,181 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 08:40:42,189 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/66epoch.pth
+[bmi2:0/4] 2024-07-10 08:40:42,189 (trainer:299) INFO: 68/100epoch started. Estimated time to finish: 1 day, 21 hours and 47 minutes
+[bmi2:0/4] 2024-07-10 08:44:59,873 (trainer:779) INFO: 68epoch:train:1-413batch: iter_time=0.001, forward_time=0.160, loss_ctc=9.333, loss_att=3.953, acc=0.976, loss=5.567, backward_time=0.256, grad_norm=72.900, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.032, optim0_lr0=6.009e-04, train_time=1.248
+[bmi2:0/4] 2024-07-10 08:49:05,359 (trainer:779) INFO: 68epoch:train:414-826batch: iter_time=2.427e-04, forward_time=0.159, loss_ctc=9.334, loss_att=3.983, acc=0.975, loss=5.588, backward_time=0.254, grad_norm=71.577, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.032, optim0_lr0=6.006e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 08:53:12,864 (trainer:779) INFO: 68epoch:train:827-1239batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.258, loss_att=3.962, acc=0.976, loss=5.551, backward_time=0.257, grad_norm=73.333, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.031, optim0_lr0=6.004e-04, train_time=1.199
+[bmi2:0/4] 2024-07-10 08:57:20,983 (trainer:779) INFO: 68epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.228, loss_att=3.943, acc=0.975, loss=5.528, backward_time=0.257, grad_norm=71.433, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.033, optim0_lr0=6.002e-04, train_time=1.201
+[bmi2:0/4] 2024-07-10 09:01:45,643 (trainer:779) INFO: 68epoch:train:1653-2065batch: iter_time=0.036, forward_time=0.160, loss_ctc=9.363, loss_att=3.979, acc=0.978, loss=5.595, backward_time=0.257, grad_norm=70.358, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.033, optim0_lr0=6.000e-04, train_time=1.282
+[bmi2:0/4] 2024-07-10 09:05:54,745 (trainer:779) INFO: 68epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.358, loss_att=3.986, acc=0.976, loss=5.598, backward_time=0.257, grad_norm=70.489, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.033, optim0_lr0=5.997e-04, train_time=1.205
+[bmi2:0/4] 2024-07-10 09:10:02,624 (trainer:779) INFO: 68epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.158, loss_ctc=9.252, loss_att=3.939, acc=0.975, loss=5.533, backward_time=0.256, grad_norm=69.020, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.033, optim0_lr0=5.995e-04, train_time=1.201
+[bmi2:0/4] 2024-07-10 09:14:13,723 (trainer:779) INFO: 68epoch:train:2892-3304batch: iter_time=2.742e-04, forward_time=0.164, loss_ctc=9.291, loss_att=3.936, acc=0.976, loss=5.542, backward_time=0.255, grad_norm=72.445, clip=100.000, loss_scale=2.276e+33, optim_step_time=0.037, optim0_lr0=5.993e-04, train_time=1.215
+[bmi2:0/4] 2024-07-10 09:18:22,569 (trainer:779) INFO: 68epoch:train:3305-3717batch: iter_time=6.718e-04, forward_time=0.163, loss_ctc=9.287, loss_att=3.937, acc=0.974, loss=5.542, backward_time=0.255, grad_norm=73.062, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.034, optim0_lr0=5.991e-04, train_time=1.206
+[bmi2:0/4] 2024-07-10 09:22:30,787 (trainer:779) INFO: 68epoch:train:3718-4130batch: iter_time=2.535e-04, forward_time=0.163, loss_ctc=9.298, loss_att=3.963, acc=0.978, loss=5.564, backward_time=0.254, grad_norm=70.624, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.989e-04, train_time=1.201
+[bmi2:0/4] 2024-07-10 09:26:35,645 (trainer:779) INFO: 68epoch:train:4131-4543batch: iter_time=6.514e-04, forward_time=0.159, loss_ctc=9.225, loss_att=3.947, acc=0.973, loss=5.530, backward_time=0.253, grad_norm=72.520, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.986e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 09:30:41,977 (trainer:779) INFO: 68epoch:train:4544-4956batch: iter_time=2.180e-04, forward_time=0.158, loss_ctc=9.291, loss_att=3.955, acc=0.976, loss=5.556, backward_time=0.254, grad_norm=73.563, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.984e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 09:34:48,930 (trainer:779) INFO: 68epoch:train:4957-5369batch: iter_time=2.414e-04, forward_time=0.160, loss_ctc=9.337, loss_att=4.021, acc=0.973, loss=5.616, backward_time=0.256, grad_norm=75.194, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.982e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 09:38:54,821 (trainer:779) INFO: 68epoch:train:5370-5782batch: iter_time=2.808e-04, forward_time=0.160, loss_ctc=9.184, loss_att=3.930, acc=0.976, loss=5.506, backward_time=0.254, grad_norm=68.565, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=5.980e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 09:42:59,508 (trainer:779) INFO: 68epoch:train:5783-6195batch: iter_time=2.423e-04, forward_time=0.160, loss_ctc=9.314, loss_att=3.983, acc=0.975, loss=5.582, backward_time=0.253, grad_norm=72.634, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.977e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 09:47:05,868 (trainer:779) INFO: 68epoch:train:6196-6608batch: iter_time=2.351e-04, forward_time=0.159, loss_ctc=9.321, loss_att=3.959, acc=0.974, loss=5.567, backward_time=0.254, grad_norm=73.701, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=5.975e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 09:51:11,296 (trainer:779) INFO: 68epoch:train:6609-7021batch: iter_time=2.356e-04, forward_time=0.160, loss_ctc=9.411, loss_att=4.006, acc=0.980, loss=5.628, backward_time=0.254, grad_norm=70.761, clip=100.000, loss_scale=2.773e+33, optim_step_time=0.031, optim0_lr0=5.973e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 09:55:18,129 (trainer:779) INFO: 68epoch:train:7022-7434batch: iter_time=2.463e-04, forward_time=0.161, loss_ctc=9.330, loss_att=3.976, acc=0.976, loss=5.582, backward_time=0.253, grad_norm=71.480, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.971e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 09:59:23,515 (trainer:779) INFO: 68epoch:train:7435-7847batch: iter_time=2.437e-04, forward_time=0.160, loss_ctc=9.277, loss_att=3.953, acc=0.976, loss=5.550, backward_time=0.253, grad_norm=73.002, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.969e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 10:03:29,779 (trainer:779) INFO: 68epoch:train:7848-8260batch: iter_time=2.505e-04, forward_time=0.160, loss_ctc=9.330, loss_att=3.994, acc=0.975, loss=5.595, backward_time=0.253, grad_norm=73.386, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=5.966e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 10:04:32,412 (trainer:365) INFO: 68epoch results: [train] iter_time=0.002, forward_time=0.160, loss_ctc=9.301, loss_att=3.965, acc=0.976, loss=5.566, backward_time=0.255, grad_norm=72.003, clip=100.000, loss_scale=2.526e+33, optim_step_time=0.032, optim0_lr0=5.987e-04, train_time=1.203, time=1 hour, 22 minutes and 52.73 seconds, total_count=562156, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.924, cer_ctc=0.040, loss_att=5.959, acc=0.949, cer=0.033, wer=0.484, loss=7.148, time=18.74 seconds, total_count=2312, gpu_max_cached_mem_GB=22.529, [att_plot] time=38.75 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 10:04:37,160 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 10:04:37,214 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/67epoch.pth
+[bmi2:0/4] 2024-07-10 10:04:37,214 (trainer:299) INFO: 69/100epoch started. Estimated time to finish: 1 day, 20 hours and 24 minutes
+[bmi2:0/4] 2024-07-10 10:08:51,733 (trainer:779) INFO: 69epoch:train:1-413batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.161, loss_att=3.904, acc=0.974, loss=5.481, backward_time=0.252, grad_norm=68.839, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=5.964e-04, train_time=1.233
+[bmi2:0/4] 2024-07-10 10:12:58,062 (trainer:779) INFO: 69epoch:train:414-826batch: iter_time=2.685e-04, forward_time=0.161, loss_ctc=9.125, loss_att=3.908, acc=0.975, loss=5.473, backward_time=0.254, grad_norm=69.626, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.034, optim0_lr0=5.962e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 10:17:04,501 (trainer:779) INFO: 69epoch:train:827-1239batch: iter_time=7.936e-04, forward_time=0.159, loss_ctc=9.220, loss_att=3.939, acc=0.975, loss=5.524, backward_time=0.255, grad_norm=74.160, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.960e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 10:17:10,034 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 10:21:10,494 (trainer:779) INFO: 69epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.157, loss_ctc=9.138, loss_att=3.925, acc=0.973, loss=5.489, backward_time=0.255, grad_norm=69.991, clip=100.000, loss_scale=2.647e+33, optim_step_time=0.030, optim0_lr0=5.958e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 10:25:14,902 (trainer:779) INFO: 69epoch:train:1653-2065batch: iter_time=8.140e-04, forward_time=0.156, loss_ctc=9.309, loss_att=3.939, acc=0.978, loss=5.550, backward_time=0.255, grad_norm=72.391, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.956e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 10:29:19,448 (trainer:779) INFO: 69epoch:train:2066-2478batch: iter_time=1.885e-04, forward_time=0.156, loss_ctc=9.263, loss_att=3.938, acc=0.976, loss=5.536, backward_time=0.255, grad_norm=73.337, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.953e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 10:33:25,266 (trainer:779) INFO: 69epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.156, loss_ctc=9.239, loss_att=3.930, acc=0.974, loss=5.523, backward_time=0.256, grad_norm=73.573, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.951e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 10:37:32,132 (trainer:779) INFO: 69epoch:train:2892-3304batch: iter_time=1.873e-04, forward_time=0.157, loss_ctc=9.167, loss_att=3.922, acc=0.979, loss=5.495, backward_time=0.256, grad_norm=75.067, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.949e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 10:41:38,072 (trainer:779) INFO: 69epoch:train:3305-3717batch: iter_time=3.306e-04, forward_time=0.156, loss_ctc=9.318, loss_att=3.958, acc=0.979, loss=5.566, backward_time=0.256, grad_norm=75.536, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.947e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 10:45:45,970 (trainer:779) INFO: 69epoch:train:3718-4130batch: iter_time=6.045e-04, forward_time=0.158, loss_ctc=9.298, loss_att=3.965, acc=0.978, loss=5.565, backward_time=0.257, grad_norm=75.732, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.945e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 10:49:51,025 (trainer:779) INFO: 69epoch:train:4131-4543batch: iter_time=8.150e-04, forward_time=0.156, loss_ctc=9.305, loss_att=3.948, acc=0.974, loss=5.555, backward_time=0.255, grad_norm=69.638, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.942e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 10:53:56,163 (trainer:779) INFO: 69epoch:train:4544-4956batch: iter_time=0.001, forward_time=0.156, loss_ctc=9.216, loss_att=3.939, acc=0.975, loss=5.522, backward_time=0.254, grad_norm=73.639, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.940e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 10:58:02,016 (trainer:779) INFO: 69epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.157, loss_ctc=9.237, loss_att=3.941, acc=0.974, loss=5.530, backward_time=0.256, grad_norm=69.829, clip=100.000, loss_scale=3.352e+33, optim_step_time=0.030, optim0_lr0=5.938e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 11:02:07,499 (trainer:779) INFO: 69epoch:train:5370-5782batch: iter_time=1.873e-04, forward_time=0.156, loss_ctc=9.295, loss_att=3.954, acc=0.976, loss=5.556, backward_time=0.255, grad_norm=71.775, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.936e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 11:06:15,201 (trainer:779) INFO: 69epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.158, loss_ctc=9.130, loss_att=3.900, acc=0.977, loss=5.469, backward_time=0.256, grad_norm=69.560, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.934e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 11:10:21,135 (trainer:779) INFO: 69epoch:train:6196-6608batch: iter_time=1.895e-04, forward_time=0.156, loss_ctc=9.316, loss_att=3.949, acc=0.977, loss=5.559, backward_time=0.256, grad_norm=70.981, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.932e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 11:14:27,096 (trainer:779) INFO: 69epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.156, loss_ctc=9.092, loss_att=3.889, acc=0.976, loss=5.450, backward_time=0.255, grad_norm=71.249, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.930e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 11:18:32,311 (trainer:779) INFO: 69epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.157, loss_ctc=9.170, loss_att=3.937, acc=0.975, loss=5.507, backward_time=0.255, grad_norm=67.774, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=5.927e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 11:22:34,107 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 11:22:39,196 (trainer:779) INFO: 69epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.156, loss_ctc=9.270, loss_att=3.972, acc=0.977, loss=5.562, backward_time=0.256, grad_norm=73.315, clip=100.000, loss_scale=5.142e+33, optim_step_time=0.029, optim0_lr0=5.925e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 11:26:44,526 (trainer:779) INFO: 69epoch:train:7848-8260batch: iter_time=6.321e-04, forward_time=0.156, loss_ctc=9.298, loss_att=3.950, acc=0.976, loss=5.554, backward_time=0.255, grad_norm=69.752, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.923e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 11:27:48,000 (trainer:365) INFO: 69epoch results: [train] iter_time=8.634e-04, forward_time=0.157, loss_ctc=9.228, loss_att=3.935, acc=0.976, loss=5.523, backward_time=0.255, grad_norm=71.776, clip=100.000, loss_scale=3.801e+33, optim_step_time=0.030, optim0_lr0=5.944e-04, train_time=1.193, time=1 hour, 22 minutes and 12.41 seconds, total_count=570423, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.880, cer_ctc=0.040, loss_att=5.852, acc=0.950, cer=0.032, wer=0.487, loss=7.060, time=18.25 seconds, total_count=2346, gpu_max_cached_mem_GB=22.529, [att_plot] time=40.12 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 11:27:52,634 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 11:27:52,724 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/53epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/68epoch.pth
+[bmi2:0/4] 2024-07-10 11:27:52,724 (trainer:299) INFO: 70/100epoch started. Estimated time to finish: 1 day, 19 hours and 1 minute
+[bmi2:0/4] 2024-07-10 11:32:11,478 (trainer:779) INFO: 70epoch:train:1-413batch: iter_time=0.002, forward_time=0.158, loss_ctc=9.102, loss_att=3.927, acc=0.976, loss=5.479, backward_time=0.256, grad_norm=72.145, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.921e-04, train_time=1.253
+[bmi2:0/4] 2024-07-10 11:36:16,024 (trainer:779) INFO: 70epoch:train:414-826batch: iter_time=3.716e-04, forward_time=0.156, loss_ctc=9.130, loss_att=3.888, acc=0.977, loss=5.461, backward_time=0.254, grad_norm=72.111, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.919e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 11:40:21,953 (trainer:779) INFO: 70epoch:train:827-1239batch: iter_time=1.941e-04, forward_time=0.158, loss_ctc=9.139, loss_att=3.922, acc=0.976, loss=5.487, backward_time=0.255, grad_norm=71.323, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.917e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 11:44:27,419 (trainer:779) INFO: 70epoch:train:1240-1652batch: iter_time=2.243e-04, forward_time=0.159, loss_ctc=9.181, loss_att=3.913, acc=0.978, loss=5.493, backward_time=0.253, grad_norm=70.201, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.914e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 11:48:32,261 (trainer:779) INFO: 70epoch:train:1653-2065batch: iter_time=2.176e-04, forward_time=0.161, loss_ctc=9.186, loss_att=3.919, acc=0.979, loss=5.499, backward_time=0.253, grad_norm=73.171, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.912e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 11:52:36,870 (trainer:779) INFO: 70epoch:train:2066-2478batch: iter_time=3.288e-04, forward_time=0.159, loss_ctc=9.088, loss_att=3.876, acc=0.975, loss=5.440, backward_time=0.252, grad_norm=71.772, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.031, optim0_lr0=5.910e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 11:56:41,145 (trainer:779) INFO: 70epoch:train:2479-2891batch: iter_time=2.112e-04, forward_time=0.160, loss_ctc=9.144, loss_att=3.890, acc=0.978, loss=5.466, backward_time=0.254, grad_norm=71.463, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.029, optim0_lr0=5.908e-04, train_time=1.183
+[bmi2:0/4] 2024-07-10 12:00:45,670 (trainer:779) INFO: 70epoch:train:2892-3304batch: iter_time=2.122e-04, forward_time=0.160, loss_ctc=9.214, loss_att=3.909, acc=0.977, loss=5.500, backward_time=0.253, grad_norm=76.058, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.030, optim0_lr0=5.906e-04, train_time=1.183
+[bmi2:0/4] 2024-07-10 12:04:50,038 (trainer:779) INFO: 70epoch:train:3305-3717batch: iter_time=2.061e-04, forward_time=0.160, loss_ctc=9.291, loss_att=3.946, acc=0.976, loss=5.550, backward_time=0.252, grad_norm=73.278, clip=100.000, loss_scale=3.504e+33, optim_step_time=0.031, optim0_lr0=5.904e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 12:08:54,721 (trainer:779) INFO: 70epoch:train:3718-4130batch: iter_time=2.267e-04, forward_time=0.160, loss_ctc=9.247, loss_att=3.917, acc=0.976, loss=5.516, backward_time=0.252, grad_norm=72.246, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.902e-04, train_time=1.184
+[bmi2:0/4] 2024-07-10 12:12:57,650 (trainer:779) INFO: 70epoch:train:4131-4543batch: iter_time=6.896e-04, forward_time=0.157, loss_ctc=9.272, loss_att=3.938, acc=0.973, loss=5.538, backward_time=0.252, grad_norm=75.307, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=5.900e-04, train_time=1.177
+[bmi2:0/4] 2024-07-10 12:17:06,724 (trainer:779) INFO: 70epoch:train:4544-4956batch: iter_time=3.214e-04, forward_time=0.163, loss_ctc=9.292, loss_att=3.950, acc=0.977, loss=5.552, backward_time=0.257, grad_norm=75.686, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.897e-04, train_time=1.205
+[bmi2:0/4] 2024-07-10 12:21:12,166 (trainer:779) INFO: 70epoch:train:4957-5369batch: iter_time=2.986e-04, forward_time=0.160, loss_ctc=9.255, loss_att=3.929, acc=0.973, loss=5.527, backward_time=0.254, grad_norm=73.790, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.895e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 12:25:19,686 (trainer:779) INFO: 70epoch:train:5370-5782batch: iter_time=0.001, forward_time=0.160, loss_ctc=9.139, loss_att=3.893, acc=0.973, loss=5.467, backward_time=0.255, grad_norm=77.937, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.893e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 12:29:25,250 (trainer:779) INFO: 70epoch:train:5783-6195batch: iter_time=2.538e-04, forward_time=0.160, loss_ctc=9.243, loss_att=3.922, acc=0.974, loss=5.518, backward_time=0.253, grad_norm=73.255, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.891e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 12:33:32,717 (trainer:779) INFO: 70epoch:train:6196-6608batch: iter_time=6.667e-04, forward_time=0.161, loss_ctc=9.180, loss_att=3.921, acc=0.972, loss=5.499, backward_time=0.255, grad_norm=75.397, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.889e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 12:37:38,025 (trainer:779) INFO: 70epoch:train:6609-7021batch: iter_time=2.404e-04, forward_time=0.159, loss_ctc=9.175, loss_att=3.889, acc=0.978, loss=5.475, backward_time=0.254, grad_norm=73.503, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.887e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 12:41:43,213 (trainer:779) INFO: 70epoch:train:7022-7434batch: iter_time=2.257e-04, forward_time=0.159, loss_ctc=9.269, loss_att=3.966, acc=0.978, loss=5.557, backward_time=0.253, grad_norm=71.010, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.885e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 12:45:30,065 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 12:45:49,457 (trainer:779) INFO: 70epoch:train:7435-7847batch: iter_time=2.344e-04, forward_time=0.160, loss_ctc=9.231, loss_att=3.919, acc=0.976, loss=5.513, backward_time=0.254, grad_norm=67.442, clip=100.000, loss_scale=8.232e+33, optim_step_time=0.032, optim0_lr0=5.883e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 12:49:54,880 (trainer:779) INFO: 70epoch:train:7848-8260batch: iter_time=2.290e-04, forward_time=0.159, loss_ctc=9.384, loss_att=3.985, acc=0.978, loss=5.605, backward_time=0.252, grad_norm=80.252, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.881e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 12:50:53,070 (trainer:365) INFO: 70epoch results: [train] iter_time=4.351e-04, forward_time=0.159, loss_ctc=9.209, loss_att=3.921, acc=0.976, loss=5.507, backward_time=0.254, grad_norm=73.359, clip=100.000, loss_scale=4.221e+33, optim_step_time=0.031, optim0_lr0=5.901e-04, train_time=1.192, time=1 hour, 22 minutes and 7.15 seconds, total_count=578690, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.822, cer_ctc=0.040, loss_att=6.054, acc=0.949, cer=0.032, wer=0.486, loss=7.185, time=18.09 seconds, total_count=2380, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.11 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 12:50:58,029 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 12:50:58,031 (trainer:299) INFO: 71/100epoch started. Estimated time to finish: 1 day, 17 hours and 38 minutes
+[bmi2:0/4] 2024-07-10 12:55:13,761 (trainer:779) INFO: 71epoch:train:1-413batch: iter_time=0.001, forward_time=0.160, loss_ctc=9.233, loss_att=3.905, acc=0.977, loss=5.503, backward_time=0.254, grad_norm=74.899, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.878e-04, train_time=1.239
+[bmi2:0/4] 2024-07-10 12:59:18,834 (trainer:779) INFO: 71epoch:train:414-826batch: iter_time=2.493e-04, forward_time=0.159, loss_ctc=9.200, loss_att=3.894, acc=0.977, loss=5.486, backward_time=0.253, grad_norm=69.896, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.876e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 13:03:23,874 (trainer:779) INFO: 71epoch:train:827-1239batch: iter_time=2.367e-04, forward_time=0.160, loss_ctc=9.202, loss_att=3.898, acc=0.978, loss=5.489, backward_time=0.252, grad_norm=68.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.874e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 13:07:29,778 (trainer:779) INFO: 71epoch:train:1240-1652batch: iter_time=2.323e-04, forward_time=0.160, loss_ctc=9.047, loss_att=3.868, acc=0.977, loss=5.422, backward_time=0.253, grad_norm=70.099, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.872e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 13:11:35,229 (trainer:779) INFO: 71epoch:train:1653-2065batch: iter_time=2.418e-04, forward_time=0.160, loss_ctc=9.061, loss_att=3.846, acc=0.975, loss=5.411, backward_time=0.253, grad_norm=70.817, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.870e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 13:15:40,094 (trainer:779) INFO: 71epoch:train:2066-2478batch: iter_time=3.923e-04, forward_time=0.159, loss_ctc=9.105, loss_att=3.904, acc=0.971, loss=5.465, backward_time=0.252, grad_norm=72.052, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.868e-04, train_time=1.185
+[bmi2:0/4] 2024-07-10 13:19:45,777 (trainer:779) INFO: 71epoch:train:2479-2891batch: iter_time=2.407e-04, forward_time=0.161, loss_ctc=9.099, loss_att=3.868, acc=0.976, loss=5.437, backward_time=0.254, grad_norm=72.898, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.866e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 13:23:52,636 (trainer:779) INFO: 71epoch:train:2892-3304batch: iter_time=2.385e-04, forward_time=0.160, loss_ctc=9.139, loss_att=3.897, acc=0.979, loss=5.470, backward_time=0.254, grad_norm=77.749, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.864e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 13:27:59,106 (trainer:779) INFO: 71epoch:train:3305-3717batch: iter_time=2.407e-04, forward_time=0.160, loss_ctc=9.127, loss_att=3.893, acc=0.974, loss=5.463, backward_time=0.255, grad_norm=68.585, clip=100.000, loss_scale=7.310e+33, optim_step_time=0.033, optim0_lr0=5.862e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 13:30:34,841 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 13:32:05,646 (trainer:779) INFO: 71epoch:train:3718-4130batch: iter_time=2.490e-04, forward_time=0.161, loss_ctc=9.049, loss_att=3.834, acc=0.974, loss=5.399, backward_time=0.253, grad_norm=69.613, clip=100.000, loss_scale=8.469e+33, optim_step_time=0.033, optim0_lr0=5.860e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 13:36:11,800 (trainer:779) INFO: 71epoch:train:4131-4543batch: iter_time=2.755e-04, forward_time=0.161, loss_ctc=9.144, loss_att=3.895, acc=0.976, loss=5.470, backward_time=0.254, grad_norm=72.704, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.858e-04, train_time=1.192
+[bmi2:0/4] 2024-07-10 13:40:17,866 (trainer:779) INFO: 71epoch:train:4544-4956batch: iter_time=2.328e-04, forward_time=0.160, loss_ctc=9.125, loss_att=3.871, acc=0.981, loss=5.448, backward_time=0.254, grad_norm=72.840, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.856e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 13:44:23,438 (trainer:779) INFO: 71epoch:train:4957-5369batch: iter_time=2.369e-04, forward_time=0.160, loss_ctc=9.145, loss_att=3.933, acc=0.977, loss=5.497, backward_time=0.254, grad_norm=81.554, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.853e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 13:48:28,916 (trainer:779) INFO: 71epoch:train:5370-5782batch: iter_time=2.398e-04, forward_time=0.159, loss_ctc=9.087, loss_att=3.869, acc=0.973, loss=5.435, backward_time=0.253, grad_norm=75.396, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.851e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 13:52:35,388 (trainer:779) INFO: 71epoch:train:5783-6195batch: iter_time=2.960e-04, forward_time=0.161, loss_ctc=9.205, loss_att=3.915, acc=0.974, loss=5.502, backward_time=0.254, grad_norm=69.302, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.849e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 13:56:41,823 (trainer:779) INFO: 71epoch:train:6196-6608batch: iter_time=2.353e-04, forward_time=0.160, loss_ctc=9.105, loss_att=3.881, acc=0.977, loss=5.448, backward_time=0.254, grad_norm=75.779, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.847e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 14:00:49,213 (trainer:779) INFO: 71epoch:train:6609-7021batch: iter_time=2.554e-04, forward_time=0.161, loss_ctc=9.079, loss_att=3.889, acc=0.974, loss=5.446, backward_time=0.255, grad_norm=72.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.845e-04, train_time=1.198
+[bmi2:0/4] 2024-07-10 14:04:54,703 (trainer:779) INFO: 71epoch:train:7022-7434batch: iter_time=2.450e-04, forward_time=0.160, loss_ctc=9.096, loss_att=3.856, acc=0.976, loss=5.428, backward_time=0.253, grad_norm=69.999, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.843e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 14:09:00,215 (trainer:779) INFO: 71epoch:train:7435-7847batch: iter_time=2.307e-04, forward_time=0.158, loss_ctc=9.275, loss_att=3.940, acc=0.978, loss=5.541, backward_time=0.253, grad_norm=72.091, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.841e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 14:13:08,884 (trainer:779) INFO: 71epoch:train:7848-8260batch: iter_time=2.540e-04, forward_time=0.162, loss_ctc=9.185, loss_att=3.893, acc=0.979, loss=5.481, backward_time=0.256, grad_norm=73.731, clip=100.000, loss_scale=8.729e+33, optim_step_time=0.035, optim0_lr0=5.839e-04, train_time=1.203
+[bmi2:0/4] 2024-07-10 14:14:09,947 (trainer:365) INFO: 71epoch results: [train] iter_time=2.969e-04, forward_time=0.160, loss_ctc=9.136, loss_att=3.888, acc=0.976, loss=5.462, backward_time=0.254, grad_norm=72.585, clip=100.000, loss_scale=5.642e+33, optim_step_time=0.033, optim0_lr0=5.859e-04, train_time=1.194, time=1 hour, 22 minutes and 16.02 seconds, total_count=586957, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.808, cer_ctc=0.041, loss_att=5.970, acc=0.949, cer=0.032, wer=0.481, loss=7.121, time=18.16 seconds, total_count=2414, gpu_max_cached_mem_GB=22.529, [att_plot] time=37.73 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 14:14:14,350 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 14:14:14,392 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/70epoch.pth
+[bmi2:0/4] 2024-07-10 14:14:14,393 (trainer:299) INFO: 72/100epoch started. Estimated time to finish: 1 day, 16 hours and 14 minutes
+[bmi2:0/4] 2024-07-10 14:15:25,705 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 14:18:30,410 (trainer:779) INFO: 72epoch:train:1-413batch: iter_time=0.001, forward_time=0.161, loss_ctc=9.119, loss_att=3.873, acc=0.978, loss=5.447, backward_time=0.253, grad_norm=77.427, clip=100.000, loss_scale=6.484e+33, optim_step_time=0.033, optim0_lr0=5.837e-04, train_time=1.240
+[bmi2:0/4] 2024-07-10 14:22:36,872 (trainer:779) INFO: 72epoch:train:414-826batch: iter_time=4.750e-04, forward_time=0.162, loss_ctc=8.888, loss_att=3.786, acc=0.974, loss=5.317, backward_time=0.254, grad_norm=66.965, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.835e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 14:26:43,632 (trainer:779) INFO: 72epoch:train:827-1239batch: iter_time=2.449e-04, forward_time=0.163, loss_ctc=9.026, loss_att=3.857, acc=0.977, loss=5.408, backward_time=0.254, grad_norm=77.648, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.833e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 14:30:49,743 (trainer:779) INFO: 72epoch:train:1240-1652batch: iter_time=2.504e-04, forward_time=0.161, loss_ctc=9.120, loss_att=3.875, acc=0.978, loss=5.449, backward_time=0.253, grad_norm=73.186, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.831e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 14:34:56,261 (trainer:779) INFO: 72epoch:train:1653-2065batch: iter_time=2.875e-04, forward_time=0.162, loss_ctc=8.943, loss_att=3.845, acc=0.978, loss=5.374, backward_time=0.254, grad_norm=67.014, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.829e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 14:39:02,725 (trainer:779) INFO: 72epoch:train:2066-2478batch: iter_time=2.571e-04, forward_time=0.161, loss_ctc=9.038, loss_att=3.864, acc=0.977, loss=5.417, backward_time=0.253, grad_norm=76.175, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.827e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 14:43:09,242 (trainer:779) INFO: 72epoch:train:2479-2891batch: iter_time=2.410e-04, forward_time=0.161, loss_ctc=9.090, loss_att=3.882, acc=0.978, loss=5.445, backward_time=0.254, grad_norm=71.222, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.825e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 14:47:15,414 (trainer:779) INFO: 72epoch:train:2892-3304batch: iter_time=2.423e-04, forward_time=0.160, loss_ctc=9.146, loss_att=3.874, acc=0.976, loss=5.455, backward_time=0.253, grad_norm=69.705, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.823e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 14:51:20,885 (trainer:779) INFO: 72epoch:train:3305-3717batch: iter_time=2.830e-04, forward_time=0.162, loss_ctc=9.140, loss_att=3.888, acc=0.976, loss=5.464, backward_time=0.254, grad_norm=73.875, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.821e-04, train_time=1.189
+[bmi2:0/4] 2024-07-10 14:55:30,086 (trainer:779) INFO: 72epoch:train:3718-4130batch: iter_time=2.624e-04, forward_time=0.163, loss_ctc=9.114, loss_att=3.888, acc=0.976, loss=5.456, backward_time=0.257, grad_norm=78.832, clip=100.000, loss_scale=5.518e+33, optim_step_time=0.033, optim0_lr0=5.819e-04, train_time=1.206
+[bmi2:0/4] 2024-07-10 14:59:35,696 (trainer:779) INFO: 72epoch:train:4131-4543batch: iter_time=2.363e-04, forward_time=0.161, loss_ctc=9.136, loss_att=3.887, acc=0.976, loss=5.462, backward_time=0.254, grad_norm=71.961, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.033, optim0_lr0=5.817e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 15:00:42,069 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 15:03:42,965 (trainer:779) INFO: 72epoch:train:4544-4956batch: iter_time=2.454e-04, forward_time=0.162, loss_ctc=9.003, loss_att=3.838, acc=0.976, loss=5.387, backward_time=0.254, grad_norm=73.251, clip=100.000, loss_scale=6.579e+33, optim_step_time=0.033, optim0_lr0=5.815e-04, train_time=1.197
+[bmi2:0/4] 2024-07-10 15:07:47,862 (trainer:779) INFO: 72epoch:train:4957-5369batch: iter_time=4.055e-04, forward_time=0.161, loss_ctc=9.119, loss_att=3.884, acc=0.972, loss=5.455, backward_time=0.253, grad_norm=69.300, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.813e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 15:11:54,428 (trainer:779) INFO: 72epoch:train:5370-5782batch: iter_time=2.436e-04, forward_time=0.160, loss_ctc=9.116, loss_att=3.883, acc=0.979, loss=5.453, backward_time=0.254, grad_norm=77.157, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.810e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 15:16:01,237 (trainer:779) INFO: 72epoch:train:5783-6195batch: iter_time=2.310e-04, forward_time=0.161, loss_ctc=9.008, loss_att=3.839, acc=0.978, loss=5.390, backward_time=0.254, grad_norm=69.959, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.808e-04, train_time=1.195
+[bmi2:0/4] 2024-07-10 15:20:09,233 (trainer:779) INFO: 72epoch:train:6196-6608batch: iter_time=2.542e-04, forward_time=0.161, loss_ctc=9.056, loss_att=3.841, acc=0.976, loss=5.405, backward_time=0.255, grad_norm=69.069, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.806e-04, train_time=1.200
+[bmi2:0/4] 2024-07-10 15:24:14,853 (trainer:779) INFO: 72epoch:train:6609-7021batch: iter_time=2.344e-04, forward_time=0.161, loss_ctc=9.131, loss_att=3.902, acc=0.977, loss=5.471, backward_time=0.254, grad_norm=75.378, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.804e-04, train_time=1.190
+[bmi2:0/4] 2024-07-10 15:28:21,542 (trainer:779) INFO: 72epoch:train:7022-7434batch: iter_time=5.884e-04, forward_time=0.160, loss_ctc=9.000, loss_att=3.847, acc=0.976, loss=5.393, backward_time=0.254, grad_norm=69.973, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.802e-04, train_time=1.194
+[bmi2:0/4] 2024-07-10 15:32:26,517 (trainer:779) INFO: 72epoch:train:7435-7847batch: iter_time=2.825e-04, forward_time=0.159, loss_ctc=9.054, loss_att=3.857, acc=0.976, loss=5.416, backward_time=0.254, grad_norm=74.323, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.029, optim0_lr0=5.800e-04, train_time=1.187
+[bmi2:0/4] 2024-07-10 15:36:30,624 (trainer:779) INFO: 72epoch:train:7848-8260batch: iter_time=6.088e-04, forward_time=0.159, loss_ctc=9.169, loss_att=3.897, acc=0.975, loss=5.479, backward_time=0.253, grad_norm=72.423, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.798e-04, train_time=1.181
+[bmi2:0/4] 2024-07-10 15:37:29,173 (trainer:365) INFO: 72epoch results: [train] iter_time=3.596e-04, forward_time=0.161, loss_ctc=9.071, loss_att=3.865, acc=0.976, loss=5.427, backward_time=0.254, grad_norm=72.739, clip=100.000, loss_scale=5.601e+33, optim_step_time=0.032, optim0_lr0=5.818e-04, train_time=1.195, time=1 hour, 22 minutes and 21.1 seconds, total_count=595224, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.893, cer_ctc=0.040, loss_att=5.751, acc=0.951, cer=0.032, wer=0.484, loss=6.994, time=18.53 seconds, total_count=2448, gpu_max_cached_mem_GB=22.529, [att_plot] time=35.15 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 15:37:33,890 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 15:37:34,009 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/69epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/71epoch.pth
+[bmi2:0/4] 2024-07-10 15:37:34,010 (trainer:299) INFO: 73/100epoch started. Estimated time to finish: 1 day, 14 hours and 51 minutes
+[bmi2:0/4] 2024-07-10 15:41:49,644 (trainer:779) INFO: 73epoch:train:1-413batch: iter_time=0.001, forward_time=0.160, loss_ctc=8.992, loss_att=3.836, acc=0.977, loss=5.383, backward_time=0.254, grad_norm=70.134, clip=100.000, loss_scale=5.495e+33, optim_step_time=0.030, optim0_lr0=5.796e-04, train_time=1.238
+[bmi2:0/4] 2024-07-10 15:42:51,905 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 15:45:55,665 (trainer:779) INFO: 73epoch:train:414-826batch: iter_time=7.431e-04, forward_time=0.161, loss_ctc=8.881, loss_att=3.776, acc=0.975, loss=5.308, backward_time=0.255, grad_norm=76.559, clip=100.000, loss_scale=6.503e+33, optim_step_time=0.031, optim0_lr0=5.794e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 15:50:06,446 (trainer:779) INFO: 73epoch:train:827-1239batch: iter_time=0.001, forward_time=0.169, loss_ctc=8.975, loss_att=3.828, acc=0.974, loss=5.372, backward_time=0.252, grad_norm=75.825, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.792e-04, train_time=1.215
+[bmi2:0/4] 2024-07-10 15:52:37,752 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 15:54:34,737 (trainer:779) INFO: 73epoch:train:1240-1652batch: iter_time=6.414e-04, forward_time=0.186, loss_ctc=9.079, loss_att=3.847, acc=0.979, loss=5.417, backward_time=0.256, grad_norm=76.193, clip=100.000, loss_scale=4.045e+33, optim_step_time=0.044, optim0_lr0=5.790e-04, train_time=1.298
+[bmi2:0/4] 2024-07-10 15:59:19,038 (trainer:779) INFO: 73epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.204, loss_ctc=9.111, loss_att=3.874, acc=0.977, loss=5.445, backward_time=0.259, grad_norm=75.468, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.051, optim0_lr0=5.788e-04, train_time=1.377
+[bmi2:0/4] 2024-07-10 16:04:04,061 (trainer:779) INFO: 73epoch:train:2066-2478batch: iter_time=6.784e-04, forward_time=0.203, loss_ctc=9.109, loss_att=3.856, acc=0.979, loss=5.432, backward_time=0.260, grad_norm=69.906, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.049, optim0_lr0=5.786e-04, train_time=1.379
+[bmi2:0/4] 2024-07-10 16:08:20,104 (trainer:779) INFO: 73epoch:train:2479-2891batch: iter_time=4.700e-04, forward_time=0.175, loss_ctc=9.141, loss_att=3.891, acc=0.977, loss=5.466, backward_time=0.256, grad_norm=75.009, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.784e-04, train_time=1.240
+[bmi2:0/4] 2024-07-10 16:12:25,256 (trainer:779) INFO: 73epoch:train:2892-3304batch: iter_time=2.247e-04, forward_time=0.160, loss_ctc=9.028, loss_att=3.820, acc=0.977, loss=5.383, backward_time=0.253, grad_norm=69.548, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=5.782e-04, train_time=1.186
+[bmi2:0/4] 2024-07-10 16:16:31,212 (trainer:779) INFO: 73epoch:train:3305-3717batch: iter_time=4.330e-04, forward_time=0.163, loss_ctc=9.176, loss_att=3.881, acc=0.976, loss=5.469, backward_time=0.254, grad_norm=71.813, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.032, optim0_lr0=5.780e-04, train_time=1.191
+[bmi2:0/4] 2024-07-10 16:21:05,387 (trainer:779) INFO: 73epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.194, loss_ctc=9.083, loss_att=3.849, acc=0.978, loss=5.420, backward_time=0.256, grad_norm=77.989, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.044, optim0_lr0=5.778e-04, train_time=1.327
+[bmi2:0/4] 2024-07-10 16:26:04,218 (trainer:779) INFO: 73epoch:train:4131-4543batch: iter_time=9.746e-04, forward_time=0.221, loss_ctc=9.106, loss_att=3.840, acc=0.975, loss=5.420, backward_time=0.266, grad_norm=74.125, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.055, optim0_lr0=5.776e-04, train_time=1.447
+[bmi2:0/4] 2024-07-10 16:31:20,625 (trainer:779) INFO: 73epoch:train:4544-4956batch: iter_time=9.540e-04, forward_time=0.244, loss_ctc=9.081, loss_att=3.865, acc=0.969, loss=5.429, backward_time=0.276, grad_norm=72.160, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.060, optim0_lr0=5.774e-04, train_time=1.531
+[bmi2:0/4] 2024-07-10 16:36:10,180 (trainer:779) INFO: 73epoch:train:4957-5369batch: iter_time=6.166e-04, forward_time=0.209, loss_ctc=9.095, loss_att=3.843, acc=0.979, loss=5.418, backward_time=0.271, grad_norm=76.467, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.048, optim0_lr0=5.772e-04, train_time=1.403
+[bmi2:0/4] 2024-07-10 16:40:17,476 (trainer:779) INFO: 73epoch:train:5370-5782batch: iter_time=2.553e-04, forward_time=0.161, loss_ctc=9.118, loss_att=3.897, acc=0.979, loss=5.464, backward_time=0.254, grad_norm=73.041, clip=100.000, loss_scale=4.553e+33, optim_step_time=0.032, optim0_lr0=5.770e-04, train_time=1.197
+[bmi2:0/4] 2024-07-10 16:44:22,728 (trainer:779) INFO: 73epoch:train:5783-6195batch: iter_time=2.810e-04, forward_time=0.161, loss_ctc=8.998, loss_att=3.815, acc=0.975, loss=5.370, backward_time=0.253, grad_norm=71.442, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.031, optim0_lr0=5.768e-04, train_time=1.188
+[bmi2:0/4] 2024-07-10 16:48:29,169 (trainer:779) INFO: 73epoch:train:6196-6608batch: iter_time=2.495e-04, forward_time=0.161, loss_ctc=9.149, loss_att=3.884, acc=0.977, loss=5.463, backward_time=0.254, grad_norm=76.819, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.766e-04, train_time=1.193
+[bmi2:0/4] 2024-07-10 16:52:32,897 (trainer:779) INFO: 73epoch:train:6609-7021batch: iter_time=2.463e-04, forward_time=0.157, loss_ctc=9.079, loss_att=3.844, acc=0.975, loss=5.415, backward_time=0.252, grad_norm=73.766, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.030, optim0_lr0=5.764e-04, train_time=1.180
+[bmi2:0/4] 2024-07-10 16:56:41,500 (trainer:779) INFO: 73epoch:train:7022-7434batch: iter_time=2.715e-04, forward_time=0.161, loss_ctc=9.087, loss_att=3.849, acc=0.976, loss=5.420, backward_time=0.257, grad_norm=68.447, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.032, optim0_lr0=5.762e-04, train_time=1.203
+[bmi2:0/4] 2024-07-10 17:00:48,265 (trainer:779) INFO: 73epoch:train:7435-7847batch: iter_time=3.027e-04, forward_time=0.162, loss_ctc=8.934, loss_att=3.827, acc=0.977, loss=5.359, backward_time=0.254, grad_norm=70.854, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.033, optim0_lr0=5.760e-04, train_time=1.196
+[bmi2:0/4] 2024-07-10 17:04:58,907 (trainer:779) INFO: 73epoch:train:7848-8260batch: iter_time=3.511e-04, forward_time=0.167, loss_ctc=9.058, loss_att=3.851, acc=0.979, loss=5.413, backward_time=0.254, grad_norm=71.356, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.037, optim0_lr0=5.759e-04, train_time=1.213
+[bmi2:0/4] 2024-07-10 17:06:15,338 (trainer:365) INFO: 73epoch results: [train] iter_time=6.209e-04, forward_time=0.179, loss_ctc=9.063, loss_att=3.848, acc=0.976, loss=5.413, backward_time=0.257, grad_norm=73.334, clip=100.000, loss_scale=4.016e+33, optim_step_time=0.039, optim0_lr0=5.777e-04, train_time=1.270, time=1 hour, 27 minutes and 30.23 seconds, total_count=603491, gpu_max_cached_mem_GB=22.529, [valid] loss_ctc=9.977, cer_ctc=0.041, loss_att=5.901, acc=0.950, cer=0.033, wer=0.484, loss=7.124, time=21.84 seconds, total_count=2482, gpu_max_cached_mem_GB=22.529, [att_plot] time=49.25 seconds, total_count=0, gpu_max_cached_mem_GB=22.529
+[bmi2:0/4] 2024-07-10 17:06:22,080 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-10 17:06:22,093 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/39epoch.pth
+[bmi2:0/4] 2024-07-10 17:06:22,093 (trainer:299) INFO: 74/100epoch started. Estimated time to finish: 1 day, 13 hours and 30 minutes
+[bmi2:0/4] 2024-07-10 17:10:46,155 (trainer:779) INFO: 74epoch:train:1-413batch: iter_time=0.002, forward_time=0.171, loss_ctc=8.916, loss_att=3.801, acc=0.979, loss=5.336, backward_time=0.253, grad_norm=71.844, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.039, optim0_lr0=5.757e-04, train_time=1.279
+[bmi2:0/4] 2024-07-10 17:15:01,190 (trainer:779) INFO: 74epoch:train:414-826batch: iter_time=0.001, forward_time=0.174, loss_ctc=8.884, loss_att=3.792, acc=0.975, loss=5.320, backward_time=0.254, grad_norm=84.087, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.041, optim0_lr0=5.755e-04, train_time=1.234
+[bmi2:0/4] 2024-07-10 17:19:14,159 (trainer:779) INFO: 74epoch:train:827-1239batch: iter_time=3.561e-04, forward_time=0.174, loss_ctc=9.004, loss_att=3.811, acc=0.979, loss=5.369, backward_time=0.255, grad_norm=71.415, clip=100.000, loss_scale=5.621e+33, optim_step_time=0.041, optim0_lr0=5.753e-04, train_time=1.225
+[bmi2:0/4] 2024-07-10 17:22:30,641 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-10 17:23:25,234 (trainer:779) INFO: 74epoch:train:1240-1652batch: iter_time=3.402e-04, forward_time=0.172, loss_ctc=8.997, loss_att=3.821, acc=0.976, loss=5.374, backward_time=0.253, grad_norm=75.624, clip=100.000, loss_scale=9.250e+33, optim_step_time=0.041, optim0_lr0=5.751e-04, train_time=1.215
+[bmi2:0/4] 2024-07-10 17:27:36,035 (trainer:779) INFO: 74epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.174, loss_ctc=8.963, loss_att=3.802, acc=0.976, loss=5.350, backward_time=0.253, grad_norm=70.238, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.041, optim0_lr0=5.749e-04, train_time=1.215
+[bmi2:0/4] 2024-07-10 17:31:47,340 (trainer:779) INFO: 74epoch:train:2066-2478batch: iter_time=7.755e-04, forward_time=0.173, loss_ctc=8.967, loss_att=3.814, acc=0.977, loss=5.360, backward_time=0.254, grad_norm=76.497, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.041, optim0_lr0=5.747e-04, train_time=1.216
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/runpy.py", line 194, in _run_module_as_main
+    return _run_code(code, main_globals, None,
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/runpy.py", line 87, in _run_code
+    exec(code, run_globals)
+  File "/home/zysong/espnet/espnet2/bin/asr_train.py", line 23, in <module>
+    main()
+  File "/home/zysong/espnet/espnet2/bin/asr_train.py", line 19, in main
+    ASRTask.main(cmd=cmd)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1219, in main
+    while not ProcessContext(processes, error_queues).join():
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/multiprocessing/spawn.py", line 109, in join
+    ready = multiprocessing.connection.wait(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/connection.py", line 931, in wait
+    ready = selector.select(timeout)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/selectors.py", line 415, in select
+    fd_event_list = self._selector.poll(timeout)
+KeyboardInterrupt
+Process SpawnProcess-1:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 700, in train_one_epoch
+    grad_norm = torch.nn.utils.clip_grad_norm_(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in clip_grad_norm_
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in <listcomp>
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/functional.py", line 1485, in norm
+    return _VF.norm(input, p, dim=_dim, keepdim=keepdim)  # type: ignore[attr-defined]
+KeyboardInterrupt
+f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO Channel 03 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:3524061:3530899 [0] NCCL INFO Connected all rings
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524061:3530899 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:3524061:3530899 [0] NCCL INFO Connected all trees
+bmi2:3524061:3530899 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:3524061:3530899 [0] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:3524061:3530899 [0] NCCL INFO comm 0xa10bec0 rank 0 nranks 4 cudaDev 0 busId 4f000 - Init COMPLETE
+Process SpawnProcess-2:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 700, in train_one_epoch
+    grad_norm = torch.nn.utils.clip_grad_norm_(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in clip_grad_norm_
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in <listcomp>
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/functional.py", line 1485, in norm
+    return _VF.norm(input, p, dim=_dim, keepdim=keepdim)  # type: ignore[attr-defined]
+KeyboardInterrupt
+064:3530922 [1] NCCL INFO Channel 03 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO Connected all rings
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 00 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 01 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 02 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524064:3530922 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:3524064:3530922 [1] NCCL INFO Channel 03 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524064:3530922 [1] NCCL INFO Connected all trees
+bmi2:3524064:3530922 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:3524064:3530922 [1] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:3524064:3530922 [1] NCCL INFO comm 0x9073ef0 rank 1 nranks 4 cudaDev 1 busId 52000 - Init COMPLETE
+Process SpawnProcess-3:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 700, in train_one_epoch
+    grad_norm = torch.nn.utils.clip_grad_norm_(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in clip_grad_norm_
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in <listcomp>
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/functional.py", line 1485, in norm
+    return _VF.norm(input, p, dim=_dim, keepdim=keepdim)  # type: ignore[attr-defined]
+KeyboardInterrupt
+071:3530901 [2] NCCL INFO Channel 03 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO Connected all rings
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 00 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 01 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 02 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524071:3530901 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:3524071:3530901 [2] NCCL INFO Channel 03 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:3524071:3530901 [2] NCCL INFO Connected all trees
+bmi2:3524071:3530901 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:3524071:3530901 [2] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:3524071:3530901 [2] NCCL INFO comm 0x9046c60 rank 2 nranks 4 cudaDev 2 busId 56000 - Init COMPLETE
+Process SpawnProcess-4:
+Traceback (most recent call last):
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
+    self.run()
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/multiprocessing/process.py", line 108, in run
+    self._target(*self._args, **self._kwargs)
+  File "/home/zysong/espnet/espnet2/tasks/abs_task.py", line 1486, in main_worker
+    cls.trainer.run(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 317, in run
+    all_steps_are_invalid = cls.train_one_epoch(
+  File "/home/zysong/espnet/espnet2/train/trainer.py", line 700, in train_one_epoch
+    grad_norm = torch.nn.utils.clip_grad_norm_(
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in clip_grad_norm_
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/nn/utils/clip_grad.py", line 43, in <listcomp>
+    total_norm = torch.norm(torch.stack([torch.norm(g.detach(), norm_type).to(device) for g in grads]), norm_type)
+  File "/home/zysong/espnet/tools/miniconda/envs/espnet/lib/python3.8/site-packages/torch/functional.py", line 1485, in norm
+    return _VF.norm(input, p, dim=_dim, keepdim=keepdim)  # type: ignore[attr-defined]
+KeyboardInterrupt
+3524072:3530900 [3] NCCL INFO Channel 03 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO Connected all rings
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 00 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 01 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 02 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:3524072:3530900 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:3524072:3530900 [3] NCCL INFO Channel 03 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:3524072:3530900 [3] NCCL INFO Connected all trees
+bmi2:3524072:3530900 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:3524072:3530900 [3] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:3524072:3530900 [3] NCCL INFO comm 0x9786fd0 rank 3 nranks 4 cudaDev 3 busId 57000 - Init COMPLETE
+bmi2:3524071:3530943 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+bmi2:3524071:3524071 [2] NCCL INFO comm 0x9046c60 rank 2 nranks 4 cudaDev 2 busId 56000 - Abort COMPLETE
+bmi2:3524072:3530941 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+bmi2:3524072:3524072 [3] NCCL INFO comm 0x9786fd0 rank 3 nranks 4 cudaDev 3 busId 57000 - Abort COMPLETE
+bmi2:3524064:3530942 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+bmi2:3524064:3524064 [1] NCCL INFO comm 0x9073ef0 rank 1 nranks 4 cudaDev 1 busId 52000 - Abort COMPLETE
+bmi2:3524061:3530944 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+bmi2:3524061:3524061 [0] NCCL INFO comm 0xa10bec0 rank 0 nranks 4 cudaDev 0 busId 4f000 - Abort COMPLETE
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.log b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.log
new file mode 100644
index 0000000000000000000000000000000000000000..de23d04657b39382696e5912da2c473cc225ffe1
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/train.log
@@ -0,0 +1,2057 @@
+# python3 -m espnet2.bin.asr_train --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True 
+# Started at Tue Jul 16 20:48:40 HKT 2024
+#
+/home/zysong/espnet/tools/miniconda/envs/espnet/bin/python3 /home/zysong/espnet/espnet2/bin/asr_train.py --use_preprocessor true --bpemodel data/en_token_list/bpe_unigram5000/bpe.model --token_type bpe --token_list data/en_token_list/bpe_unigram5000/tokens.txt --non_linguistic_symbols none --cleaner none --g2p none --valid_data_path_and_name_and_type dump/raw/dev/wav.scp,speech,sound --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape --resume true --ignore_init_mismatch false --fold_length 80000 --output_dir exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7 --config conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml --frontend_conf fs=16k --normalize=global_mvn --normalize_conf stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz --train_data_path_and_name_and_type dump/raw/train_960_sp/wav.scp,speech,sound --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/speech_shape --fold_length 150 --train_data_path_and_name_and_type dump/raw/train_960_sp/text,text,text --train_shape_file exp/asr_stats_raw_en_bpe5000_sp/train/text_shape.bpe --valid_data_path_and_name_and_type dump/raw/dev/text,text,text --valid_shape_file exp/asr_stats_raw_en_bpe5000_sp/valid/text_shape.bpe --ngpu 4 --multiprocessing_distributed True
+[bmi2:0/4] 2024-07-16 20:48:55,407 (distributed_c10d:319) INFO: Added key: store_based_barrier_key:1 to store for rank: 0
+[bmi2:0/4] 2024-07-16 20:48:55,408 (distributed_c10d:353) INFO: Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes.
+[bmi2:0/4] 2024-07-16 20:48:55,417 (asr:523) INFO: Vocabulary size: 5000
+[bmi2:0/4] 2024-07-16 20:48:56,997 (initialize:88) INFO: Initialize encoder.embed.conv.0.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.embed.conv.2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.embed.out.0.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,998 (initialize:88) INFO: Initialize encoder.encoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.6.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.6.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.6.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.6.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.7.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.7.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.7.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.7.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.8.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.8.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.8.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:56,999 (initialize:88) INFO: Initialize encoder.encoders.8.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.9.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.9.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.9.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.9.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.10.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.10.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.10.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.10.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.11.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.11.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.11.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.11.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.12.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.12.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.12.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.12.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.13.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.13.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.13.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.13.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,000 (initialize:88) INFO: Initialize encoder.encoders.14.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.14.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.14.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.14.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.15.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.15.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.15.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.15.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.16.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.16.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.16.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.16.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.17.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.17.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.17.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.encoders.17.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize encoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.after_norm.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.output_layer.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,001 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.0.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.1.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,002 (initialize:88) INFO: Initialize decoder.decoders.2.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.2.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.2.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.2.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.3.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.4.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,003 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.self_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_q.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_k.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_v.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.src_attn.linear_out.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.feed_forward.w_2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.norm1.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.norm2.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize decoder.decoders.5.norm3.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:57,004 (initialize:88) INFO: Initialize ctc.ctc_lo.bias to zeros
+[bmi2:0/4] 2024-07-16 20:48:58,177 (abs_task:1320) INFO: pytorch.version=1.13.1+cu117, cuda.available=True, cudnn.version=8500, cudnn.benchmark=False, cudnn.deterministic=True
+[bmi2:0/4] 2024-07-16 20:48:58,183 (abs_task:1321) INFO: Model structure:
+ESPnetASRModel(
+  (frontend): DefaultFrontend(
+    (stft): Stft(n_fft=512, win_length=512, hop_length=128, center=True, normalized=False, onesided=True)
+    (frontend): Frontend()
+    (logmel): LogMel(sr=16000, n_fft=512, n_mels=80, fmin=0, fmax=8000.0, htk=False)
+  )
+  (specaug): SpecAug(
+    (time_warp): TimeWarp(window=5, mode=bicubic)
+    (freq_mask): MaskAlongAxis(mask_width_range=[0, 30], num_mask=2, axis=freq)
+    (time_mask): MaskAlongAxis(mask_width_range=[0, 40], num_mask=2, axis=time)
+  )
+  (normalize): GlobalMVN(stats_file=exp/asr_stats_raw_en_bpe5000_sp/train/feats_stats.npz, norm_means=True, norm_vars=True)
+  (encoder): Q_TransformerEncoder(
+    (embed): Conv2dSubsampling6(
+      (conv): Sequential(
+        (0): Conv2d(1, 512, kernel_size=(3, 3), stride=(2, 2))
+        (1): ReLU()
+        (2): Conv2d(512, 512, kernel_size=(5, 5), stride=(3, 3))
+        (3): ReLU()
+      )
+      (out): Sequential(
+        (0): Linear(in_features=6144, out_features=512, bias=True)
+        (1): PositionalEncoding(
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+      )
+    )
+    (encoders): MultiSequential(
+      (0): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (1): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (2): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (3): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (4): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (5): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (6): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (7): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (8): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (9): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (10): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (11): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (12): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (13): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (14): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (15): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (16): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+      (17): Q_Transformer_EncoderLayer(
+        (self_attn): Q_MultiHeadedAttention_HierDecay_v2(
+          (linear_q): Linear(in_features=512, out_features=512, bias=False)
+          (linear_k): Linear(in_features=512, out_features=512, bias=False)
+          (linear_v): Linear(in_features=512, out_features=512, bias=False)
+          (q_sn): MultiSpike()
+          (k_sn): MultiSpike()
+          (v_sn): MultiSpike()
+          (output_sn): MultiSpike()
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): Q_PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): MultiSpike()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+        (ATT_sn): MultiSpike()
+        (FFN_sn): MultiSpike()
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+  )
+  (decoder): TransformerDecoder(
+    (embed): Sequential(
+      (0): Embedding(5000, 512)
+      (1): PositionalEncoding(
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+    (after_norm): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+    (output_layer): Linear(in_features=512, out_features=5000, bias=True)
+    (decoders): MultiSequential(
+      (0): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (1): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (2): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (3): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (4): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+      (5): DecoderLayer(
+        (self_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (src_attn): MultiHeadedAttention(
+          (linear_q): Linear(in_features=512, out_features=512, bias=True)
+          (linear_k): Linear(in_features=512, out_features=512, bias=True)
+          (linear_v): Linear(in_features=512, out_features=512, bias=True)
+          (linear_out): Linear(in_features=512, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+        )
+        (feed_forward): PositionwiseFeedForward(
+          (w_1): Linear(in_features=512, out_features=2048, bias=True)
+          (w_2): Linear(in_features=2048, out_features=512, bias=True)
+          (dropout): Dropout(p=0.1, inplace=False)
+          (activation): ReLU()
+        )
+        (norm1): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm2): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (norm3): LayerNorm((512,), eps=1e-12, elementwise_affine=True)
+        (dropout): Dropout(p=0.1, inplace=False)
+      )
+    )
+  )
+  (criterion_att): LabelSmoothingLoss(
+    (criterion): KLDivLoss()
+  )
+  (ctc): CTC(
+    (ctc_lo): Linear(in_features=512, out_features=5000, bias=True)
+    (ctc_loss): CTCLoss()
+  )
+)
+
+Model summary:
+    Class Name: ESPnetASRModel
+    Total Number of model parameters: 99.34 M
+    Number of trainable parameters: 99.34 M (100.0%)
+    Size: 397.35 MB
+    Type: torch.float32
+[bmi2:0/4] 2024-07-16 20:48:58,184 (abs_task:1324) INFO: Optimizer:
+Adam (
+Parameter Group 0
+    amsgrad: False
+    betas: (0.9, 0.999)
+    capturable: False
+    differentiable: False
+    eps: 1e-08
+    foreach: None
+    fused: False
+    initial_lr: 0.002
+    lr: 8e-08
+    maximize: False
+    weight_decay: 0
+)
+[bmi2:0/4] 2024-07-16 20:48:58,184 (abs_task:1325) INFO: Scheduler: WarmupLR(warmup_steps=25000)
+[bmi2:0/4] 2024-07-16 20:48:58,197 (abs_task:1334) INFO: Saving the configuration in exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/config.yaml
+[bmi2:0/4] 2024-07-16 20:48:59,779 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:49:06,719 (abs_task:1714) INFO: [train] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/train_960_sp/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/train_960_sp/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f70267afac0>)
+[bmi2:0/4] 2024-07-16 20:49:06,719 (abs_task:1715) INFO: [train] Batch sampler: NumElementsBatchSampler(N-batch=8267, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-16 20:49:06,720 (abs_task:1716) INFO: [train] mini-batch sizes summary: N-batch=8267, mean=102.1, min=41, max=683
+[bmi2:0/4] 2024-07-16 20:49:06,763 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:49:06,790 (abs_task:1714) INFO: [valid] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f70264012b0>)
+[bmi2:0/4] 2024-07-16 20:49:06,790 (abs_task:1715) INFO: [valid] Batch sampler: NumElementsBatchSampler(N-batch=34, batch_bins=45000000, sort_in_batch=descending, sort_batch=descending)
+[bmi2:0/4] 2024-07-16 20:49:06,790 (abs_task:1716) INFO: [valid] mini-batch sizes summary: N-batch=34, mean=163.3, min=7, max=445
+[bmi2:0/4] 2024-07-16 20:49:06,797 (asr:495) INFO: Optional Data Names: ('text_spk2', 'text_spk3', 'text_spk4', 'prompt')
+[bmi2:0/4] 2024-07-16 20:49:06,802 (abs_task:1714) INFO: [plot_att] dataset:
+ESPnetDataset(
+  speech: {"path": "dump/raw/dev/wav.scp", "type": "sound"}
+  text: {"path": "dump/raw/dev/text", "type": "text"}
+  preprocess: <espnet2.train.preprocessor.CommonPreprocessor object at 0x7f7026401a90>)
+[bmi2:0/4] 2024-07-16 20:49:06,802 (abs_task:1715) INFO: [plot_att] Batch sampler: UnsortedBatchSampler(N-batch=5551, batch_size=1, key_file=exp/asr_stats_raw_en_bpe5000_sp/valid/speech_shape, 
+[bmi2:0/4] 2024-07-16 20:49:06,802 (abs_task:1716) INFO: [plot_att] mini-batch sizes summary: N-batch=3, mean=1.0, min=1, max=1
+[bmi2:0/4] 2024-07-16 20:49:07,645 (trainer:174) INFO: The training was resumed using exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/checkpoint.pth
+bmi2:1856035:1856035 [0] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1856035:1856035 [0] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1856035:1856035 [0] NCCL INFO cudaDriverVersion 12040
+NCCL version 2.14.3+cuda11.7
+bmi2:1856044:1856044 [1] NCCL INFO cudaDriverVersion 12040
+bmi2:1856044:1856044 [1] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1856044:1856044 [1] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1856044:1859957 [1] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1856044:1859957 [1] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1856044:1859957 [1] NCCL INFO Using network Socket
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) abmi2:1856046:1856046 [3] NCCL INFO cudaDriverVersion 12040
+bmi2:1856046:1856046 [3] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1856046:1856046 [3] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1856046:1859959 [3] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1856046:1859959 [3] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1856046:1859959 [3] NCCL INFO Using network Socket
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) abmi2:1856035:1859955 [0] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1856035:1859955 [0] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1856035:1859955 [0] NCCL INFO Using network Socket
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) andbmi2:1856045:1856045 [2] NCCL INFO cudaDriverVersion 12040
+bmi2:1856045:1856045 [2] NCCL INFO Bootstrap : Using eno1:10.21.4.69<0>
+bmi2:1856045:1856045 [2] NCCL INFO NET/Plugin : No plugin found (libnccl-net.so), using internal implementation
+bmi2:1856045:1859958 [2] NCCL INFO Failed to open libibverbs.so[.1]
+bmi2:1856045:1859958 [2] NCCL INFO NET/Socket : Using [0]eno1:10.21.4.69<0> [1]usb0:169.254.3.1<0> [2]br-3b06e69e1a27:172.18.0.1<0> [3]tailscale0:100.65.56.18<0> [4]veth1e33792:fe80::f89c:ebff:feac:f6d6%veth1e33792<0>
+bmi2:1856045:1859958 [2] NCCL INFO Using network Socket
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856044:1859957 [1] NCCL INFO Setting affinity for GPU 1 to ffffffff,00000000,ffffffff
+bmi2:1856044:1859957 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 00 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 01 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 02 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFnd dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856046:1859959 [3] NCCL INFO Setting affinity for GPU 3 to ffffffff,00000000,ffffffff
+bmi2:1856046:1859959 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 00 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 01 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 02 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 0(=4f000)
+bmi2:1856046:1859959 [3] NCCLnd dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO Setting affinity for GPU 2 to ffffffff,00000000,ffffffff
+bmi2:1856045:1859958 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 00 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 01 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 02 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INF dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 1 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO Setting affinity for GPU 0 to ffffffff,00000000,ffffffff
+bmi2:1856035:1859955 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+bmi2:1856035:1859955 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+bmi2:1856035:1859955 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+bmi2:1856035:1859955 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+bmi2:1856035:1859955 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 3(=57000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO Channel 00 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO Channel 01 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO Channel 02 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+[bmi2:0/4] 2024-07-16 20:49:08,935 (trainer:311) INFO: 74/100epoch started
+[bmi2:0/4] 2024-07-16 20:49:20,376 (distributed:1027) INFO: Reducer buckets have been rebuilt in this iteration.
+[bmi2:0/4] 2024-07-16 20:53:21,898 (trainer:779) INFO: 74epoch:train:1-413batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.938, loss_att=3.802, acc=0.979, loss=5.343, backward_time=0.255, grad_norm=69.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.757e-04, train_time=1.225
+[bmi2:0/4] 2024-07-16 20:57:22,038 (trainer:779) INFO: 74epoch:train:414-826batch: iter_time=8.446e-04, forward_time=0.155, loss_ctc=8.900, loss_att=3.786, acc=0.975, loss=5.320, backward_time=0.254, grad_norm=72.489, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.755e-04, train_time=1.162
+[bmi2:0/4] 2024-07-16 21:01:24,113 (trainer:779) INFO: 74epoch:train:827-1239batch: iter_time=3.029e-04, forward_time=0.156, loss_ctc=9.057, loss_att=3.832, acc=0.979, loss=5.399, backward_time=0.256, grad_norm=71.410, clip=100.000, loss_scale=5.621e+33, optim_step_time=0.036, optim0_lr0=5.753e-04, train_time=1.173
+[bmi2:0/4] 2024-07-16 21:01:55,189 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 21:05:26,183 (trainer:779) INFO: 74epoch:train:1240-1652batch: iter_time=6.949e-04, forward_time=0.156, loss_ctc=9.006, loss_att=3.827, acc=0.976, loss=5.381, backward_time=0.255, grad_norm=77.391, clip=100.000, loss_scale=5.848e+33, optim_step_time=0.036, optim0_lr0=5.751e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 21:09:27,185 (trainer:779) INFO: 74epoch:train:1653-2065batch: iter_time=6.111e-04, forward_time=0.155, loss_ctc=8.957, loss_att=3.800, acc=0.976, loss=5.347, backward_time=0.255, grad_norm=75.296, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.749e-04, train_time=1.167
+[bmi2:0/4] 2024-07-16 21:13:29,190 (trainer:779) INFO: 74epoch:train:2066-2478batch: iter_time=4.105e-04, forward_time=0.157, loss_ctc=9.001, loss_att=3.806, acc=0.977, loss=5.365, backward_time=0.256, grad_norm=78.475, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.747e-04, train_time=1.171
+[bmi2:0/4] 2024-07-16 21:16:05,067 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 21:17:32,063 (trainer:779) INFO: 74epoch:train:2479-2891batch: iter_time=0.003, forward_time=0.156, loss_ctc=9.028, loss_att=3.809, acc=0.977, loss=5.375, backward_time=0.256, grad_norm=72.805, clip=100.000, loss_scale=4.255e+33, optim_step_time=0.035, optim0_lr0=5.745e-04, train_time=1.176
+[bmi2:0/4] 2024-07-16 21:21:33,656 (trainer:779) INFO: 74epoch:train:2892-3304batch: iter_time=1.719e-04, forward_time=0.155, loss_ctc=9.015, loss_att=3.835, acc=0.980, loss=5.389, backward_time=0.256, grad_norm=74.509, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.743e-04, train_time=1.169
+[bmi2:0/4] 2024-07-16 21:25:34,148 (trainer:779) INFO: 74epoch:train:3305-3717batch: iter_time=7.501e-04, forward_time=0.154, loss_ctc=8.928, loss_att=3.794, acc=0.976, loss=5.334, backward_time=0.255, grad_norm=74.816, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.741e-04, train_time=1.165
+[bmi2:0/4] 2024-07-16 21:29:36,345 (trainer:779) INFO: 74epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.156, loss_ctc=9.079, loss_att=3.856, acc=0.975, loss=5.423, backward_time=0.256, grad_norm=71.523, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.739e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 21:33:39,284 (trainer:779) INFO: 74epoch:train:4131-4543batch: iter_time=1.746e-04, forward_time=0.156, loss_ctc=9.195, loss_att=3.924, acc=0.981, loss=5.505, backward_time=0.257, grad_norm=79.709, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.737e-04, train_time=1.177
+[bmi2:0/4] 2024-07-16 21:37:42,536 (trainer:779) INFO: 74epoch:train:4544-4956batch: iter_time=0.003, forward_time=0.157, loss_ctc=9.131, loss_att=3.857, acc=0.972, loss=5.439, backward_time=0.256, grad_norm=77.148, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.735e-04, train_time=1.177
+[bmi2:0/4] 2024-07-16 21:41:43,421 (trainer:779) INFO: 74epoch:train:4957-5369batch: iter_time=9.412e-04, forward_time=0.156, loss_ctc=9.174, loss_att=3.881, acc=0.975, loss=5.469, backward_time=0.255, grad_norm=75.241, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.733e-04, train_time=1.167
+[bmi2:0/4] 2024-07-16 21:45:46,099 (trainer:779) INFO: 74epoch:train:5370-5782batch: iter_time=5.897e-04, forward_time=0.157, loss_ctc=9.080, loss_att=3.843, acc=0.978, loss=5.414, backward_time=0.257, grad_norm=71.587, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.731e-04, train_time=1.175
+[bmi2:0/4] 2024-07-16 21:49:48,526 (trainer:779) INFO: 74epoch:train:5783-6195batch: iter_time=8.688e-04, forward_time=0.156, loss_ctc=9.135, loss_att=3.863, acc=0.977, loss=5.444, backward_time=0.256, grad_norm=78.395, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.729e-04, train_time=1.174
+[bmi2:0/4] 2024-07-16 21:53:51,622 (trainer:779) INFO: 74epoch:train:6196-6608batch: iter_time=6.584e-04, forward_time=0.158, loss_ctc=9.086, loss_att=3.859, acc=0.978, loss=5.427, backward_time=0.257, grad_norm=73.121, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.727e-04, train_time=1.177
+[bmi2:0/4] 2024-07-16 21:57:54,256 (trainer:779) INFO: 74epoch:train:6609-7021batch: iter_time=7.544e-04, forward_time=0.157, loss_ctc=9.086, loss_att=3.866, acc=0.974, loss=5.432, backward_time=0.257, grad_norm=80.928, clip=100.000, loss_scale=4.348e+33, optim_step_time=0.035, optim0_lr0=5.725e-04, train_time=1.175
+[bmi2:0/4] 2024-07-16 22:01:42,030 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 22:01:55,931 (trainer:779) INFO: 74epoch:train:7022-7434batch: iter_time=9.296e-04, forward_time=0.155, loss_ctc=9.141, loss_att=3.844, acc=0.976, loss=5.433, backward_time=0.256, grad_norm=74.310, clip=100.000, loss_scale=5.041e+33, optim_step_time=0.036, optim0_lr0=5.723e-04, train_time=1.169
+[bmi2:0/4] 2024-07-16 22:05:59,680 (trainer:779) INFO: 74epoch:train:7435-7847batch: iter_time=0.005, forward_time=0.156, loss_ctc=9.073, loss_att=3.881, acc=0.974, loss=5.439, backward_time=0.257, grad_norm=82.649, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.721e-04, train_time=1.181
+[bmi2:0/4] 2024-07-16 22:10:00,480 (trainer:779) INFO: 74epoch:train:7848-8260batch: iter_time=1.972e-04, forward_time=0.154, loss_ctc=9.166, loss_att=3.880, acc=0.976, loss=5.466, backward_time=0.255, grad_norm=71.055, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.719e-04, train_time=1.165
+[bmi2:0/4] 2024-07-16 22:12:00,552 (trainer:365) INFO: 74epoch results: [train] iter_time=0.001, forward_time=0.156, loss_ctc=9.059, loss_att=3.842, acc=0.976, loss=5.407, backward_time=0.256, grad_norm=75.148, clip=100.000, loss_scale=3.720e+33, optim_step_time=0.036, optim0_lr0=5.738e-04, train_time=1.175, time=1 hour, 20 minutes and 56.25 seconds, total_count=611758, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.721, cer_ctc=0.040, loss_att=5.779, acc=0.951, cer=0.031, wer=0.478, loss=6.962, time=17.78 seconds, total_count=2516, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 37.59 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-16 22:12:05,926 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-16 22:12:05,929 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/62epoch.pth
+[bmi2:0/4] 2024-07-16 22:12:05,929 (trainer:299) INFO: 75/100epoch started. Estimated time to finish: 1 day, 11 hours and 56 minutes
+[bmi2:0/4] 2024-07-16 22:16:15,923 (trainer:779) INFO: 75epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.949, loss_att=3.789, acc=0.977, loss=5.337, backward_time=0.256, grad_norm=74.931, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.718e-04, train_time=1.211
+[bmi2:0/4] 2024-07-16 22:20:18,794 (trainer:779) INFO: 75epoch:train:414-826batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.900, loss_att=3.794, acc=0.976, loss=5.326, backward_time=0.256, grad_norm=74.844, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.716e-04, train_time=1.175
+[bmi2:0/4] 2024-07-16 22:24:21,432 (trainer:779) INFO: 75epoch:train:827-1239batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.934, loss_att=3.783, acc=0.976, loss=5.329, backward_time=0.255, grad_norm=74.666, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.714e-04, train_time=1.175
+[bmi2:0/4] 2024-07-16 22:28:25,023 (trainer:779) INFO: 75epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.159, loss_ctc=9.016, loss_att=3.831, acc=0.976, loss=5.386, backward_time=0.256, grad_norm=73.552, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.712e-04, train_time=1.179
+[bmi2:0/4] 2024-07-16 22:32:26,557 (trainer:779) INFO: 75epoch:train:1653-2065batch: iter_time=7.423e-04, forward_time=0.156, loss_ctc=8.951, loss_att=3.816, acc=0.977, loss=5.357, backward_time=0.254, grad_norm=69.343, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.710e-04, train_time=1.170
+[bmi2:0/4] 2024-07-16 22:36:31,029 (trainer:779) INFO: 75epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.860, loss_att=3.765, acc=0.975, loss=5.294, backward_time=0.257, grad_norm=67.827, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.708e-04, train_time=1.183
+[bmi2:0/4] 2024-07-16 22:40:33,743 (trainer:779) INFO: 75epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.937, loss_att=3.800, acc=0.977, loss=5.341, backward_time=0.256, grad_norm=75.369, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.706e-04, train_time=1.176
+[bmi2:0/4] 2024-07-16 22:44:35,439 (trainer:779) INFO: 75epoch:train:2892-3304batch: iter_time=0.002, forward_time=0.156, loss_ctc=9.049, loss_att=3.831, acc=0.978, loss=5.396, backward_time=0.255, grad_norm=74.337, clip=100.000, loss_scale=3.599e+33, optim_step_time=0.036, optim0_lr0=5.704e-04, train_time=1.170
+[bmi2:0/4] 2024-07-16 22:48:38,391 (trainer:779) INFO: 75epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.909, loss_att=3.782, acc=0.976, loss=5.320, backward_time=0.256, grad_norm=72.043, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.702e-04, train_time=1.177
+[bmi2:0/4] 2024-07-16 22:52:40,878 (trainer:779) INFO: 75epoch:train:3718-4130batch: iter_time=8.815e-04, forward_time=0.157, loss_ctc=8.923, loss_att=3.800, acc=0.976, loss=5.337, backward_time=0.255, grad_norm=82.340, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.700e-04, train_time=1.174
+[bmi2:0/4] 2024-07-16 22:56:42,933 (trainer:779) INFO: 75epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.157, loss_ctc=9.008, loss_att=3.822, acc=0.978, loss=5.378, backward_time=0.255, grad_norm=76.163, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.698e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 23:00:44,029 (trainer:779) INFO: 75epoch:train:4544-4956batch: iter_time=4.323e-04, forward_time=0.154, loss_ctc=8.956, loss_att=3.801, acc=0.975, loss=5.347, backward_time=0.255, grad_norm=72.450, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.696e-04, train_time=1.167
+[bmi2:0/4] 2024-07-16 23:04:43,689 (trainer:779) INFO: 75epoch:train:4957-5369batch: iter_time=2.756e-04, forward_time=0.155, loss_ctc=8.985, loss_att=3.835, acc=0.976, loss=5.380, backward_time=0.254, grad_norm=73.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.695e-04, train_time=1.161
+[bmi2:0/4] 2024-07-16 23:08:45,439 (trainer:779) INFO: 75epoch:train:5370-5782batch: iter_time=3.856e-04, forward_time=0.156, loss_ctc=8.915, loss_att=3.783, acc=0.978, loss=5.323, backward_time=0.256, grad_norm=71.857, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.693e-04, train_time=1.170
+[bmi2:0/4] 2024-07-16 23:12:47,403 (trainer:779) INFO: 75epoch:train:5783-6195batch: iter_time=4.333e-04, forward_time=0.156, loss_ctc=8.936, loss_att=3.819, acc=0.976, loss=5.354, backward_time=0.256, grad_norm=66.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.691e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 23:16:48,139 (trainer:779) INFO: 75epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.935, loss_att=3.816, acc=0.976, loss=5.351, backward_time=0.254, grad_norm=74.335, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.689e-04, train_time=1.165
+[bmi2:0/4] 2024-07-16 23:20:50,136 (trainer:779) INFO: 75epoch:train:6609-7021batch: iter_time=4.547e-04, forward_time=0.157, loss_ctc=9.095, loss_att=3.848, acc=0.977, loss=5.422, backward_time=0.255, grad_norm=71.722, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.687e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 23:23:55,251 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 23:24:52,818 (trainer:779) INFO: 75epoch:train:7022-7434batch: iter_time=1.657e-04, forward_time=0.158, loss_ctc=8.875, loss_att=3.783, acc=0.980, loss=5.310, backward_time=0.256, grad_norm=69.600, clip=100.000, loss_scale=7.587e+33, optim_step_time=0.036, optim0_lr0=5.685e-04, train_time=1.174
+[bmi2:0/4] 2024-07-16 23:28:56,109 (trainer:779) INFO: 75epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.956, loss_att=3.817, acc=0.977, loss=5.359, backward_time=0.255, grad_norm=75.400, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.683e-04, train_time=1.178
+[bmi2:0/4] 2024-07-16 23:32:57,473 (trainer:779) INFO: 75epoch:train:7848-8260batch: iter_time=5.086e-04, forward_time=0.157, loss_ctc=8.993, loss_att=3.819, acc=0.976, loss=5.371, backward_time=0.254, grad_norm=71.221, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.681e-04, train_time=1.168
+[bmi2:0/4] 2024-07-16 23:34:44,859 (trainer:365) INFO: 75epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.955, loss_att=3.807, acc=0.977, loss=5.351, backward_time=0.255, grad_norm=73.123, clip=100.000, loss_scale=4.324e+33, optim_step_time=0.036, optim0_lr0=5.699e-04, train_time=1.174, time=1 hour, 20 minutes and 56.39 seconds, total_count=620025, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.671, cer_ctc=0.040, loss_att=5.799, acc=0.949, cer=0.032, wer=0.490, loss=6.961, time=17.33 seconds, total_count=2550, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 25.2 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-16 23:34:49,177 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-16 23:34:49,178 (trainer:299) INFO: 76/100epoch started. Estimated time to finish: 1 day, 10 hours and 30 minutes
+[bmi2:0/4] 2024-07-16 23:39:01,988 (trainer:779) INFO: 76epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.701, loss_att=3.707, acc=0.976, loss=5.205, backward_time=0.257, grad_norm=69.089, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.679e-04, train_time=1.225
+[bmi2:0/4] 2024-07-16 23:43:04,217 (trainer:779) INFO: 76epoch:train:414-826batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.849, loss_att=3.763, acc=0.978, loss=5.289, backward_time=0.255, grad_norm=74.221, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.677e-04, train_time=1.172
+[bmi2:0/4] 2024-07-16 23:45:41,778 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-16 23:47:06,944 (trainer:779) INFO: 76epoch:train:827-1239batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.778, loss_att=3.722, acc=0.976, loss=5.239, backward_time=0.255, grad_norm=68.112, clip=100.000, loss_scale=4.280e+33, optim_step_time=0.036, optim0_lr0=5.676e-04, train_time=1.176
+[bmi2:0/4] 2024-07-16 23:51:09,881 (trainer:779) INFO: 76epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.971, loss_att=3.831, acc=0.979, loss=5.373, backward_time=0.256, grad_norm=71.018, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.674e-04, train_time=1.176
+[bmi2:0/4] 2024-07-16 23:55:11,400 (trainer:779) INFO: 76epoch:train:1653-2065batch: iter_time=2.006e-04, forward_time=0.155, loss_ctc=8.965, loss_att=3.831, acc=0.978, loss=5.371, backward_time=0.256, grad_norm=80.094, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.672e-04, train_time=1.170
+[bmi2:0/4] 2024-07-16 23:59:12,689 (trainer:779) INFO: 76epoch:train:2066-2478batch: iter_time=2.969e-04, forward_time=0.156, loss_ctc=8.838, loss_att=3.752, acc=0.980, loss=5.278, backward_time=0.255, grad_norm=68.032, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.670e-04, train_time=1.168
+[bmi2:0/4] 2024-07-17 00:03:14,121 (trainer:779) INFO: 76epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.967, loss_att=3.816, acc=0.974, loss=5.361, backward_time=0.254, grad_norm=69.738, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.668e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 00:07:16,091 (trainer:779) INFO: 76epoch:train:2892-3304batch: iter_time=7.143e-04, forward_time=0.158, loss_ctc=8.848, loss_att=3.785, acc=0.976, loss=5.304, backward_time=0.255, grad_norm=79.838, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.666e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 00:11:17,590 (trainer:779) INFO: 76epoch:train:3305-3717batch: iter_time=1.719e-04, forward_time=0.155, loss_ctc=8.973, loss_att=3.801, acc=0.979, loss=5.352, backward_time=0.255, grad_norm=76.200, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.664e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 00:15:18,713 (trainer:779) INFO: 76epoch:train:3718-4130batch: iter_time=8.038e-04, forward_time=0.156, loss_ctc=8.943, loss_att=3.796, acc=0.976, loss=5.340, backward_time=0.254, grad_norm=72.707, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.662e-04, train_time=1.167
+[bmi2:0/4] 2024-07-17 00:19:21,223 (trainer:779) INFO: 76epoch:train:4131-4543batch: iter_time=1.692e-04, forward_time=0.158, loss_ctc=8.917, loss_att=3.801, acc=0.977, loss=5.336, backward_time=0.255, grad_norm=74.020, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.660e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 00:23:22,754 (trainer:779) INFO: 76epoch:train:4544-4956batch: iter_time=5.163e-04, forward_time=0.156, loss_ctc=9.011, loss_att=3.817, acc=0.978, loss=5.375, backward_time=0.254, grad_norm=73.430, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.659e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 00:27:25,368 (trainer:779) INFO: 76epoch:train:4957-5369batch: iter_time=5.505e-04, forward_time=0.157, loss_ctc=8.825, loss_att=3.751, acc=0.978, loss=5.273, backward_time=0.255, grad_norm=72.184, clip=100.000, loss_scale=4.323e+33, optim_step_time=0.036, optim0_lr0=5.657e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 00:31:26,317 (trainer:779) INFO: 76epoch:train:5370-5782batch: iter_time=1.736e-04, forward_time=0.156, loss_ctc=8.889, loss_att=3.777, acc=0.978, loss=5.311, backward_time=0.254, grad_norm=70.457, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.655e-04, train_time=1.166
+[bmi2:0/4] 2024-07-17 00:35:28,741 (trainer:779) INFO: 76epoch:train:5783-6195batch: iter_time=2.572e-04, forward_time=0.158, loss_ctc=8.939, loss_att=3.802, acc=0.974, loss=5.343, backward_time=0.255, grad_norm=72.428, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.653e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 00:39:31,596 (trainer:779) INFO: 76epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.924, loss_att=3.786, acc=0.975, loss=5.327, backward_time=0.255, grad_norm=82.587, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.651e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 00:43:33,408 (trainer:779) INFO: 76epoch:train:6609-7021batch: iter_time=6.003e-04, forward_time=0.156, loss_ctc=8.945, loss_att=3.784, acc=0.978, loss=5.332, backward_time=0.255, grad_norm=72.977, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.649e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 00:47:34,821 (trainer:779) INFO: 76epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.908, loss_att=3.784, acc=0.977, loss=5.321, backward_time=0.254, grad_norm=74.010, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.647e-04, train_time=1.168
+[bmi2:0/4] 2024-07-17 00:51:34,885 (trainer:779) INFO: 76epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.154, loss_ctc=9.008, loss_att=3.827, acc=0.976, loss=5.382, backward_time=0.253, grad_norm=73.096, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.646e-04, train_time=1.163
+[bmi2:0/4] 2024-07-17 00:55:35,127 (trainer:779) INFO: 76epoch:train:7848-8260batch: iter_time=1.786e-04, forward_time=0.155, loss_ctc=8.885, loss_att=3.782, acc=0.975, loss=5.313, backward_time=0.253, grad_norm=69.809, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.644e-04, train_time=1.163
+[bmi2:0/4] 2024-07-17 00:57:22,020 (trainer:365) INFO: 76epoch results: [train] iter_time=8.253e-04, forward_time=0.156, loss_ctc=8.904, loss_att=3.785, acc=0.977, loss=5.321, backward_time=0.255, grad_norm=73.199, clip=100.000, loss_scale=3.936e+33, optim_step_time=0.036, optim0_lr0=5.661e-04, train_time=1.173, time=1 hour, 20 minutes and 50.86 seconds, total_count=628292, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.051, cer_ctc=0.040, loss_att=5.853, acc=0.949, cer=0.032, wer=0.488, loss=7.112, time=17.75 seconds, total_count=2584, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 24.22 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 00:57:26,814 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 00:57:26,867 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/75epoch.pth
+[bmi2:0/4] 2024-07-17 00:57:26,868 (trainer:299) INFO: 77/100epoch started. Estimated time to finish: 1 day, 9 hours and 6 minutes
+[bmi2:0/4] 2024-07-17 01:01:37,497 (trainer:779) INFO: 77epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.741, loss_att=3.712, acc=0.976, loss=5.221, backward_time=0.256, grad_norm=79.411, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.642e-04, train_time=1.214
+[bmi2:0/4] 2024-07-17 01:05:40,580 (trainer:779) INFO: 77epoch:train:414-826batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.763, loss_att=3.712, acc=0.977, loss=5.227, backward_time=0.256, grad_norm=75.501, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.640e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 01:07:05,793 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 01:09:41,892 (trainer:779) INFO: 77epoch:train:827-1239batch: iter_time=6.986e-04, forward_time=0.157, loss_ctc=8.999, loss_att=3.807, acc=0.977, loss=5.365, backward_time=0.254, grad_norm=73.188, clip=100.000, loss_scale=6.991e+33, optim_step_time=0.036, optim0_lr0=5.638e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 01:13:42,556 (trainer:779) INFO: 77epoch:train:1240-1652batch: iter_time=1.719e-04, forward_time=0.156, loss_ctc=8.918, loss_att=3.768, acc=0.979, loss=5.313, backward_time=0.254, grad_norm=74.280, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.636e-04, train_time=1.165
+[bmi2:0/4] 2024-07-17 01:17:44,360 (trainer:779) INFO: 77epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.756, loss_att=3.729, acc=0.977, loss=5.237, backward_time=0.255, grad_norm=72.781, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.634e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 01:21:46,563 (trainer:779) INFO: 77epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.771, loss_att=3.716, acc=0.973, loss=5.232, backward_time=0.255, grad_norm=72.867, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.633e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 01:25:48,929 (trainer:779) INFO: 77epoch:train:2479-2891batch: iter_time=1.712e-04, forward_time=0.156, loss_ctc=8.917, loss_att=3.795, acc=0.977, loss=5.332, backward_time=0.255, grad_norm=71.758, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.631e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 01:29:50,618 (trainer:779) INFO: 77epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.894, loss_att=3.784, acc=0.977, loss=5.317, backward_time=0.254, grad_norm=73.767, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.629e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 01:33:52,121 (trainer:779) INFO: 77epoch:train:3305-3717batch: iter_time=2.172e-04, forward_time=0.156, loss_ctc=8.894, loss_att=3.760, acc=0.980, loss=5.301, backward_time=0.254, grad_norm=72.510, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.627e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 01:37:55,519 (trainer:779) INFO: 77epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.972, loss_att=3.791, acc=0.977, loss=5.345, backward_time=0.258, grad_norm=75.533, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.625e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 01:41:55,713 (trainer:779) INFO: 77epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.833, loss_att=3.771, acc=0.978, loss=5.290, backward_time=0.254, grad_norm=71.178, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.623e-04, train_time=1.164
+[bmi2:0/4] 2024-07-17 01:45:56,724 (trainer:779) INFO: 77epoch:train:4544-4956batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.862, loss_att=3.799, acc=0.978, loss=5.318, backward_time=0.255, grad_norm=74.903, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.622e-04, train_time=1.166
+[bmi2:0/4] 2024-07-17 01:48:18,827 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 01:50:00,309 (trainer:779) INFO: 77epoch:train:4957-5369batch: iter_time=9.655e-04, forward_time=0.159, loss_ctc=8.834, loss_att=3.757, acc=0.975, loss=5.280, backward_time=0.256, grad_norm=74.044, clip=100.000, loss_scale=8.004e+33, optim_step_time=0.036, optim0_lr0=5.620e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 01:54:03,383 (trainer:779) INFO: 77epoch:train:5370-5782batch: iter_time=5.919e-04, forward_time=0.158, loss_ctc=8.779, loss_att=3.747, acc=0.978, loss=5.257, backward_time=0.256, grad_norm=70.200, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.618e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 01:58:05,185 (trainer:779) INFO: 77epoch:train:5783-6195batch: iter_time=3.119e-04, forward_time=0.158, loss_ctc=8.922, loss_att=3.783, acc=0.978, loss=5.325, backward_time=0.254, grad_norm=71.555, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.616e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 02:02:07,209 (trainer:779) INFO: 77epoch:train:6196-6608batch: iter_time=2.825e-04, forward_time=0.157, loss_ctc=8.929, loss_att=3.779, acc=0.976, loss=5.324, backward_time=0.255, grad_norm=70.617, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.614e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 02:02:32,924 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 02:06:09,073 (trainer:779) INFO: 77epoch:train:6609-7021batch: iter_time=2.448e-04, forward_time=0.157, loss_ctc=8.846, loss_att=3.745, acc=0.978, loss=5.275, backward_time=0.255, grad_norm=71.303, clip=100.000, loss_scale=2.862e+33, optim_step_time=0.035, optim0_lr0=5.612e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 02:10:10,864 (trainer:779) INFO: 77epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.916, loss_att=3.784, acc=0.975, loss=5.323, backward_time=0.254, grad_norm=88.873, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.611e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 02:14:13,614 (trainer:779) INFO: 77epoch:train:7435-7847batch: iter_time=9.122e-04, forward_time=0.158, loss_ctc=8.882, loss_att=3.763, acc=0.976, loss=5.299, backward_time=0.256, grad_norm=72.718, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.609e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 02:18:16,148 (trainer:779) INFO: 77epoch:train:7848-8260batch: iter_time=6.422e-04, forward_time=0.158, loss_ctc=8.922, loss_att=3.786, acc=0.977, loss=5.326, backward_time=0.255, grad_norm=72.815, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.607e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 02:20:03,871 (trainer:365) INFO: 77epoch results: [train] iter_time=8.817e-04, forward_time=0.157, loss_ctc=8.867, loss_att=3.764, acc=0.977, loss=5.295, backward_time=0.255, grad_norm=74.011, clip=100.000, loss_scale=4.914e+33, optim_step_time=0.035, optim0_lr0=5.624e-04, train_time=1.174, time=1 hour, 20 minutes and 54.25 seconds, total_count=636559, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.081, cer_ctc=0.040, loss_att=5.866, acc=0.950, cer=0.032, wer=0.484, loss=7.131, time=17.16 seconds, total_count=2618, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 25.59 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 02:20:09,189 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 02:20:09,245 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/65epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/76epoch.pth
+[bmi2:0/4] 2024-07-17 02:20:09,245 (trainer:299) INFO: 78/100epoch started. Estimated time to finish: 1 day, 7 hours and 43 minutes
+[bmi2:0/4] 2024-07-17 02:24:22,095 (trainer:779) INFO: 78epoch:train:1-413batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.777, loss_att=3.729, acc=0.976, loss=5.243, backward_time=0.255, grad_norm=69.096, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.605e-04, train_time=1.225
+[bmi2:0/4] 2024-07-17 02:28:24,575 (trainer:779) INFO: 78epoch:train:414-826batch: iter_time=9.112e-04, forward_time=0.157, loss_ctc=8.781, loss_att=3.723, acc=0.978, loss=5.240, backward_time=0.256, grad_norm=69.990, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.603e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 02:32:25,116 (trainer:779) INFO: 78epoch:train:827-1239batch: iter_time=1.873e-04, forward_time=0.155, loss_ctc=8.669, loss_att=3.710, acc=0.977, loss=5.198, backward_time=0.254, grad_norm=67.615, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.601e-04, train_time=1.165
+[bmi2:0/4] 2024-07-17 02:36:27,191 (trainer:779) INFO: 78epoch:train:1240-1652batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.905, loss_att=3.800, acc=0.973, loss=5.331, backward_time=0.255, grad_norm=73.176, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.600e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 02:40:30,341 (trainer:779) INFO: 78epoch:train:1653-2065batch: iter_time=0.003, forward_time=0.158, loss_ctc=8.803, loss_att=3.745, acc=0.977, loss=5.263, backward_time=0.255, grad_norm=71.852, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.598e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 02:44:34,119 (trainer:779) INFO: 78epoch:train:2066-2478batch: iter_time=0.003, forward_time=0.158, loss_ctc=8.746, loss_att=3.732, acc=0.977, loss=5.236, backward_time=0.256, grad_norm=70.036, clip=100.000, loss_scale=3.173e+33, optim_step_time=0.035, optim0_lr0=5.596e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 02:48:37,701 (trainer:779) INFO: 78epoch:train:2479-2891batch: iter_time=6.833e-04, forward_time=0.159, loss_ctc=8.718, loss_att=3.721, acc=0.980, loss=5.220, backward_time=0.256, grad_norm=72.700, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.594e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 02:52:40,905 (trainer:779) INFO: 78epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.879, loss_att=3.766, acc=0.976, loss=5.300, backward_time=0.256, grad_norm=77.285, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.592e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 02:56:43,744 (trainer:779) INFO: 78epoch:train:3305-3717batch: iter_time=8.894e-04, forward_time=0.158, loss_ctc=8.737, loss_att=3.714, acc=0.978, loss=5.221, backward_time=0.255, grad_norm=68.410, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.591e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 03:00:45,410 (trainer:779) INFO: 78epoch:train:3718-4130batch: iter_time=2.077e-04, forward_time=0.155, loss_ctc=8.971, loss_att=3.804, acc=0.978, loss=5.354, backward_time=0.255, grad_norm=70.394, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.589e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 03:04:48,676 (trainer:779) INFO: 78epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.982, loss_att=3.788, acc=0.974, loss=5.346, backward_time=0.255, grad_norm=68.389, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.587e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 03:08:51,653 (trainer:779) INFO: 78epoch:train:4544-4956batch: iter_time=2.791e-04, forward_time=0.158, loss_ctc=8.751, loss_att=3.708, acc=0.976, loss=5.221, backward_time=0.256, grad_norm=68.491, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.585e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 03:12:53,485 (trainer:779) INFO: 78epoch:train:4957-5369batch: iter_time=1.761e-04, forward_time=0.157, loss_ctc=8.710, loss_att=3.701, acc=0.979, loss=5.203, backward_time=0.254, grad_norm=70.334, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.583e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 03:16:55,156 (trainer:779) INFO: 78epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.848, loss_att=3.770, acc=0.976, loss=5.293, backward_time=0.254, grad_norm=71.078, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.582e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 03:20:55,582 (trainer:779) INFO: 78epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.829, loss_att=3.715, acc=0.977, loss=5.249, backward_time=0.253, grad_norm=69.061, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.580e-04, train_time=1.165
+[bmi2:0/4] 2024-07-17 03:24:55,487 (trainer:779) INFO: 78epoch:train:6196-6608batch: iter_time=8.593e-04, forward_time=0.155, loss_ctc=8.790, loss_att=3.735, acc=0.975, loss=5.252, backward_time=0.253, grad_norm=70.868, clip=100.000, loss_scale=7.977e+33, optim_step_time=0.035, optim0_lr0=5.578e-04, train_time=1.161
+[bmi2:0/4] 2024-07-17 03:28:17,177 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 03:28:58,558 (trainer:779) INFO: 78epoch:train:6609-7021batch: iter_time=1.648e-04, forward_time=0.158, loss_ctc=8.776, loss_att=3.703, acc=0.980, loss=5.225, backward_time=0.255, grad_norm=69.424, clip=100.000, loss_scale=9.498e+33, optim_step_time=0.036, optim0_lr0=5.576e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 03:33:01,119 (trainer:779) INFO: 78epoch:train:7022-7434batch: iter_time=3.411e-04, forward_time=0.158, loss_ctc=8.751, loss_att=3.701, acc=0.979, loss=5.216, backward_time=0.256, grad_norm=69.134, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.574e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 03:37:02,725 (trainer:779) INFO: 78epoch:train:7435-7847batch: iter_time=1.620e-04, forward_time=0.157, loss_ctc=8.963, loss_att=3.799, acc=0.979, loss=5.348, backward_time=0.254, grad_norm=73.461, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.573e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 03:41:05,289 (trainer:779) INFO: 78epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.830, loss_att=3.732, acc=0.977, loss=5.261, backward_time=0.255, grad_norm=73.905, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.571e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 03:42:55,324 (trainer:365) INFO: 78epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.811, loss_att=3.740, acc=0.977, loss=5.261, backward_time=0.255, grad_norm=70.719, clip=100.000, loss_scale=4.796e+33, optim_step_time=0.036, optim0_lr0=5.588e-04, train_time=1.176, time=1 hour, 21 minutes and 1.21 seconds, total_count=644826, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.069, cer_ctc=0.040, loss_att=5.849, acc=0.950, cer=0.032, wer=0.480, loss=7.115, time=17.85 seconds, total_count=2652, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 27.02 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 03:43:00,023 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 03:43:00,027 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/49epoch.pth
+[bmi2:0/4] 2024-07-17 03:43:00,027 (trainer:299) INFO: 79/100epoch started. Estimated time to finish: 1 day, 6 hours and 20 minutes
+[bmi2:0/4] 2024-07-17 03:47:10,813 (trainer:779) INFO: 79epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.776, loss_att=3.736, acc=0.978, loss=5.248, backward_time=0.254, grad_norm=71.067, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.569e-04, train_time=1.215
+[bmi2:0/4] 2024-07-17 03:50:31,899 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 03:51:14,120 (trainer:779) INFO: 79epoch:train:414-826batch: iter_time=7.596e-04, forward_time=0.159, loss_ctc=8.782, loss_att=3.738, acc=0.978, loss=5.251, backward_time=0.256, grad_norm=71.173, clip=100.000, loss_scale=4.739e+33, optim_step_time=0.036, optim0_lr0=5.567e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 03:55:17,481 (trainer:779) INFO: 79epoch:train:827-1239batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.778, loss_att=3.725, acc=0.976, loss=5.241, backward_time=0.255, grad_norm=68.891, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.566e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 03:59:20,759 (trainer:779) INFO: 79epoch:train:1240-1652batch: iter_time=8.646e-04, forward_time=0.158, loss_ctc=8.622, loss_att=3.668, acc=0.976, loss=5.154, backward_time=0.255, grad_norm=72.213, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.564e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 04:03:24,531 (trainer:779) INFO: 79epoch:train:1653-2065batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.644, loss_att=3.683, acc=0.975, loss=5.171, backward_time=0.255, grad_norm=66.746, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.562e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 04:07:27,886 (trainer:779) INFO: 79epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.775, loss_att=3.720, acc=0.978, loss=5.236, backward_time=0.255, grad_norm=79.386, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.560e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 04:11:30,396 (trainer:779) INFO: 79epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.876, loss_att=3.744, acc=0.977, loss=5.283, backward_time=0.254, grad_norm=73.288, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.558e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 04:15:33,248 (trainer:779) INFO: 79epoch:train:2892-3304batch: iter_time=1.768e-04, forward_time=0.159, loss_ctc=8.852, loss_att=3.750, acc=0.976, loss=5.280, backward_time=0.256, grad_norm=73.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.557e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 04:19:37,577 (trainer:779) INFO: 79epoch:train:3305-3717batch: iter_time=1.868e-04, forward_time=0.162, loss_ctc=8.805, loss_att=3.729, acc=0.979, loss=5.252, backward_time=0.258, grad_norm=72.606, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.037, optim0_lr0=5.555e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 04:23:40,433 (trainer:779) INFO: 79epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.857, loss_att=3.748, acc=0.975, loss=5.280, backward_time=0.256, grad_norm=75.410, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.553e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 04:27:44,657 (trainer:779) INFO: 79epoch:train:4131-4543batch: iter_time=5.390e-04, forward_time=0.159, loss_ctc=8.923, loss_att=3.756, acc=0.978, loss=5.306, backward_time=0.257, grad_norm=74.557, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.551e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 04:31:46,786 (trainer:779) INFO: 79epoch:train:4544-4956batch: iter_time=1.868e-04, forward_time=0.158, loss_ctc=8.829, loss_att=3.733, acc=0.978, loss=5.262, backward_time=0.255, grad_norm=66.742, clip=100.000, loss_scale=3.863e+33, optim_step_time=0.036, optim0_lr0=5.550e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 04:34:00,743 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 04:35:49,624 (trainer:779) INFO: 79epoch:train:4957-5369batch: iter_time=1.901e-04, forward_time=0.159, loss_ctc=8.833, loss_att=3.735, acc=0.977, loss=5.264, backward_time=0.256, grad_norm=75.081, clip=100.000, loss_scale=4.027e+33, optim_step_time=0.036, optim0_lr0=5.548e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 04:39:52,089 (trainer:779) INFO: 79epoch:train:5370-5782batch: iter_time=6.050e-04, forward_time=0.158, loss_ctc=8.804, loss_att=3.748, acc=0.979, loss=5.265, backward_time=0.256, grad_norm=70.336, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.546e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 04:43:55,962 (trainer:779) INFO: 79epoch:train:5783-6195batch: iter_time=2.307e-04, forward_time=0.160, loss_ctc=8.784, loss_att=3.721, acc=0.978, loss=5.240, backward_time=0.256, grad_norm=71.525, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.544e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 04:47:59,365 (trainer:779) INFO: 79epoch:train:6196-6608batch: iter_time=1.797e-04, forward_time=0.160, loss_ctc=8.884, loss_att=3.738, acc=0.979, loss=5.282, backward_time=0.257, grad_norm=79.797, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.543e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 04:52:02,610 (trainer:779) INFO: 79epoch:train:6609-7021batch: iter_time=1.880e-04, forward_time=0.161, loss_ctc=8.886, loss_att=3.758, acc=0.976, loss=5.296, backward_time=0.256, grad_norm=69.830, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.541e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 04:56:05,463 (trainer:779) INFO: 79epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.793, loss_att=3.718, acc=0.975, loss=5.241, backward_time=0.255, grad_norm=70.320, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.539e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 05:00:07,807 (trainer:779) INFO: 79epoch:train:7435-7847batch: iter_time=1.679e-04, forward_time=0.157, loss_ctc=8.872, loss_att=3.748, acc=0.977, loss=5.285, backward_time=0.255, grad_norm=72.010, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.537e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 05:04:10,844 (trainer:779) INFO: 79epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.873, loss_att=3.772, acc=0.977, loss=5.303, backward_time=0.256, grad_norm=79.065, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.536e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 05:06:38,229 (trainer:365) INFO: 79epoch results: [train] iter_time=9.151e-04, forward_time=0.159, loss_ctc=8.812, loss_att=3.733, acc=0.977, loss=5.257, backward_time=0.256, grad_norm=72.695, clip=100.000, loss_scale=2.967e+33, optim_step_time=0.036, optim0_lr0=5.552e-04, train_time=1.179, time=1 hour, 21 minutes and 15.78 seconds, total_count=653093, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.929, cer_ctc=0.040, loss_att=6.296, acc=0.948, cer=0.033, wer=0.489, loss=7.385, time=38.78 seconds, total_count=2686, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 43.64 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 05:06:42,975 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 05:06:42,976 (trainer:299) INFO: 80/100epoch started. Estimated time to finish: 1 day, 5 hours and 1 minute
+[bmi2:0/4] 2024-07-17 05:11:26,229 (trainer:779) INFO: 80epoch:train:1-413batch: iter_time=0.002, forward_time=0.154, loss_ctc=8.839, loss_att=3.743, acc=0.975, loss=5.272, backward_time=0.255, grad_norm=71.859, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.534e-04, train_time=1.372
+[bmi2:0/4] 2024-07-17 05:15:30,015 (trainer:779) INFO: 80epoch:train:414-826batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.818, loss_att=3.742, acc=0.976, loss=5.265, backward_time=0.255, grad_norm=76.070, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.532e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 05:19:32,381 (trainer:779) INFO: 80epoch:train:827-1239batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.738, loss_att=3.693, acc=0.975, loss=5.207, backward_time=0.254, grad_norm=73.940, clip=100.000, loss_scale=4.613e+33, optim_step_time=0.036, optim0_lr0=5.530e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 05:23:36,039 (trainer:779) INFO: 80epoch:train:1240-1652batch: iter_time=9.142e-04, forward_time=0.158, loss_ctc=8.848, loss_att=3.754, acc=0.977, loss=5.282, backward_time=0.256, grad_norm=78.693, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.528e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 05:27:38,907 (trainer:779) INFO: 80epoch:train:1653-2065batch: iter_time=2.042e-04, forward_time=0.157, loss_ctc=8.723, loss_att=3.672, acc=0.979, loss=5.187, backward_time=0.256, grad_norm=72.075, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.527e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 05:31:42,508 (trainer:779) INFO: 80epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.562, loss_att=3.645, acc=0.975, loss=5.121, backward_time=0.256, grad_norm=71.910, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.525e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 05:35:45,049 (trainer:779) INFO: 80epoch:train:2479-2891batch: iter_time=5.590e-04, forward_time=0.156, loss_ctc=8.756, loss_att=3.712, acc=0.978, loss=5.225, backward_time=0.255, grad_norm=70.664, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.523e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 05:39:48,751 (trainer:779) INFO: 80epoch:train:2892-3304batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.662, loss_att=3.685, acc=0.974, loss=5.178, backward_time=0.256, grad_norm=67.286, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.522e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 05:42:15,643 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 05:43:52,479 (trainer:779) INFO: 80epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.759, loss_att=3.730, acc=0.979, loss=5.239, backward_time=0.255, grad_norm=75.716, clip=100.000, loss_scale=4.154e+33, optim_step_time=0.036, optim0_lr0=5.520e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 05:47:55,406 (trainer:779) INFO: 80epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.897, loss_att=3.781, acc=0.977, loss=5.316, backward_time=0.255, grad_norm=77.965, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.518e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 05:51:59,508 (trainer:779) INFO: 80epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.885, loss_att=3.758, acc=0.976, loss=5.296, backward_time=0.256, grad_norm=68.866, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.516e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 05:56:02,172 (trainer:779) INFO: 80epoch:train:4544-4956batch: iter_time=0.005, forward_time=0.154, loss_ctc=8.616, loss_att=3.644, acc=0.974, loss=5.136, backward_time=0.254, grad_norm=67.824, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.515e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 06:00:04,253 (trainer:779) INFO: 80epoch:train:4957-5369batch: iter_time=1.884e-04, forward_time=0.156, loss_ctc=8.726, loss_att=3.711, acc=0.979, loss=5.215, backward_time=0.255, grad_norm=68.306, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.513e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 06:04:06,393 (trainer:779) INFO: 80epoch:train:5370-5782batch: iter_time=1.683e-04, forward_time=0.156, loss_ctc=8.793, loss_att=3.701, acc=0.979, loss=5.229, backward_time=0.255, grad_norm=69.462, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.511e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 06:08:09,829 (trainer:779) INFO: 80epoch:train:5783-6195batch: iter_time=8.281e-04, forward_time=0.158, loss_ctc=8.736, loss_att=3.711, acc=0.978, loss=5.219, backward_time=0.255, grad_norm=66.120, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.509e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 06:12:13,379 (trainer:779) INFO: 80epoch:train:6196-6608batch: iter_time=8.845e-04, forward_time=0.157, loss_ctc=8.761, loss_att=3.679, acc=0.979, loss=5.203, backward_time=0.255, grad_norm=70.977, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.508e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 06:16:15,875 (trainer:779) INFO: 80epoch:train:6609-7021batch: iter_time=4.291e-04, forward_time=0.156, loss_ctc=8.836, loss_att=3.702, acc=0.979, loss=5.242, backward_time=0.255, grad_norm=72.977, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.506e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 06:20:19,484 (trainer:779) INFO: 80epoch:train:7022-7434batch: iter_time=1.639e-04, forward_time=0.159, loss_ctc=8.765, loss_att=3.718, acc=0.979, loss=5.232, backward_time=0.256, grad_norm=70.379, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.504e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 06:24:21,960 (trainer:779) INFO: 80epoch:train:7435-7847batch: iter_time=8.676e-04, forward_time=0.156, loss_ctc=8.715, loss_att=3.693, acc=0.978, loss=5.200, backward_time=0.255, grad_norm=69.872, clip=100.000, loss_scale=4.449e+33, optim_step_time=0.035, optim0_lr0=5.503e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 06:28:25,163 (trainer:779) INFO: 80epoch:train:7848-8260batch: iter_time=2.542e-04, forward_time=0.157, loss_ctc=8.943, loss_att=3.778, acc=0.979, loss=5.328, backward_time=0.255, grad_norm=70.000, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.501e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 06:31:06,574 (trainer:365) INFO: 80epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.768, loss_att=3.712, acc=0.977, loss=5.229, backward_time=0.255, grad_norm=71.561, clip=100.000, loss_scale=3.647e+33, optim_step_time=0.036, optim0_lr0=5.517e-04, train_time=1.187, time=1 hour, 21 minutes and 47.15 seconds, total_count=661360, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.891, cer_ctc=0.039, loss_att=5.917, acc=0.950, cer=0.032, wer=0.485, loss=7.109, time=43.72 seconds, total_count=2720, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 52.72 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 06:31:10,729 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 06:31:10,803 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/73epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/79epoch.pth
+[bmi2:0/4] 2024-07-17 06:31:10,803 (trainer:299) INFO: 81/100epoch started. Estimated time to finish: 1 day, 3 hours and 42 minutes
+[bmi2:0/4] 2024-07-17 06:35:55,532 (trainer:779) INFO: 81epoch:train:1-413batch: iter_time=0.004, forward_time=0.154, loss_ctc=8.576, loss_att=3.659, acc=0.975, loss=5.134, backward_time=0.254, grad_norm=66.923, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.499e-04, train_time=1.379
+[bmi2:0/4] 2024-07-17 06:39:59,307 (trainer:779) INFO: 81epoch:train:414-826batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.612, loss_att=3.633, acc=0.978, loss=5.127, backward_time=0.255, grad_norm=67.939, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.497e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 06:44:01,303 (trainer:779) INFO: 81epoch:train:827-1239batch: iter_time=3.123e-04, forward_time=0.157, loss_ctc=8.758, loss_att=3.700, acc=0.979, loss=5.217, backward_time=0.254, grad_norm=70.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.496e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 06:48:03,969 (trainer:779) INFO: 81epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.693, loss_att=3.669, acc=0.977, loss=5.176, backward_time=0.255, grad_norm=72.231, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.494e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 06:52:04,901 (trainer:779) INFO: 81epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.678, loss_att=3.665, acc=0.975, loss=5.169, backward_time=0.254, grad_norm=66.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.492e-04, train_time=1.167
+[bmi2:0/4] 2024-07-17 06:56:08,725 (trainer:779) INFO: 81epoch:train:2066-2478batch: iter_time=2.132e-04, forward_time=0.158, loss_ctc=8.817, loss_att=3.708, acc=0.979, loss=5.241, backward_time=0.257, grad_norm=71.261, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.490e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 07:00:11,209 (trainer:779) INFO: 81epoch:train:2479-2891batch: iter_time=2.141e-04, forward_time=0.156, loss_ctc=8.643, loss_att=3.679, acc=0.978, loss=5.168, backward_time=0.255, grad_norm=72.786, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.489e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 07:04:14,258 (trainer:779) INFO: 81epoch:train:2892-3304batch: iter_time=7.279e-04, forward_time=0.158, loss_ctc=8.694, loss_att=3.664, acc=0.979, loss=5.173, backward_time=0.256, grad_norm=71.068, clip=100.000, loss_scale=5.418e+33, optim_step_time=0.036, optim0_lr0=5.487e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 07:05:44,188 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 07:07:27,091 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 07:08:18,075 (trainer:779) INFO: 81epoch:train:3305-3717batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.616, loss_att=3.651, acc=0.975, loss=5.140, backward_time=0.255, grad_norm=72.503, clip=100.000, loss_scale=6.554e+33, optim_step_time=0.036, optim0_lr0=5.485e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 07:12:21,770 (trainer:779) INFO: 81epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.683, loss_att=3.696, acc=0.976, loss=5.192, backward_time=0.256, grad_norm=70.889, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.484e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 07:16:25,020 (trainer:779) INFO: 81epoch:train:4131-4543batch: iter_time=3.728e-04, forward_time=0.157, loss_ctc=8.811, loss_att=3.723, acc=0.979, loss=5.250, backward_time=0.255, grad_norm=73.160, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.482e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 07:20:28,265 (trainer:779) INFO: 81epoch:train:4544-4956batch: iter_time=1.705e-04, forward_time=0.158, loss_ctc=8.683, loss_att=3.681, acc=0.979, loss=5.182, backward_time=0.256, grad_norm=74.307, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.480e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 07:24:31,693 (trainer:779) INFO: 81epoch:train:4957-5369batch: iter_time=1.662e-04, forward_time=0.157, loss_ctc=8.768, loss_att=3.702, acc=0.979, loss=5.221, backward_time=0.256, grad_norm=69.404, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.479e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 07:28:33,696 (trainer:779) INFO: 81epoch:train:5370-5782batch: iter_time=7.399e-04, forward_time=0.156, loss_ctc=8.761, loss_att=3.698, acc=0.978, loss=5.217, backward_time=0.254, grad_norm=70.457, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.477e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 07:32:36,982 (trainer:779) INFO: 81epoch:train:5783-6195batch: iter_time=9.327e-04, forward_time=0.158, loss_ctc=8.794, loss_att=3.716, acc=0.975, loss=5.239, backward_time=0.255, grad_norm=70.062, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.475e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 07:36:40,012 (trainer:779) INFO: 81epoch:train:6196-6608batch: iter_time=1.732e-04, forward_time=0.157, loss_ctc=8.770, loss_att=3.707, acc=0.980, loss=5.226, backward_time=0.256, grad_norm=71.147, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.473e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 07:40:43,853 (trainer:779) INFO: 81epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.156, loss_ctc=8.780, loss_att=3.759, acc=0.975, loss=5.265, backward_time=0.256, grad_norm=75.139, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.472e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 07:44:45,049 (trainer:779) INFO: 81epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.604, loss_att=3.665, acc=0.977, loss=5.147, backward_time=0.253, grad_norm=67.970, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.470e-04, train_time=1.167
+[bmi2:0/4] 2024-07-17 07:48:47,632 (trainer:779) INFO: 81epoch:train:7435-7847batch: iter_time=7.069e-04, forward_time=0.158, loss_ctc=8.686, loss_att=3.694, acc=0.977, loss=5.192, backward_time=0.256, grad_norm=71.011, clip=100.000, loss_scale=3.957e+33, optim_step_time=0.036, optim0_lr0=5.468e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 07:52:51,939 (trainer:779) INFO: 81epoch:train:7848-8260batch: iter_time=1.671e-04, forward_time=0.159, loss_ctc=8.693, loss_att=3.697, acc=0.980, loss=5.196, backward_time=0.256, grad_norm=73.508, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.467e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 07:55:27,175 (trainer:365) INFO: 81epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.706, loss_att=3.688, acc=0.977, loss=5.193, backward_time=0.255, grad_norm=70.956, clip=100.000, loss_scale=4.041e+33, optim_step_time=0.036, optim0_lr0=5.483e-04, train_time=1.187, time=1 hour, 21 minutes and 46.08 seconds, total_count=669627, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.816, cer_ctc=0.039, loss_att=5.732, acc=0.951, cer=0.031, wer=0.480, loss=6.957, time=43.27 seconds, total_count=2754, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 47.02 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 07:55:32,609 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 07:55:32,612 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/50epoch.pth
+[bmi2:0/4] 2024-07-17 07:55:32,612 (trainer:299) INFO: 82/100epoch started. Estimated time to finish: 1 day, 2 hours and 22 minutes
+[bmi2:0/4] 2024-07-17 07:58:39,264 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 08:00:18,116 (trainer:779) INFO: 82epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.649, loss_att=3.632, acc=0.979, loss=5.137, backward_time=0.255, grad_norm=73.013, clip=100.000, loss_scale=4.129e+33, optim_step_time=0.035, optim0_lr0=5.465e-04, train_time=1.384
+[bmi2:0/4] 2024-07-17 08:04:21,357 (trainer:779) INFO: 82epoch:train:414-826batch: iter_time=8.071e-04, forward_time=0.157, loss_ctc=8.609, loss_att=3.641, acc=0.978, loss=5.131, backward_time=0.255, grad_norm=77.562, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.463e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 08:07:39,224 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 08:08:23,088 (trainer:779) INFO: 82epoch:train:827-1239batch: iter_time=4.356e-04, forward_time=0.156, loss_ctc=8.542, loss_att=3.625, acc=0.979, loss=5.100, backward_time=0.254, grad_norm=72.994, clip=100.000, loss_scale=2.362e+33, optim_step_time=0.036, optim0_lr0=5.462e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 08:12:26,766 (trainer:779) INFO: 82epoch:train:1240-1652batch: iter_time=1.707e-04, forward_time=0.158, loss_ctc=8.613, loss_att=3.666, acc=0.981, loss=5.150, backward_time=0.255, grad_norm=70.358, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.460e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 08:16:29,701 (trainer:779) INFO: 82epoch:train:1653-2065batch: iter_time=5.103e-04, forward_time=0.158, loss_ctc=8.615, loss_att=3.659, acc=0.977, loss=5.146, backward_time=0.256, grad_norm=65.827, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.458e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 08:20:31,679 (trainer:779) INFO: 82epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.585, loss_att=3.645, acc=0.976, loss=5.127, backward_time=0.254, grad_norm=68.412, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.035, optim0_lr0=5.457e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 08:24:34,360 (trainer:779) INFO: 82epoch:train:2479-2891batch: iter_time=6.630e-04, forward_time=0.157, loss_ctc=8.614, loss_att=3.661, acc=0.976, loss=5.147, backward_time=0.256, grad_norm=75.406, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.035, optim0_lr0=5.455e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 08:28:37,269 (trainer:779) INFO: 82epoch:train:2892-3304batch: iter_time=1.669e-04, forward_time=0.156, loss_ctc=8.697, loss_att=3.652, acc=0.977, loss=5.166, backward_time=0.255, grad_norm=70.533, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.453e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 08:32:42,084 (trainer:779) INFO: 82epoch:train:3305-3717batch: iter_time=0.004, forward_time=0.158, loss_ctc=8.555, loss_att=3.656, acc=0.975, loss=5.126, backward_time=0.257, grad_norm=76.877, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.035, optim0_lr0=5.452e-04, train_time=1.186
+[bmi2:0/4] 2024-07-17 08:36:44,954 (trainer:779) INFO: 82epoch:train:3718-4130batch: iter_time=6.808e-04, forward_time=0.156, loss_ctc=8.643, loss_att=3.658, acc=0.977, loss=5.154, backward_time=0.255, grad_norm=68.145, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.450e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 08:40:47,467 (trainer:779) INFO: 82epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.762, loss_att=3.690, acc=0.976, loss=5.211, backward_time=0.254, grad_norm=78.316, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.448e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 08:44:50,328 (trainer:779) INFO: 82epoch:train:4544-4956batch: iter_time=2.198e-04, forward_time=0.157, loss_ctc=8.667, loss_att=3.649, acc=0.978, loss=5.154, backward_time=0.255, grad_norm=70.813, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.035, optim0_lr0=5.447e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 08:48:53,004 (trainer:779) INFO: 82epoch:train:4957-5369batch: iter_time=3.030e-04, forward_time=0.157, loss_ctc=8.738, loss_att=3.714, acc=0.981, loss=5.222, backward_time=0.255, grad_norm=73.844, clip=100.000, loss_scale=1.941e+33, optim_step_time=0.036, optim0_lr0=5.445e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 08:52:54,938 (trainer:779) INFO: 82epoch:train:5370-5782batch: iter_time=8.218e-04, forward_time=0.155, loss_ctc=8.694, loss_att=3.696, acc=0.979, loss=5.196, backward_time=0.254, grad_norm=72.703, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.443e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 08:56:58,874 (trainer:779) INFO: 82epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.654, loss_att=3.663, acc=0.976, loss=5.160, backward_time=0.256, grad_norm=69.266, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.442e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 09:01:02,489 (trainer:779) INFO: 82epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.659, loss_att=3.670, acc=0.975, loss=5.167, backward_time=0.255, grad_norm=72.623, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.440e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 09:05:04,739 (trainer:779) INFO: 82epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.712, loss_att=3.705, acc=0.974, loss=5.207, backward_time=0.255, grad_norm=71.272, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.438e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 09:09:09,050 (trainer:779) INFO: 82epoch:train:7022-7434batch: iter_time=5.771e-04, forward_time=0.158, loss_ctc=8.651, loss_att=3.641, acc=0.978, loss=5.144, backward_time=0.256, grad_norm=74.035, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.437e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 09:13:12,369 (trainer:779) INFO: 82epoch:train:7435-7847batch: iter_time=5.382e-04, forward_time=0.158, loss_ctc=8.809, loss_att=3.722, acc=0.980, loss=5.248, backward_time=0.256, grad_norm=77.344, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.435e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 09:17:17,001 (trainer:779) INFO: 82epoch:train:7848-8260batch: iter_time=3.816e-04, forward_time=0.158, loss_ctc=8.684, loss_att=3.647, acc=0.979, loss=5.158, backward_time=0.256, grad_norm=78.416, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.433e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 09:19:43,012 (trainer:365) INFO: 82epoch results: [train] iter_time=9.829e-04, forward_time=0.157, loss_ctc=8.657, loss_att=3.664, acc=0.978, loss=5.162, backward_time=0.255, grad_norm=72.877, clip=100.000, loss_scale=2.044e+33, optim_step_time=0.036, optim0_lr0=5.449e-04, train_time=1.187, time=1 hour, 21 minutes and 49.38 seconds, total_count=677894, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.042, cer_ctc=0.040, loss_att=6.115, acc=0.949, cer=0.033, wer=0.487, loss=7.293, time=39 seconds, total_count=2788, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 42.01 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 09:19:47,548 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 09:19:47,549 (trainer:299) INFO: 83/100epoch started. Estimated time to finish: 1 day, 1 hour and 1 minute
+[bmi2:0/4] 2024-07-17 09:24:28,105 (trainer:779) INFO: 83epoch:train:1-413batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.674, loss_att=3.669, acc=0.979, loss=5.171, backward_time=0.255, grad_norm=74.447, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.432e-04, train_time=1.359
+[bmi2:0/4] 2024-07-17 09:28:28,514 (trainer:779) INFO: 83epoch:train:414-826batch: iter_time=7.478e-04, forward_time=0.154, loss_ctc=8.677, loss_att=3.676, acc=0.978, loss=5.176, backward_time=0.253, grad_norm=73.618, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.430e-04, train_time=1.164
+[bmi2:0/4] 2024-07-17 09:32:31,426 (trainer:779) INFO: 83epoch:train:827-1239batch: iter_time=3.016e-04, forward_time=0.157, loss_ctc=8.643, loss_att=3.674, acc=0.979, loss=5.165, backward_time=0.256, grad_norm=71.854, clip=100.000, loss_scale=4.739e+33, optim_step_time=0.035, optim0_lr0=5.428e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 09:36:34,145 (trainer:779) INFO: 83epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.447, loss_att=3.590, acc=0.979, loss=5.047, backward_time=0.254, grad_norm=74.075, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.427e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 09:40:36,229 (trainer:779) INFO: 83epoch:train:1653-2065batch: iter_time=9.829e-04, forward_time=0.157, loss_ctc=8.576, loss_att=3.621, acc=0.978, loss=5.108, backward_time=0.255, grad_norm=71.203, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.425e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 09:44:38,907 (trainer:779) INFO: 83epoch:train:2066-2478batch: iter_time=4.560e-04, forward_time=0.156, loss_ctc=8.547, loss_att=3.634, acc=0.977, loss=5.108, backward_time=0.255, grad_norm=70.404, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.423e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 09:45:31,874 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 09:48:41,501 (trainer:779) INFO: 83epoch:train:2479-2891batch: iter_time=9.014e-04, forward_time=0.157, loss_ctc=8.598, loss_att=3.668, acc=0.977, loss=5.147, backward_time=0.255, grad_norm=75.912, clip=100.000, loss_scale=3.153e+33, optim_step_time=0.035, optim0_lr0=5.422e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 09:52:45,523 (trainer:779) INFO: 83epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.726, loss_att=3.681, acc=0.978, loss=5.194, backward_time=0.256, grad_norm=70.904, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.420e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 09:56:48,423 (trainer:779) INFO: 83epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.734, loss_att=3.684, acc=0.977, loss=5.199, backward_time=0.255, grad_norm=74.956, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.418e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 10:00:52,469 (trainer:779) INFO: 83epoch:train:3718-4130batch: iter_time=2.209e-04, forward_time=0.157, loss_ctc=8.620, loss_att=3.633, acc=0.979, loss=5.129, backward_time=0.257, grad_norm=69.981, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.417e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 10:04:55,100 (trainer:779) INFO: 83epoch:train:4131-4543batch: iter_time=4.975e-04, forward_time=0.156, loss_ctc=8.687, loss_att=3.655, acc=0.979, loss=5.164, backward_time=0.255, grad_norm=68.979, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.415e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 10:08:59,289 (trainer:779) INFO: 83epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.793, loss_att=3.708, acc=0.977, loss=5.233, backward_time=0.255, grad_norm=76.977, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.414e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 10:13:02,526 (trainer:779) INFO: 83epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.760, loss_att=3.685, acc=0.977, loss=5.208, backward_time=0.255, grad_norm=74.738, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.412e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 10:17:07,490 (trainer:779) INFO: 83epoch:train:5370-5782batch: iter_time=5.140e-04, forward_time=0.158, loss_ctc=8.726, loss_att=3.699, acc=0.979, loss=5.207, backward_time=0.257, grad_norm=72.068, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.410e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 10:21:11,751 (trainer:779) INFO: 83epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.712, loss_att=3.683, acc=0.977, loss=5.192, backward_time=0.256, grad_norm=72.661, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.409e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 10:25:14,744 (trainer:779) INFO: 83epoch:train:6196-6608batch: iter_time=8.946e-04, forward_time=0.156, loss_ctc=8.708, loss_att=3.693, acc=0.975, loss=5.197, backward_time=0.255, grad_norm=70.717, clip=100.000, loss_scale=2.847e+33, optim_step_time=0.036, optim0_lr0=5.407e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 10:29:16,992 (trainer:779) INFO: 83epoch:train:6609-7021batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.730, loss_att=3.693, acc=0.974, loss=5.204, backward_time=0.254, grad_norm=71.631, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.405e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 10:33:18,563 (trainer:779) INFO: 83epoch:train:7022-7434batch: iter_time=5.213e-04, forward_time=0.156, loss_ctc=8.642, loss_att=3.639, acc=0.976, loss=5.140, backward_time=0.253, grad_norm=71.033, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.404e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 10:37:22,823 (trainer:779) INFO: 83epoch:train:7435-7847batch: iter_time=9.108e-04, forward_time=0.158, loss_ctc=8.716, loss_att=3.642, acc=0.978, loss=5.165, backward_time=0.256, grad_norm=74.226, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.402e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 10:41:26,794 (trainer:779) INFO: 83epoch:train:7848-8260batch: iter_time=5.113e-04, forward_time=0.158, loss_ctc=8.762, loss_att=3.660, acc=0.979, loss=5.191, backward_time=0.256, grad_norm=71.502, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.400e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 10:44:09,209 (trainer:365) INFO: 83epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.674, loss_att=3.664, acc=0.978, loss=5.167, backward_time=0.255, grad_norm=72.581, clip=100.000, loss_scale=3.654e+33, optim_step_time=0.036, optim0_lr0=5.416e-04, train_time=1.186, time=1 hour, 21 minutes and 44.25 seconds, total_count=686161, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.043, cer_ctc=0.040, loss_att=6.065, acc=0.949, cer=0.033, wer=0.482, loss=7.258, time=44.19 seconds, total_count=2822, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 53.22 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 10:44:14,797 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 10:44:14,870 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/82epoch.pth
+[bmi2:0/4] 2024-07-17 10:44:14,870 (trainer:299) INFO: 84/100epoch started. Estimated time to finish: 23 hours, 39 minutes and 40.09 seconds
+[bmi2:0/4] 2024-07-17 10:48:59,755 (trainer:779) INFO: 84epoch:train:1-413batch: iter_time=0.001, forward_time=0.154, loss_ctc=8.465, loss_att=3.607, acc=0.980, loss=5.064, backward_time=0.255, grad_norm=69.330, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.399e-04, train_time=1.381
+[bmi2:0/4] 2024-07-17 10:53:03,442 (trainer:779) INFO: 84epoch:train:414-826batch: iter_time=4.651e-04, forward_time=0.157, loss_ctc=8.665, loss_att=3.659, acc=0.977, loss=5.161, backward_time=0.257, grad_norm=75.031, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.397e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 10:57:06,529 (trainer:779) INFO: 84epoch:train:827-1239batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.677, loss_att=3.643, acc=0.978, loss=5.153, backward_time=0.255, grad_norm=68.191, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.396e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 11:01:09,351 (trainer:779) INFO: 84epoch:train:1240-1652batch: iter_time=8.747e-04, forward_time=0.157, loss_ctc=8.483, loss_att=3.597, acc=0.980, loss=5.063, backward_time=0.255, grad_norm=74.650, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.394e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 11:02:00,895 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 11:05:11,484 (trainer:779) INFO: 84epoch:train:1653-2065batch: iter_time=1.695e-04, forward_time=0.157, loss_ctc=8.575, loss_att=3.678, acc=0.978, loss=5.147, backward_time=0.255, grad_norm=70.806, clip=100.000, loss_scale=3.141e+33, optim_step_time=0.036, optim0_lr0=5.392e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 11:09:14,827 (trainer:779) INFO: 84epoch:train:2066-2478batch: iter_time=2.035e-04, forward_time=0.155, loss_ctc=8.561, loss_att=3.638, acc=0.979, loss=5.115, backward_time=0.257, grad_norm=67.202, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.391e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 11:13:17,579 (trainer:779) INFO: 84epoch:train:2479-2891batch: iter_time=6.345e-04, forward_time=0.156, loss_ctc=8.569, loss_att=3.620, acc=0.978, loss=5.104, backward_time=0.256, grad_norm=72.379, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.389e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 11:17:20,656 (trainer:779) INFO: 84epoch:train:2892-3304batch: iter_time=9.183e-04, forward_time=0.156, loss_ctc=8.529, loss_att=3.606, acc=0.977, loss=5.083, backward_time=0.255, grad_norm=67.742, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.387e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 11:21:24,248 (trainer:779) INFO: 84epoch:train:3305-3717batch: iter_time=1.901e-04, forward_time=0.159, loss_ctc=8.554, loss_att=3.621, acc=0.979, loss=5.100, backward_time=0.259, grad_norm=69.878, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.037, optim0_lr0=5.386e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 11:25:28,733 (trainer:779) INFO: 84epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.647, loss_att=3.661, acc=0.975, loss=5.156, backward_time=0.258, grad_norm=71.419, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.384e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 11:29:32,461 (trainer:779) INFO: 84epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.585, loss_att=3.645, acc=0.974, loss=5.127, backward_time=0.256, grad_norm=69.658, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.383e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 11:33:35,124 (trainer:779) INFO: 84epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.686, loss_att=3.674, acc=0.976, loss=5.177, backward_time=0.254, grad_norm=72.301, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.381e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 11:37:38,317 (trainer:779) INFO: 84epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.653, loss_att=3.688, acc=0.977, loss=5.177, backward_time=0.255, grad_norm=71.508, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.379e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 11:41:41,885 (trainer:779) INFO: 84epoch:train:5370-5782batch: iter_time=9.781e-04, forward_time=0.157, loss_ctc=8.563, loss_att=3.607, acc=0.976, loss=5.094, backward_time=0.255, grad_norm=72.198, clip=100.000, loss_scale=2.860e+33, optim_step_time=0.036, optim0_lr0=5.378e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 11:45:45,620 (trainer:779) INFO: 84epoch:train:5783-6195batch: iter_time=8.304e-04, forward_time=0.159, loss_ctc=8.474, loss_att=3.590, acc=0.975, loss=5.055, backward_time=0.255, grad_norm=67.991, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.376e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 11:49:50,036 (trainer:779) INFO: 84epoch:train:6196-6608batch: iter_time=1.746e-04, forward_time=0.158, loss_ctc=8.628, loss_att=3.622, acc=0.980, loss=5.124, backward_time=0.257, grad_norm=70.275, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.375e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 11:53:53,460 (trainer:779) INFO: 84epoch:train:6609-7021batch: iter_time=9.969e-04, forward_time=0.158, loss_ctc=8.531, loss_att=3.610, acc=0.979, loss=5.086, backward_time=0.255, grad_norm=75.726, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.373e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 11:57:55,138 (trainer:779) INFO: 84epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.567, loss_att=3.603, acc=0.979, loss=5.092, backward_time=0.255, grad_norm=79.764, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.371e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 12:01:57,113 (trainer:779) INFO: 84epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.155, loss_ctc=8.700, loss_att=3.681, acc=0.977, loss=5.187, backward_time=0.253, grad_norm=75.401, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.370e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 12:06:00,124 (trainer:779) INFO: 84epoch:train:7848-8260batch: iter_time=7.341e-04, forward_time=0.157, loss_ctc=8.626, loss_att=3.635, acc=0.980, loss=5.132, backward_time=0.255, grad_norm=77.953, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.368e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 12:08:39,932 (trainer:365) INFO: 84epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.587, loss_att=3.634, acc=0.978, loss=5.120, backward_time=0.256, grad_norm=71.962, clip=100.000, loss_scale=3.936e+33, optim_step_time=0.036, optim0_lr0=5.383e-04, train_time=1.187, time=1 hour, 21 minutes and 50.12 seconds, total_count=694428, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.045, cer_ctc=0.040, loss_att=6.120, acc=0.949, cer=0.033, wer=0.485, loss=7.298, time=41.98 seconds, total_count=2856, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 52.96 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 12:08:45,139 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 12:08:45,196 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/83epoch.pth
+[bmi2:0/4] 2024-07-17 12:08:45,197 (trainer:299) INFO: 85/100epoch started. Estimated time to finish: 22 hours, 17 minutes and 36.38 seconds
+[bmi2:0/4] 2024-07-17 12:13:29,813 (trainer:779) INFO: 85epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.642, loss_att=3.640, acc=0.977, loss=5.141, backward_time=0.255, grad_norm=74.404, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.367e-04, train_time=1.379
+[bmi2:0/4] 2024-07-17 12:17:33,025 (trainer:779) INFO: 85epoch:train:414-826batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.668, loss_att=3.669, acc=0.978, loss=5.169, backward_time=0.255, grad_norm=79.062, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.365e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 12:21:35,120 (trainer:779) INFO: 85epoch:train:827-1239batch: iter_time=7.507e-04, forward_time=0.156, loss_ctc=8.786, loss_att=3.704, acc=0.975, loss=5.229, backward_time=0.254, grad_norm=80.376, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.363e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 12:25:37,654 (trainer:779) INFO: 85epoch:train:1240-1652batch: iter_time=2.422e-04, forward_time=0.156, loss_ctc=8.819, loss_att=3.690, acc=0.976, loss=5.228, backward_time=0.254, grad_norm=74.300, clip=100.000, loss_scale=7.425e+33, optim_step_time=0.036, optim0_lr0=5.362e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 12:29:15,501 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 12:29:39,227 (trainer:779) INFO: 85epoch:train:1653-2065batch: iter_time=1.803e-04, forward_time=0.156, loss_ctc=8.650, loss_att=3.636, acc=0.978, loss=5.140, backward_time=0.255, grad_norm=74.084, clip=100.000, loss_scale=9.878e+33, optim_step_time=0.035, optim0_lr0=5.360e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 12:33:42,784 (trainer:779) INFO: 85epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.674, loss_att=3.676, acc=0.977, loss=5.175, backward_time=0.257, grad_norm=71.498, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.359e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 12:37:44,436 (trainer:779) INFO: 85epoch:train:2479-2891batch: iter_time=3.451e-04, forward_time=0.155, loss_ctc=8.623, loss_att=3.626, acc=0.978, loss=5.125, backward_time=0.255, grad_norm=81.897, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.357e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 12:41:47,212 (trainer:779) INFO: 85epoch:train:2892-3304batch: iter_time=5.164e-04, forward_time=0.156, loss_ctc=8.594, loss_att=3.620, acc=0.979, loss=5.112, backward_time=0.255, grad_norm=75.988, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.355e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 12:45:49,847 (trainer:779) INFO: 85epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.729, loss_att=3.671, acc=0.975, loss=5.189, backward_time=0.256, grad_norm=74.840, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.354e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 12:49:52,962 (trainer:779) INFO: 85epoch:train:3718-4130batch: iter_time=0.003, forward_time=0.155, loss_ctc=8.550, loss_att=3.618, acc=0.978, loss=5.097, backward_time=0.256, grad_norm=73.864, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.352e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 12:52:04,397 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 12:53:54,105 (trainer:779) INFO: 85epoch:train:4131-4543batch: iter_time=2.542e-04, forward_time=0.155, loss_ctc=8.525, loss_att=3.601, acc=0.978, loss=5.079, backward_time=0.255, grad_norm=73.955, clip=100.000, loss_scale=4.015e+33, optim_step_time=0.035, optim0_lr0=5.351e-04, train_time=1.168
+[bmi2:0/4] 2024-07-17 12:57:56,581 (trainer:779) INFO: 85epoch:train:4544-4956batch: iter_time=7.529e-04, forward_time=0.155, loss_ctc=8.678, loss_att=3.646, acc=0.979, loss=5.156, backward_time=0.255, grad_norm=72.959, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.349e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 13:02:00,258 (trainer:779) INFO: 85epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.697, loss_att=3.651, acc=0.978, loss=5.165, backward_time=0.256, grad_norm=74.311, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.348e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 13:06:03,594 (trainer:779) INFO: 85epoch:train:5370-5782batch: iter_time=8.058e-04, forward_time=0.156, loss_ctc=8.543, loss_att=3.640, acc=0.978, loss=5.111, backward_time=0.255, grad_norm=70.380, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.346e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 13:10:05,800 (trainer:779) INFO: 85epoch:train:5783-6195batch: iter_time=6.926e-04, forward_time=0.158, loss_ctc=8.633, loss_att=3.640, acc=0.979, loss=5.138, backward_time=0.254, grad_norm=72.792, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.344e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 13:14:09,756 (trainer:779) INFO: 85epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.629, loss_att=3.661, acc=0.977, loss=5.151, backward_time=0.255, grad_norm=69.701, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.343e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 13:18:12,144 (trainer:779) INFO: 85epoch:train:6609-7021batch: iter_time=8.434e-04, forward_time=0.156, loss_ctc=8.640, loss_att=3.653, acc=0.979, loss=5.149, backward_time=0.255, grad_norm=73.789, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.341e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 13:22:15,138 (trainer:779) INFO: 85epoch:train:7022-7434batch: iter_time=2.007e-04, forward_time=0.157, loss_ctc=8.466, loss_att=3.586, acc=0.979, loss=5.050, backward_time=0.255, grad_norm=74.007, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.340e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 13:26:19,510 (trainer:779) INFO: 85epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.471, loss_att=3.585, acc=0.977, loss=5.051, backward_time=0.256, grad_norm=66.794, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.338e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 13:30:23,167 (trainer:779) INFO: 85epoch:train:7848-8260batch: iter_time=1.974e-04, forward_time=0.158, loss_ctc=8.577, loss_att=3.604, acc=0.980, loss=5.096, backward_time=0.256, grad_norm=69.322, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.337e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 13:33:01,782 (trainer:365) INFO: 85epoch results: [train] iter_time=0.001, forward_time=0.156, loss_ctc=8.631, loss_att=3.641, acc=0.978, loss=5.138, backward_time=0.255, grad_norm=73.900, clip=100.000, loss_scale=4.308e+33, optim_step_time=0.036, optim0_lr0=5.352e-04, train_time=1.186, time=1 hour, 21 minutes and 42.84 seconds, total_count=702695, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.096, cer_ctc=0.039, loss_att=5.943, acc=0.948, cer=0.032, wer=0.484, loss=7.189, time=44.29 seconds, total_count=2890, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 49.44 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 13:33:06,168 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 13:33:06,226 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/84epoch.pth
+[bmi2:0/4] 2024-07-17 13:33:06,226 (trainer:299) INFO: 86/100epoch started. Estimated time to finish: 20 hours, 54 minutes and 56.61 seconds
+[bmi2:0/4] 2024-07-17 13:37:48,705 (trainer:779) INFO: 86epoch:train:1-413batch: iter_time=0.003, forward_time=0.154, loss_ctc=8.375, loss_att=3.573, acc=0.976, loss=5.014, backward_time=0.253, grad_norm=69.903, clip=100.000, loss_scale=4.625e+33, optim_step_time=0.036, optim0_lr0=5.335e-04, train_time=1.369
+[bmi2:0/4] 2024-07-17 13:41:51,504 (trainer:779) INFO: 86epoch:train:414-826batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.365, loss_att=3.542, acc=0.979, loss=4.989, backward_time=0.256, grad_norm=70.505, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.333e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 13:45:54,648 (trainer:779) INFO: 86epoch:train:827-1239batch: iter_time=9.212e-04, forward_time=0.157, loss_ctc=8.559, loss_att=3.616, acc=0.980, loss=5.099, backward_time=0.255, grad_norm=69.793, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.332e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 13:49:57,503 (trainer:779) INFO: 86epoch:train:1240-1652batch: iter_time=6.695e-04, forward_time=0.156, loss_ctc=8.387, loss_att=3.547, acc=0.979, loss=4.999, backward_time=0.256, grad_norm=65.740, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.330e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 13:53:59,724 (trainer:779) INFO: 86epoch:train:1653-2065batch: iter_time=7.279e-04, forward_time=0.157, loss_ctc=8.505, loss_att=3.597, acc=0.979, loss=5.069, backward_time=0.254, grad_norm=71.216, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.329e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 13:58:01,567 (trainer:779) INFO: 86epoch:train:2066-2478batch: iter_time=7.333e-04, forward_time=0.156, loss_ctc=8.447, loss_att=3.581, acc=0.978, loss=5.041, backward_time=0.253, grad_norm=68.384, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.327e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 14:02:04,650 (trainer:779) INFO: 86epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.452, loss_att=3.585, acc=0.978, loss=5.045, backward_time=0.256, grad_norm=68.173, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.326e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 14:06:09,402 (trainer:779) INFO: 86epoch:train:2892-3304batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.508, loss_att=3.617, acc=0.972, loss=5.084, backward_time=0.256, grad_norm=69.595, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.324e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 14:10:11,528 (trainer:779) INFO: 86epoch:train:3305-3717batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.438, loss_att=3.572, acc=0.975, loss=5.032, backward_time=0.255, grad_norm=72.141, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.322e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 14:14:15,065 (trainer:779) INFO: 86epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.519, loss_att=3.614, acc=0.979, loss=5.085, backward_time=0.255, grad_norm=64.844, clip=100.000, loss_scale=5.694e+33, optim_step_time=0.036, optim0_lr0=5.321e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 14:17:21,977 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 14:18:16,579 (trainer:779) INFO: 86epoch:train:4131-4543batch: iter_time=5.524e-04, forward_time=0.155, loss_ctc=8.476, loss_att=3.574, acc=0.978, loss=5.045, backward_time=0.254, grad_norm=68.934, clip=100.000, loss_scale=9.219e+33, optim_step_time=0.035, optim0_lr0=5.319e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 14:22:21,130 (trainer:779) INFO: 86epoch:train:4544-4956batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.579, loss_att=3.629, acc=0.979, loss=5.114, backward_time=0.256, grad_norm=69.871, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.318e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 14:26:24,625 (trainer:779) INFO: 86epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.556, loss_att=3.618, acc=0.977, loss=5.100, backward_time=0.255, grad_norm=68.761, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.316e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 14:30:27,349 (trainer:779) INFO: 86epoch:train:5370-5782batch: iter_time=9.899e-04, forward_time=0.156, loss_ctc=8.472, loss_att=3.571, acc=0.978, loss=5.042, backward_time=0.255, grad_norm=68.166, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.315e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 14:33:33,255 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 14:34:27,167 (trainer:779) INFO: 86epoch:train:5783-6195batch: iter_time=7.323e-04, forward_time=0.153, loss_ctc=8.517, loss_att=3.621, acc=0.980, loss=5.090, backward_time=0.254, grad_norm=72.495, clip=100.000, loss_scale=4.610e+33, optim_step_time=0.035, optim0_lr0=5.313e-04, train_time=1.162
+[bmi2:0/4] 2024-07-17 14:38:30,720 (trainer:779) INFO: 86epoch:train:6196-6608batch: iter_time=1.777e-04, forward_time=0.157, loss_ctc=8.439, loss_att=3.576, acc=0.980, loss=5.035, backward_time=0.256, grad_norm=73.463, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.312e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 14:42:34,399 (trainer:779) INFO: 86epoch:train:6609-7021batch: iter_time=8.797e-04, forward_time=0.158, loss_ctc=8.476, loss_att=3.582, acc=0.978, loss=5.050, backward_time=0.256, grad_norm=69.529, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.310e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 14:46:38,339 (trainer:779) INFO: 86epoch:train:7022-7434batch: iter_time=1.799e-04, forward_time=0.158, loss_ctc=8.536, loss_att=3.575, acc=0.979, loss=5.063, backward_time=0.256, grad_norm=76.692, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.309e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 14:50:40,136 (trainer:779) INFO: 86epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.528, loss_att=3.600, acc=0.978, loss=5.079, backward_time=0.254, grad_norm=69.863, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.307e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 14:54:43,277 (trainer:779) INFO: 86epoch:train:7848-8260batch: iter_time=1.759e-04, forward_time=0.158, loss_ctc=8.616, loss_att=3.618, acc=0.980, loss=5.117, backward_time=0.256, grad_norm=71.293, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.305e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 14:57:25,570 (trainer:365) INFO: 86epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.487, loss_att=3.590, acc=0.978, loss=5.059, backward_time=0.255, grad_norm=69.962, clip=100.000, loss_scale=4.709e+33, optim_step_time=0.036, optim0_lr0=5.320e-04, train_time=1.186, time=1 hour, 21 minutes and 42 seconds, total_count=710962, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.962, cer_ctc=0.039, loss_att=5.961, acc=0.951, cer=0.032, wer=0.480, loss=7.161, time=42.88 seconds, total_count=2924, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 54.47 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 14:57:29,447 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 14:57:29,502 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/54epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/85epoch.pth
+[bmi2:0/4] 2024-07-17 14:57:29,502 (trainer:299) INFO: 87/100epoch started. Estimated time to finish: 19 hours, 32 minutes and 3.69 seconds
+[bmi2:0/4] 2024-07-17 15:02:10,812 (trainer:779) INFO: 87epoch:train:1-413batch: iter_time=0.002, forward_time=0.154, loss_ctc=8.346, loss_att=3.545, acc=0.976, loss=4.985, backward_time=0.254, grad_norm=66.965, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.304e-04, train_time=1.363
+[bmi2:0/4] 2024-07-17 15:06:14,543 (trainer:779) INFO: 87epoch:train:414-826batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.297, loss_att=3.522, acc=0.980, loss=4.955, backward_time=0.255, grad_norm=66.888, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.302e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 15:10:17,770 (trainer:779) INFO: 87epoch:train:827-1239batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.487, loss_att=3.608, acc=0.975, loss=5.072, backward_time=0.254, grad_norm=66.405, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.301e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 15:14:21,782 (trainer:779) INFO: 87epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.465, loss_att=3.573, acc=0.978, loss=5.040, backward_time=0.256, grad_norm=68.630, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.299e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 15:18:26,365 (trainer:779) INFO: 87epoch:train:1653-2065batch: iter_time=1.919e-04, forward_time=0.159, loss_ctc=8.497, loss_att=3.575, acc=0.981, loss=5.052, backward_time=0.257, grad_norm=71.887, clip=100.000, loss_scale=4.033e+33, optim_step_time=0.037, optim0_lr0=5.298e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 15:22:29,476 (trainer:779) INFO: 87epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.579, loss_att=3.604, acc=0.980, loss=5.097, backward_time=0.255, grad_norm=72.820, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.296e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 15:26:31,764 (trainer:779) INFO: 87epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.579, loss_att=3.600, acc=0.979, loss=5.094, backward_time=0.255, grad_norm=73.524, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.295e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 15:30:34,516 (trainer:779) INFO: 87epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.498, loss_att=3.577, acc=0.979, loss=5.054, backward_time=0.254, grad_norm=71.782, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.293e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 15:33:19,485 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 15:34:35,845 (trainer:779) INFO: 87epoch:train:3305-3717batch: iter_time=3.342e-04, forward_time=0.155, loss_ctc=8.631, loss_att=3.616, acc=0.980, loss=5.121, backward_time=0.254, grad_norm=75.767, clip=100.000, loss_scale=4.369e+33, optim_step_time=0.035, optim0_lr0=5.292e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 15:38:39,495 (trainer:779) INFO: 87epoch:train:3718-4130batch: iter_time=3.224e-04, forward_time=0.157, loss_ctc=8.415, loss_att=3.553, acc=0.979, loss=5.012, backward_time=0.256, grad_norm=71.142, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.290e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 15:42:40,288 (trainer:779) INFO: 87epoch:train:4131-4543batch: iter_time=8.250e-04, forward_time=0.155, loss_ctc=8.577, loss_att=3.605, acc=0.978, loss=5.097, backward_time=0.254, grad_norm=73.690, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.289e-04, train_time=1.166
+[bmi2:0/4] 2024-07-17 15:46:43,010 (trainer:779) INFO: 87epoch:train:4544-4956batch: iter_time=4.624e-04, forward_time=0.156, loss_ctc=8.547, loss_att=3.585, acc=0.979, loss=5.074, backward_time=0.254, grad_norm=71.891, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.287e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 15:50:45,493 (trainer:779) INFO: 87epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.608, loss_att=3.656, acc=0.976, loss=5.142, backward_time=0.254, grad_norm=73.970, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.285e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 15:54:48,759 (trainer:779) INFO: 87epoch:train:5370-5782batch: iter_time=2.477e-04, forward_time=0.156, loss_ctc=8.541, loss_att=3.580, acc=0.978, loss=5.068, backward_time=0.254, grad_norm=74.183, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.284e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 15:58:51,600 (trainer:779) INFO: 87epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.465, loss_att=3.586, acc=0.977, loss=5.049, backward_time=0.255, grad_norm=84.303, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.282e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 16:02:53,697 (trainer:779) INFO: 87epoch:train:6196-6608batch: iter_time=1.750e-04, forward_time=0.156, loss_ctc=8.646, loss_att=3.622, acc=0.978, loss=5.130, backward_time=0.253, grad_norm=76.380, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.281e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 16:06:55,274 (trainer:779) INFO: 87epoch:train:6609-7021batch: iter_time=4.871e-04, forward_time=0.156, loss_ctc=8.482, loss_att=3.574, acc=0.977, loss=5.047, backward_time=0.254, grad_norm=70.605, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.279e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 16:10:57,786 (trainer:779) INFO: 87epoch:train:7022-7434batch: iter_time=1.811e-04, forward_time=0.156, loss_ctc=8.489, loss_att=3.573, acc=0.979, loss=5.048, backward_time=0.254, grad_norm=71.924, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.278e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 16:15:01,241 (trainer:779) INFO: 87epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.156, loss_ctc=8.497, loss_att=3.598, acc=0.979, loss=5.067, backward_time=0.256, grad_norm=73.610, clip=100.000, loss_scale=4.234e+33, optim_step_time=0.035, optim0_lr0=5.276e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 16:19:02,495 (trainer:779) INFO: 87epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.596, loss_att=3.622, acc=0.974, loss=5.114, backward_time=0.253, grad_norm=73.341, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.275e-04, train_time=1.168
+[bmi2:0/4] 2024-07-17 16:21:44,941 (trainer:365) INFO: 87epoch results: [train] iter_time=0.001, forward_time=0.156, loss_ctc=8.512, loss_att=3.589, acc=0.978, loss=5.066, backward_time=0.255, grad_norm=72.473, clip=100.000, loss_scale=3.359e+33, optim_step_time=0.036, optim0_lr0=5.289e-04, train_time=1.185, time=1 hour, 21 minutes and 37.82 seconds, total_count=719229, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.145, cer_ctc=0.039, loss_att=6.024, acc=0.950, cer=0.032, wer=0.477, loss=7.261, time=43.67 seconds, total_count=2958, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 53.95 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 16:21:49,075 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 16:21:49,076 (trainer:299) INFO: 88/100epoch started. Estimated time to finish: 18 hours, 8 minutes and 54.42 seconds
+[bmi2:0/4] 2024-07-17 16:26:33,666 (trainer:779) INFO: 88epoch:train:1-413batch: iter_time=0.003, forward_time=0.155, loss_ctc=8.386, loss_att=3.559, acc=0.975, loss=5.007, backward_time=0.255, grad_norm=66.867, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.273e-04, train_time=1.379
+[bmi2:0/4] 2024-07-17 16:30:36,076 (trainer:779) INFO: 88epoch:train:414-826batch: iter_time=7.056e-04, forward_time=0.156, loss_ctc=8.475, loss_att=3.599, acc=0.981, loss=5.062, backward_time=0.254, grad_norm=75.114, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.272e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 16:34:38,410 (trainer:779) INFO: 88epoch:train:827-1239batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.518, loss_att=3.585, acc=0.980, loss=5.065, backward_time=0.254, grad_norm=76.716, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.270e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 16:38:40,704 (trainer:779) INFO: 88epoch:train:1240-1652batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.445, loss_att=3.548, acc=0.974, loss=5.017, backward_time=0.253, grad_norm=67.422, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.269e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 16:42:43,171 (trainer:779) INFO: 88epoch:train:1653-2065batch: iter_time=3.492e-04, forward_time=0.156, loss_ctc=8.398, loss_att=3.530, acc=0.978, loss=4.991, backward_time=0.255, grad_norm=71.024, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.267e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 16:46:46,068 (trainer:779) INFO: 88epoch:train:2066-2478batch: iter_time=3.800e-04, forward_time=0.157, loss_ctc=8.376, loss_att=3.540, acc=0.977, loss=4.991, backward_time=0.255, grad_norm=73.942, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.266e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 16:50:49,397 (trainer:779) INFO: 88epoch:train:2479-2891batch: iter_time=6.854e-04, forward_time=0.158, loss_ctc=8.504, loss_att=3.579, acc=0.980, loss=5.057, backward_time=0.255, grad_norm=73.922, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.264e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 16:54:52,246 (trainer:779) INFO: 88epoch:train:2892-3304batch: iter_time=1.720e-04, forward_time=0.157, loss_ctc=8.579, loss_att=3.605, acc=0.980, loss=5.097, backward_time=0.255, grad_norm=75.600, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.263e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 16:58:54,900 (trainer:779) INFO: 88epoch:train:3305-3717batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.545, loss_att=3.592, acc=0.978, loss=5.078, backward_time=0.255, grad_norm=79.363, clip=100.000, loss_scale=1.018e+34, optim_step_time=0.035, optim0_lr0=5.261e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 17:02:56,537 (trainer:779) INFO: 88epoch:train:3718-4130batch: iter_time=0.003, forward_time=0.155, loss_ctc=8.425, loss_att=3.566, acc=0.976, loss=5.024, backward_time=0.253, grad_norm=69.966, clip=100.000, loss_scale=1.038e+34, optim_step_time=0.035, optim0_lr0=5.260e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 17:03:40,891 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 17:06:57,064 (trainer:779) INFO: 88epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.153, loss_ctc=8.454, loss_att=3.592, acc=0.979, loss=5.050, backward_time=0.255, grad_norm=71.089, clip=100.000, loss_scale=6.129e+33, optim_step_time=0.035, optim0_lr0=5.258e-04, train_time=1.165
+[bmi2:0/4] 2024-07-17 17:10:59,887 (trainer:779) INFO: 88epoch:train:4544-4956batch: iter_time=4.827e-04, forward_time=0.156, loss_ctc=8.422, loss_att=3.554, acc=0.980, loss=5.015, backward_time=0.255, grad_norm=69.998, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.257e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 17:15:03,141 (trainer:779) INFO: 88epoch:train:4957-5369batch: iter_time=7.750e-04, forward_time=0.158, loss_ctc=8.455, loss_att=3.561, acc=0.979, loss=5.029, backward_time=0.256, grad_norm=71.101, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.255e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 17:19:06,565 (trainer:779) INFO: 88epoch:train:5370-5782batch: iter_time=2.813e-04, forward_time=0.157, loss_ctc=8.533, loss_att=3.598, acc=0.980, loss=5.078, backward_time=0.255, grad_norm=70.291, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.254e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 17:23:11,116 (trainer:779) INFO: 88epoch:train:5783-6195batch: iter_time=5.310e-04, forward_time=0.160, loss_ctc=8.491, loss_att=3.570, acc=0.978, loss=5.047, backward_time=0.256, grad_norm=73.692, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.252e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 17:27:15,977 (trainer:779) INFO: 88epoch:train:6196-6608batch: iter_time=0.003, forward_time=0.158, loss_ctc=8.478, loss_att=3.573, acc=0.976, loss=5.044, backward_time=0.256, grad_norm=70.163, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.251e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 17:31:17,996 (trainer:779) INFO: 88epoch:train:6609-7021batch: iter_time=3.714e-04, forward_time=0.156, loss_ctc=8.441, loss_att=3.546, acc=0.980, loss=5.014, backward_time=0.254, grad_norm=70.129, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.249e-04, train_time=1.172
+[bmi2:0/4] 2024-07-17 17:34:44,587 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 17:35:22,398 (trainer:779) INFO: 88epoch:train:7022-7434batch: iter_time=9.675e-04, forward_time=0.159, loss_ctc=8.566, loss_att=3.624, acc=0.979, loss=5.107, backward_time=0.257, grad_norm=79.084, clip=100.000, loss_scale=4.789e+33, optim_step_time=0.035, optim0_lr0=5.248e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 17:39:26,084 (trainer:779) INFO: 88epoch:train:7435-7847batch: iter_time=3.378e-04, forward_time=0.158, loss_ctc=8.526, loss_att=3.603, acc=0.977, loss=5.080, backward_time=0.256, grad_norm=68.649, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.246e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 17:43:29,728 (trainer:779) INFO: 88epoch:train:7848-8260batch: iter_time=2.078e-04, forward_time=0.158, loss_ctc=8.451, loss_att=3.550, acc=0.981, loss=5.020, backward_time=0.256, grad_norm=68.153, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.245e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 17:46:11,692 (trainer:365) INFO: 88epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.472, loss_att=3.573, acc=0.978, loss=5.043, backward_time=0.255, grad_norm=72.103, clip=100.000, loss_scale=5.466e+33, optim_step_time=0.036, optim0_lr0=5.259e-04, train_time=1.186, time=1 hour, 21 minutes and 45.51 seconds, total_count=727496, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.111, cer_ctc=0.040, loss_att=6.104, acc=0.947, cer=0.032, wer=0.484, loss=7.307, time=43.36 seconds, total_count=2992, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 53.75 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 17:46:17,063 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 17:46:17,149 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/87epoch.pth
+[bmi2:0/4] 2024-07-17 17:46:17,149 (trainer:299) INFO: 89/100epoch started. Estimated time to finish: 16 hours, 45 minutes and 42.57 seconds
+[bmi2:0/4] 2024-07-17 17:51:04,469 (trainer:779) INFO: 89epoch:train:1-413batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.371, loss_att=3.540, acc=0.979, loss=4.989, backward_time=0.255, grad_norm=71.293, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.243e-04, train_time=1.392
+[bmi2:0/4] 2024-07-17 17:55:07,525 (trainer:779) INFO: 89epoch:train:414-826batch: iter_time=3.862e-04, forward_time=0.158, loss_ctc=8.414, loss_att=3.548, acc=0.978, loss=5.008, backward_time=0.256, grad_norm=70.097, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.242e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 17:59:09,335 (trainer:779) INFO: 89epoch:train:827-1239batch: iter_time=3.052e-04, forward_time=0.156, loss_ctc=8.289, loss_att=3.491, acc=0.981, loss=4.930, backward_time=0.256, grad_norm=70.756, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.240e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 18:03:11,335 (trainer:779) INFO: 89epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.398, loss_att=3.537, acc=0.977, loss=4.995, backward_time=0.255, grad_norm=73.435, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.239e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 18:07:15,170 (trainer:779) INFO: 89epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.424, loss_att=3.540, acc=0.980, loss=5.005, backward_time=0.256, grad_norm=71.014, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.237e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 18:09:19,217 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 18:11:18,409 (trainer:779) INFO: 89epoch:train:2066-2478batch: iter_time=9.867e-04, forward_time=0.157, loss_ctc=8.403, loss_att=3.534, acc=0.976, loss=4.995, backward_time=0.255, grad_norm=75.133, clip=100.000, loss_scale=1.960e+33, optim_step_time=0.036, optim0_lr0=5.236e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 18:15:21,381 (trainer:779) INFO: 89epoch:train:2479-2891batch: iter_time=8.141e-04, forward_time=0.158, loss_ctc=8.371, loss_att=3.541, acc=0.981, loss=4.990, backward_time=0.256, grad_norm=68.681, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.234e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 18:19:25,820 (trainer:779) INFO: 89epoch:train:2892-3304batch: iter_time=1.656e-04, forward_time=0.158, loss_ctc=8.342, loss_att=3.513, acc=0.981, loss=4.962, backward_time=0.257, grad_norm=72.255, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.233e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 18:23:28,684 (trainer:779) INFO: 89epoch:train:3305-3717batch: iter_time=2.103e-04, forward_time=0.159, loss_ctc=8.436, loss_att=3.578, acc=0.979, loss=5.035, backward_time=0.255, grad_norm=71.788, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.231e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 18:27:31,042 (trainer:779) INFO: 89epoch:train:3718-4130batch: iter_time=9.025e-04, forward_time=0.156, loss_ctc=8.346, loss_att=3.549, acc=0.976, loss=4.988, backward_time=0.255, grad_norm=73.350, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.230e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 18:31:34,043 (trainer:779) INFO: 89epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.336, loss_att=3.498, acc=0.978, loss=4.949, backward_time=0.256, grad_norm=72.576, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.228e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 18:35:37,391 (trainer:779) INFO: 89epoch:train:4544-4956batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.500, loss_att=3.596, acc=0.978, loss=5.067, backward_time=0.255, grad_norm=74.129, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.227e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 18:39:38,812 (trainer:779) INFO: 89epoch:train:4957-5369batch: iter_time=4.991e-04, forward_time=0.156, loss_ctc=8.547, loss_att=3.605, acc=0.978, loss=5.087, backward_time=0.254, grad_norm=76.074, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.226e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 18:43:41,955 (trainer:779) INFO: 89epoch:train:5370-5782batch: iter_time=4.309e-04, forward_time=0.157, loss_ctc=8.385, loss_att=3.523, acc=0.978, loss=4.982, backward_time=0.255, grad_norm=68.558, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.224e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 18:47:44,072 (trainer:779) INFO: 89epoch:train:5783-6195batch: iter_time=4.688e-04, forward_time=0.157, loss_ctc=8.353, loss_att=3.522, acc=0.979, loss=4.971, backward_time=0.255, grad_norm=63.622, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.035, optim0_lr0=5.223e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 18:51:48,112 (trainer:779) INFO: 89epoch:train:6196-6608batch: iter_time=9.402e-04, forward_time=0.157, loss_ctc=8.388, loss_att=3.529, acc=0.976, loss=4.986, backward_time=0.256, grad_norm=67.654, clip=100.000, loss_scale=2.339e+33, optim_step_time=0.036, optim0_lr0=5.221e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 18:55:52,245 (trainer:779) INFO: 89epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.338, loss_att=3.550, acc=0.979, loss=4.986, backward_time=0.256, grad_norm=70.072, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.220e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 18:59:55,779 (trainer:779) INFO: 89epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.463, loss_att=3.561, acc=0.979, loss=5.032, backward_time=0.255, grad_norm=70.818, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.218e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 19:03:59,200 (trainer:779) INFO: 89epoch:train:7435-7847batch: iter_time=5.526e-04, forward_time=0.158, loss_ctc=8.470, loss_att=3.576, acc=0.976, loss=5.044, backward_time=0.255, grad_norm=67.352, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.217e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 19:08:04,076 (trainer:779) INFO: 89epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.158, loss_ctc=8.275, loss_att=3.504, acc=0.979, loss=4.935, backward_time=0.256, grad_norm=66.225, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.215e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 19:10:43,686 (trainer:365) INFO: 89epoch results: [train] iter_time=9.238e-04, forward_time=0.157, loss_ctc=8.393, loss_att=3.542, acc=0.978, loss=4.997, backward_time=0.255, grad_norm=70.733, clip=100.000, loss_scale=1.968e+33, optim_step_time=0.036, optim0_lr0=5.229e-04, train_time=1.188, time=1 hour, 21 minutes and 51.77 seconds, total_count=735763, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.086, cer_ctc=0.039, loss_att=6.363, acc=0.947, cer=0.033, wer=0.482, loss=7.480, time=42.01 seconds, total_count=3026, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 52.75 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 19:10:48,961 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 19:10:49,046 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/88epoch.pth
+[bmi2:0/4] 2024-07-17 19:10:49,046 (trainer:299) INFO: 90/100epoch started. Estimated time to finish: 15 hours, 22 minutes and 23.83 seconds
+[bmi2:0/4] 2024-07-17 19:15:35,558 (trainer:779) INFO: 90epoch:train:1-413batch: iter_time=0.004, forward_time=0.156, loss_ctc=8.235, loss_att=3.483, acc=0.975, loss=4.908, backward_time=0.254, grad_norm=69.843, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.214e-04, train_time=1.388
+[bmi2:0/4] 2024-07-17 19:19:40,233 (trainer:779) INFO: 90epoch:train:414-826batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.380, loss_att=3.548, acc=0.977, loss=4.998, backward_time=0.256, grad_norm=69.962, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.212e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 19:23:43,721 (trainer:779) INFO: 90epoch:train:827-1239batch: iter_time=8.977e-04, forward_time=0.158, loss_ctc=8.271, loss_att=3.507, acc=0.979, loss=4.936, backward_time=0.255, grad_norm=67.301, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.211e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 19:27:47,451 (trainer:779) INFO: 90epoch:train:1240-1652batch: iter_time=3.042e-04, forward_time=0.158, loss_ctc=8.239, loss_att=3.481, acc=0.979, loss=4.908, backward_time=0.256, grad_norm=64.551, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.209e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 19:31:33,710 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 19:31:50,591 (trainer:779) INFO: 90epoch:train:1653-2065batch: iter_time=3.146e-04, forward_time=0.158, loss_ctc=8.334, loss_att=3.536, acc=0.980, loss=4.975, backward_time=0.256, grad_norm=71.399, clip=100.000, loss_scale=2.748e+33, optim_step_time=0.035, optim0_lr0=5.208e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 19:35:52,973 (trainer:779) INFO: 90epoch:train:2066-2478batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.346, loss_att=3.533, acc=0.978, loss=4.977, backward_time=0.255, grad_norm=68.419, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.206e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 19:39:55,730 (trainer:779) INFO: 90epoch:train:2479-2891batch: iter_time=5.584e-04, forward_time=0.158, loss_ctc=8.285, loss_att=3.498, acc=0.978, loss=4.934, backward_time=0.256, grad_norm=71.992, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.205e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 19:43:57,182 (trainer:779) INFO: 90epoch:train:2892-3304batch: iter_time=1.934e-04, forward_time=0.155, loss_ctc=8.468, loss_att=3.579, acc=0.979, loss=5.046, backward_time=0.255, grad_norm=68.460, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.204e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 19:47:59,505 (trainer:779) INFO: 90epoch:train:3305-3717batch: iter_time=1.744e-04, forward_time=0.157, loss_ctc=8.311, loss_att=3.557, acc=0.980, loss=4.983, backward_time=0.255, grad_norm=75.836, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.202e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 19:52:02,480 (trainer:779) INFO: 90epoch:train:3718-4130batch: iter_time=5.915e-04, forward_time=0.155, loss_ctc=8.410, loss_att=3.560, acc=0.979, loss=5.015, backward_time=0.256, grad_norm=68.397, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.201e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 19:56:06,389 (trainer:779) INFO: 90epoch:train:4131-4543batch: iter_time=0.004, forward_time=0.157, loss_ctc=8.343, loss_att=3.512, acc=0.975, loss=4.961, backward_time=0.255, grad_norm=69.187, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.199e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 20:00:09,740 (trainer:779) INFO: 90epoch:train:4544-4956batch: iter_time=5.067e-04, forward_time=0.159, loss_ctc=8.335, loss_att=3.508, acc=0.980, loss=4.956, backward_time=0.255, grad_norm=67.610, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.198e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 20:04:11,220 (trainer:779) INFO: 90epoch:train:4957-5369batch: iter_time=9.599e-04, forward_time=0.157, loss_ctc=8.369, loss_att=3.536, acc=0.977, loss=4.986, backward_time=0.254, grad_norm=70.206, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.196e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 20:08:15,237 (trainer:779) INFO: 90epoch:train:5370-5782batch: iter_time=8.284e-04, forward_time=0.158, loss_ctc=8.335, loss_att=3.518, acc=0.981, loss=4.963, backward_time=0.256, grad_norm=68.549, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.195e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 20:12:18,269 (trainer:779) INFO: 90epoch:train:5783-6195batch: iter_time=4.962e-04, forward_time=0.158, loss_ctc=8.434, loss_att=3.552, acc=0.978, loss=5.017, backward_time=0.256, grad_norm=73.791, clip=100.000, loss_scale=3.592e+33, optim_step_time=0.036, optim0_lr0=5.193e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 20:16:21,100 (trainer:779) INFO: 90epoch:train:6196-6608batch: iter_time=5.661e-04, forward_time=0.157, loss_ctc=8.392, loss_att=3.539, acc=0.978, loss=4.995, backward_time=0.255, grad_norm=70.824, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.192e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 20:18:26,592 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 20:20:23,593 (trainer:779) INFO: 90epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.451, loss_att=3.559, acc=0.979, loss=5.027, backward_time=0.254, grad_norm=76.713, clip=100.000, loss_scale=3.939e+33, optim_step_time=0.036, optim0_lr0=5.190e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 20:24:26,794 (trainer:779) INFO: 90epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.482, loss_att=3.573, acc=0.977, loss=5.046, backward_time=0.255, grad_norm=71.531, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.189e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 20:28:29,103 (trainer:779) INFO: 90epoch:train:7435-7847batch: iter_time=2.296e-04, forward_time=0.157, loss_ctc=8.502, loss_att=3.586, acc=0.980, loss=5.061, backward_time=0.254, grad_norm=76.158, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.188e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 20:32:30,933 (trainer:779) INFO: 90epoch:train:7848-8260batch: iter_time=1.891e-04, forward_time=0.155, loss_ctc=8.389, loss_att=3.541, acc=0.980, loss=4.995, backward_time=0.255, grad_norm=71.533, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.186e-04, train_time=1.170
+[bmi2:0/4] 2024-07-17 20:35:04,289 (trainer:365) INFO: 90epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.365, loss_att=3.535, acc=0.978, loss=4.984, backward_time=0.255, grad_norm=70.607, clip=100.000, loss_scale=2.850e+33, optim_step_time=0.036, optim0_lr0=5.200e-04, train_time=1.187, time=1 hour, 21 minutes and 46.7 seconds, total_count=744030, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.096, cer_ctc=0.039, loss_att=6.034, acc=0.949, cer=0.032, wer=0.478, loss=7.253, time=40.83 seconds, total_count=3060, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 47.71 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 20:35:09,965 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 20:35:10,041 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/89epoch.pth
+[bmi2:0/4] 2024-07-17 20:35:10,041 (trainer:299) INFO: 91/100epoch started. Estimated time to finish: 13 hours, 58 minutes and 50.06 seconds
+[bmi2:0/4] 2024-07-17 20:39:55,400 (trainer:779) INFO: 91epoch:train:1-413batch: iter_time=0.003, forward_time=0.156, loss_ctc=8.212, loss_att=3.484, acc=0.977, loss=4.902, backward_time=0.256, grad_norm=67.696, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.185e-04, train_time=1.383
+[bmi2:0/4] 2024-07-17 20:43:59,250 (trainer:779) INFO: 91epoch:train:414-826batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.266, loss_att=3.493, acc=0.977, loss=4.925, backward_time=0.256, grad_norm=70.491, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.183e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 20:48:00,648 (trainer:779) INFO: 91epoch:train:827-1239batch: iter_time=1.713e-04, forward_time=0.156, loss_ctc=8.300, loss_att=3.546, acc=0.977, loss=4.972, backward_time=0.254, grad_norm=78.237, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.182e-04, train_time=1.169
+[bmi2:0/4] 2024-07-17 20:52:03,631 (trainer:779) INFO: 91epoch:train:1240-1652batch: iter_time=1.727e-04, forward_time=0.157, loss_ctc=8.205, loss_att=3.452, acc=0.982, loss=4.878, backward_time=0.255, grad_norm=66.510, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.180e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 20:56:06,597 (trainer:779) INFO: 91epoch:train:1653-2065batch: iter_time=1.808e-04, forward_time=0.158, loss_ctc=8.309, loss_att=3.543, acc=0.980, loss=4.973, backward_time=0.255, grad_norm=69.389, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.179e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 21:00:09,886 (trainer:779) INFO: 91epoch:train:2066-2478batch: iter_time=4.263e-04, forward_time=0.158, loss_ctc=8.274, loss_att=3.486, acc=0.981, loss=4.922, backward_time=0.255, grad_norm=68.960, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.178e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 21:04:11,544 (trainer:779) INFO: 91epoch:train:2479-2891batch: iter_time=1.747e-04, forward_time=0.156, loss_ctc=8.404, loss_att=3.548, acc=0.982, loss=5.005, backward_time=0.255, grad_norm=69.429, clip=100.000, loss_scale=4.701e+33, optim_step_time=0.036, optim0_lr0=5.176e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 21:08:14,404 (trainer:779) INFO: 91epoch:train:2892-3304batch: iter_time=7.365e-04, forward_time=0.157, loss_ctc=8.321, loss_att=3.511, acc=0.978, loss=4.954, backward_time=0.255, grad_norm=75.327, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.175e-04, train_time=1.175
+[bmi2:0/4] 2024-07-17 21:12:18,071 (trainer:779) INFO: 91epoch:train:3305-3717batch: iter_time=5.794e-04, forward_time=0.158, loss_ctc=8.390, loss_att=3.524, acc=0.979, loss=4.984, backward_time=0.256, grad_norm=74.095, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.173e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 21:16:21,437 (trainer:779) INFO: 91epoch:train:3718-4130batch: iter_time=5.940e-04, forward_time=0.157, loss_ctc=8.271, loss_att=3.518, acc=0.981, loss=4.944, backward_time=0.255, grad_norm=71.472, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.172e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 21:20:25,234 (trainer:779) INFO: 91epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.386, loss_att=3.522, acc=0.977, loss=4.981, backward_time=0.255, grad_norm=68.188, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.170e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 21:24:27,531 (trainer:779) INFO: 91epoch:train:4544-4956batch: iter_time=6.683e-04, forward_time=0.155, loss_ctc=8.320, loss_att=3.499, acc=0.979, loss=4.945, backward_time=0.255, grad_norm=69.385, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.169e-04, train_time=1.173
+[bmi2:0/4] 2024-07-17 21:28:31,101 (trainer:779) INFO: 91epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.321, loss_att=3.535, acc=0.975, loss=4.971, backward_time=0.255, grad_norm=70.561, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.168e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 21:32:34,752 (trainer:779) INFO: 91epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.315, loss_att=3.525, acc=0.977, loss=4.962, backward_time=0.255, grad_norm=69.025, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.166e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 21:36:38,540 (trainer:779) INFO: 91epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.265, loss_att=3.506, acc=0.976, loss=4.934, backward_time=0.255, grad_norm=67.591, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.165e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 21:40:41,563 (trainer:779) INFO: 91epoch:train:6196-6608batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.303, loss_att=3.522, acc=0.979, loss=4.956, backward_time=0.255, grad_norm=65.813, clip=100.000, loss_scale=5.844e+33, optim_step_time=0.036, optim0_lr0=5.163e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 21:42:05,781 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 21:44:44,397 (trainer:779) INFO: 91epoch:train:6609-7021batch: iter_time=3.652e-04, forward_time=0.157, loss_ctc=8.289, loss_att=3.525, acc=0.978, loss=4.954, backward_time=0.255, grad_norm=68.241, clip=100.000, loss_scale=6.965e+33, optim_step_time=0.036, optim0_lr0=5.162e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 21:48:19,924 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 21:48:46,918 (trainer:779) INFO: 91epoch:train:7022-7434batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.400, loss_att=3.546, acc=0.979, loss=5.002, backward_time=0.254, grad_norm=67.756, clip=100.000, loss_scale=4.902e+33, optim_step_time=0.035, optim0_lr0=5.160e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 21:52:51,374 (trainer:779) INFO: 91epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.277, loss_att=3.487, acc=0.979, loss=4.924, backward_time=0.256, grad_norm=63.916, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.159e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 21:56:54,684 (trainer:779) INFO: 91epoch:train:7848-8260batch: iter_time=6.806e-04, forward_time=0.157, loss_ctc=8.282, loss_att=3.505, acc=0.981, loss=4.938, backward_time=0.255, grad_norm=67.215, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.158e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 21:59:39,232 (trainer:365) INFO: 91epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.305, loss_att=3.514, acc=0.979, loss=4.951, backward_time=0.255, grad_norm=69.454, clip=100.000, loss_scale=4.234e+33, optim_step_time=0.036, optim0_lr0=5.171e-04, train_time=1.187, time=1 hour, 21 minutes and 49.71 seconds, total_count=752297, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.044, cer_ctc=0.039, loss_att=5.861, acc=0.949, cer=0.032, wer=0.481, loss=7.116, time=42.9 seconds, total_count=3094, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 56.57 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 21:59:44,207 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 21:59:44,254 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/90epoch.pth
+[bmi2:0/4] 2024-07-17 21:59:44,254 (trainer:299) INFO: 92/100epoch started. Estimated time to finish: 12 hours, 35 minutes and 17.66 seconds
+[bmi2:0/4] 2024-07-17 22:04:28,047 (trainer:779) INFO: 92epoch:train:1-413batch: iter_time=0.002, forward_time=0.154, loss_ctc=8.210, loss_att=3.468, acc=0.977, loss=4.891, backward_time=0.254, grad_norm=67.384, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.156e-04, train_time=1.375
+[bmi2:0/4] 2024-07-17 22:08:30,640 (trainer:779) INFO: 92epoch:train:414-826batch: iter_time=6.713e-04, forward_time=0.157, loss_ctc=8.301, loss_att=3.513, acc=0.976, loss=4.949, backward_time=0.254, grad_norm=72.116, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.155e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 22:12:33,949 (trainer:779) INFO: 92epoch:train:827-1239batch: iter_time=0.004, forward_time=0.156, loss_ctc=8.259, loss_att=3.491, acc=0.977, loss=4.921, backward_time=0.254, grad_norm=68.308, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.153e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 22:16:36,282 (trainer:779) INFO: 92epoch:train:1240-1652batch: iter_time=8.621e-04, forward_time=0.156, loss_ctc=8.207, loss_att=3.486, acc=0.981, loss=4.902, backward_time=0.255, grad_norm=73.528, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.152e-04, train_time=1.174
+[bmi2:0/4] 2024-07-17 22:20:39,189 (trainer:779) INFO: 92epoch:train:1653-2065batch: iter_time=1.864e-04, forward_time=0.157, loss_ctc=8.307, loss_att=3.494, acc=0.981, loss=4.938, backward_time=0.256, grad_norm=73.011, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.151e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 22:24:43,508 (trainer:779) INFO: 92epoch:train:2066-2478batch: iter_time=8.028e-04, forward_time=0.158, loss_ctc=8.267, loss_att=3.468, acc=0.981, loss=4.908, backward_time=0.256, grad_norm=69.205, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.149e-04, train_time=1.182
+[bmi2:0/4] 2024-07-17 22:28:46,593 (trainer:779) INFO: 92epoch:train:2479-2891batch: iter_time=9.212e-04, forward_time=0.157, loss_ctc=8.336, loss_att=3.515, acc=0.979, loss=4.961, backward_time=0.255, grad_norm=69.496, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.148e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 22:31:09,424 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 22:32:49,895 (trainer:779) INFO: 92epoch:train:2892-3304batch: iter_time=6.037e-04, forward_time=0.158, loss_ctc=8.338, loss_att=3.528, acc=0.978, loss=4.971, backward_time=0.255, grad_norm=71.721, clip=100.000, loss_scale=2.659e+33, optim_step_time=0.036, optim0_lr0=5.146e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 22:36:52,533 (trainer:779) INFO: 92epoch:train:3305-3717batch: iter_time=1.783e-04, forward_time=0.157, loss_ctc=8.206, loss_att=3.460, acc=0.982, loss=4.884, backward_time=0.255, grad_norm=68.584, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.145e-04, train_time=1.176
+[bmi2:0/4] 2024-07-17 22:40:57,041 (trainer:779) INFO: 92epoch:train:3718-4130batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.284, loss_att=3.498, acc=0.978, loss=4.934, backward_time=0.257, grad_norm=67.996, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.143e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 22:45:00,916 (trainer:779) INFO: 92epoch:train:4131-4543batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.397, loss_att=3.526, acc=0.975, loss=4.987, backward_time=0.255, grad_norm=68.276, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.142e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 22:49:04,314 (trainer:779) INFO: 92epoch:train:4544-4956batch: iter_time=1.749e-04, forward_time=0.158, loss_ctc=8.385, loss_att=3.538, acc=0.978, loss=4.992, backward_time=0.255, grad_norm=67.929, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.141e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 22:53:08,918 (trainer:779) INFO: 92epoch:train:4957-5369batch: iter_time=4.315e-04, forward_time=0.159, loss_ctc=8.316, loss_att=3.502, acc=0.980, loss=4.946, backward_time=0.255, grad_norm=72.025, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.037, optim0_lr0=5.139e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 22:57:12,382 (trainer:779) INFO: 92epoch:train:5370-5782batch: iter_time=1.788e-04, forward_time=0.158, loss_ctc=8.317, loss_att=3.503, acc=0.980, loss=4.947, backward_time=0.256, grad_norm=73.561, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.138e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 23:01:16,136 (trainer:779) INFO: 92epoch:train:5783-6195batch: iter_time=6.741e-04, forward_time=0.158, loss_ctc=8.377, loss_att=3.509, acc=0.980, loss=4.970, backward_time=0.256, grad_norm=68.544, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.136e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 23:05:19,204 (trainer:779) INFO: 92epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.313, loss_att=3.500, acc=0.977, loss=4.944, backward_time=0.255, grad_norm=71.956, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.135e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 23:09:22,277 (trainer:779) INFO: 92epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.328, loss_att=3.510, acc=0.978, loss=4.955, backward_time=0.256, grad_norm=71.530, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.134e-04, train_time=1.178
+[bmi2:0/4] 2024-07-17 23:13:24,266 (trainer:779) INFO: 92epoch:train:7022-7434batch: iter_time=1.789e-04, forward_time=0.156, loss_ctc=8.368, loss_att=3.519, acc=0.978, loss=4.974, backward_time=0.254, grad_norm=70.862, clip=100.000, loss_scale=4.477e+33, optim_step_time=0.036, optim0_lr0=5.132e-04, train_time=1.171
+[bmi2:0/4] 2024-07-17 23:17:28,699 (trainer:779) INFO: 92epoch:train:7435-7847batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.390, loss_att=3.494, acc=0.978, loss=4.962, backward_time=0.257, grad_norm=71.546, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.131e-04, train_time=1.184
+[bmi2:0/4] 2024-07-17 23:21:31,963 (trainer:779) INFO: 92epoch:train:7848-8260batch: iter_time=0.003, forward_time=0.157, loss_ctc=8.442, loss_att=3.565, acc=0.978, loss=5.028, backward_time=0.254, grad_norm=72.757, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.129e-04, train_time=1.177
+[bmi2:0/4] 2024-07-17 23:24:08,115 (trainer:365) INFO: 92epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.318, loss_att=3.505, acc=0.979, loss=4.949, backward_time=0.255, grad_norm=70.509, clip=100.000, loss_scale=2.955e+33, optim_step_time=0.036, optim0_lr0=5.143e-04, train_time=1.188, time=1 hour, 21 minutes and 52.85 seconds, total_count=760564, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.965, cer_ctc=0.039, loss_att=5.879, acc=0.951, cer=0.032, wer=0.479, loss=7.105, time=43.21 seconds, total_count=3128, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 47.79 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-17 23:24:12,539 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-17 23:24:12,606 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/78epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/91epoch.pth
+[bmi2:0/4] 2024-07-17 23:24:12,606 (trainer:299) INFO: 93/100epoch started. Estimated time to finish: 11 hours, 11 minutes and 36.28 seconds
+[bmi2:0/4] 2024-07-17 23:28:54,951 (trainer:779) INFO: 93epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.315, loss_att=3.501, acc=0.982, loss=4.945, backward_time=0.254, grad_norm=74.702, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.128e-04, train_time=1.368
+[bmi2:0/4] 2024-07-17 23:32:58,615 (trainer:779) INFO: 93epoch:train:414-826batch: iter_time=0.003, forward_time=0.156, loss_ctc=8.192, loss_att=3.459, acc=0.979, loss=4.879, backward_time=0.255, grad_norm=73.807, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.127e-04, train_time=1.179
+[bmi2:0/4] 2024-07-17 23:37:02,779 (trainer:779) INFO: 93epoch:train:827-1239batch: iter_time=8.666e-04, forward_time=0.159, loss_ctc=8.212, loss_att=3.480, acc=0.976, loss=4.899, backward_time=0.256, grad_norm=69.683, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.125e-04, train_time=1.183
+[bmi2:0/4] 2024-07-17 23:41:06,698 (trainer:779) INFO: 93epoch:train:1240-1652batch: iter_time=7.127e-04, forward_time=0.158, loss_ctc=8.281, loss_att=3.512, acc=0.976, loss=4.943, backward_time=0.256, grad_norm=72.248, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.124e-04, train_time=1.180
+[bmi2:0/4] 2024-07-17 23:45:11,181 (trainer:779) INFO: 93epoch:train:1653-2065batch: iter_time=5.289e-04, forward_time=0.159, loss_ctc=8.291, loss_att=3.494, acc=0.977, loss=4.933, backward_time=0.257, grad_norm=66.065, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.123e-04, train_time=1.185
+[bmi2:0/4] 2024-07-17 23:49:15,214 (trainer:779) INFO: 93epoch:train:2066-2478batch: iter_time=3.130e-04, forward_time=0.157, loss_ctc=8.255, loss_att=3.483, acc=0.982, loss=4.914, backward_time=0.256, grad_norm=69.556, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.121e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 23:53:19,013 (trainer:779) INFO: 93epoch:train:2479-2891batch: iter_time=1.742e-04, forward_time=0.158, loss_ctc=8.380, loss_att=3.553, acc=0.981, loss=5.001, backward_time=0.256, grad_norm=75.100, clip=100.000, loss_scale=5.470e+33, optim_step_time=0.036, optim0_lr0=5.120e-04, train_time=1.181
+[bmi2:0/4] 2024-07-17 23:55:25,046 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-17 23:57:22,096 (trainer:779) INFO: 93epoch:train:2892-3304batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.310, loss_att=3.482, acc=0.978, loss=4.930, backward_time=0.254, grad_norm=68.866, clip=100.000, loss_scale=7.864e+33, optim_step_time=0.036, optim0_lr0=5.118e-04, train_time=1.176
+[bmi2:0/4] 2024-07-18 00:01:22,577 (trainer:779) INFO: 93epoch:train:3305-3717batch: iter_time=1.906e-04, forward_time=0.154, loss_ctc=8.356, loss_att=3.526, acc=0.981, loss=4.975, backward_time=0.255, grad_norm=68.786, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.117e-04, train_time=1.165
+[bmi2:0/4] 2024-07-18 00:04:42,240 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 00:05:26,185 (trainer:779) INFO: 93epoch:train:3718-4130batch: iter_time=5.493e-04, forward_time=0.157, loss_ctc=8.313, loss_att=3.487, acc=0.979, loss=4.935, backward_time=0.256, grad_norm=65.510, clip=100.000, loss_scale=4.726e+33, optim_step_time=0.036, optim0_lr0=5.116e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 00:09:29,988 (trainer:779) INFO: 93epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.140, loss_att=3.435, acc=0.978, loss=4.847, backward_time=0.256, grad_norm=64.973, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.114e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 00:13:33,707 (trainer:779) INFO: 93epoch:train:4544-4956batch: iter_time=1.845e-04, forward_time=0.157, loss_ctc=8.263, loss_att=3.505, acc=0.981, loss=4.933, backward_time=0.256, grad_norm=71.798, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.113e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 00:17:37,417 (trainer:779) INFO: 93epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.200, loss_att=3.459, acc=0.977, loss=4.882, backward_time=0.256, grad_norm=68.514, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.111e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 00:21:40,972 (trainer:779) INFO: 93epoch:train:5370-5782batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.274, loss_att=3.497, acc=0.979, loss=4.930, backward_time=0.255, grad_norm=67.354, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.110e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 00:25:46,537 (trainer:779) INFO: 93epoch:train:5783-6195batch: iter_time=9.941e-04, forward_time=0.158, loss_ctc=8.308, loss_att=3.497, acc=0.977, loss=4.940, backward_time=0.256, grad_norm=64.040, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.037, optim0_lr0=5.109e-04, train_time=1.190
+[bmi2:0/4] 2024-07-18 00:29:50,632 (trainer:779) INFO: 93epoch:train:6196-6608batch: iter_time=7.070e-04, forward_time=0.157, loss_ctc=8.199, loss_att=3.448, acc=0.979, loss=4.874, backward_time=0.256, grad_norm=69.047, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.107e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 00:33:54,873 (trainer:779) INFO: 93epoch:train:6609-7021batch: iter_time=6.080e-04, forward_time=0.159, loss_ctc=8.316, loss_att=3.493, acc=0.980, loss=4.940, backward_time=0.256, grad_norm=75.151, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.106e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 00:37:57,841 (trainer:779) INFO: 93epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.247, loss_att=3.462, acc=0.977, loss=4.897, backward_time=0.254, grad_norm=65.772, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.105e-04, train_time=1.176
+[bmi2:0/4] 2024-07-18 00:42:02,183 (trainer:779) INFO: 93epoch:train:7435-7847batch: iter_time=4.112e-04, forward_time=0.159, loss_ctc=8.194, loss_att=3.444, acc=0.979, loss=4.869, backward_time=0.256, grad_norm=67.127, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.103e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 00:46:06,422 (trainer:779) INFO: 93epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.155, loss_att=3.464, acc=0.977, loss=4.871, backward_time=0.256, grad_norm=65.263, clip=100.000, loss_scale=3.875e+33, optim_step_time=0.037, optim0_lr0=5.102e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 00:48:34,811 (trainer:365) INFO: 93epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.260, loss_att=3.484, acc=0.979, loss=4.917, backward_time=0.256, grad_norm=69.162, clip=100.000, loss_scale=4.082e+33, optim_step_time=0.036, optim0_lr0=5.115e-04, train_time=1.190, time=1 hour, 21 minutes and 58.56 seconds, total_count=768831, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.896, cer_ctc=0.039, loss_att=5.957, acc=0.950, cer=0.032, wer=0.482, loss=7.138, time=39.85 seconds, total_count=3162, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 43.79 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 00:48:39,233 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 00:48:39,234 (trainer:299) INFO: 94/100epoch started. Estimated time to finish: 9 hours, 47 minutes and 49.6 seconds
+[bmi2:0/4] 2024-07-18 00:52:51,123 (trainer:779) INFO: 94epoch:train:1-413batch: iter_time=0.001, forward_time=0.154, loss_ctc=8.123, loss_att=3.434, acc=0.977, loss=4.841, backward_time=0.256, grad_norm=68.684, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.035, optim0_lr0=5.100e-04, train_time=1.220
+[bmi2:0/4] 2024-07-18 00:56:53,982 (trainer:779) INFO: 94epoch:train:414-826batch: iter_time=5.281e-04, forward_time=0.157, loss_ctc=8.236, loss_att=3.467, acc=0.979, loss=4.898, backward_time=0.255, grad_norm=65.700, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.099e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 01:00:58,013 (trainer:779) INFO: 94epoch:train:827-1239batch: iter_time=8.509e-04, forward_time=0.158, loss_ctc=8.133, loss_att=3.461, acc=0.981, loss=4.863, backward_time=0.256, grad_norm=70.206, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.098e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 01:05:02,154 (trainer:779) INFO: 94epoch:train:1240-1652batch: iter_time=7.591e-04, forward_time=0.158, loss_ctc=8.081, loss_att=3.434, acc=0.979, loss=4.828, backward_time=0.256, grad_norm=66.329, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.096e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 01:09:05,573 (trainer:779) INFO: 94epoch:train:1653-2065batch: iter_time=3.203e-04, forward_time=0.158, loss_ctc=8.191, loss_att=3.489, acc=0.980, loss=4.900, backward_time=0.256, grad_norm=69.936, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.095e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 01:13:09,259 (trainer:779) INFO: 94epoch:train:2066-2478batch: iter_time=3.704e-04, forward_time=0.158, loss_ctc=8.185, loss_att=3.464, acc=0.980, loss=4.880, backward_time=0.256, grad_norm=69.834, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.094e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 01:17:14,049 (trainer:779) INFO: 94epoch:train:2479-2891batch: iter_time=6.450e-04, forward_time=0.159, loss_ctc=8.129, loss_att=3.431, acc=0.980, loss=4.841, backward_time=0.257, grad_norm=67.190, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.092e-04, train_time=1.186
+[bmi2:0/4] 2024-07-18 01:17:49,951 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 01:21:17,624 (trainer:779) INFO: 94epoch:train:2892-3304batch: iter_time=6.048e-04, forward_time=0.156, loss_ctc=8.157, loss_att=3.448, acc=0.980, loss=4.861, backward_time=0.255, grad_norm=66.779, clip=100.000, loss_scale=2.974e+33, optim_step_time=0.036, optim0_lr0=5.091e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 01:25:21,578 (trainer:779) INFO: 94epoch:train:3305-3717batch: iter_time=5.057e-04, forward_time=0.159, loss_ctc=8.248, loss_att=3.477, acc=0.978, loss=4.908, backward_time=0.256, grad_norm=66.461, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.090e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 01:29:25,608 (trainer:779) INFO: 94epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.242, loss_att=3.466, acc=0.978, loss=4.899, backward_time=0.255, grad_norm=67.460, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.088e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 01:33:28,619 (trainer:779) INFO: 94epoch:train:4131-4543batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.221, loss_att=3.472, acc=0.974, loss=4.897, backward_time=0.255, grad_norm=65.996, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.087e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 01:37:32,906 (trainer:779) INFO: 94epoch:train:4544-4956batch: iter_time=5.062e-04, forward_time=0.159, loss_ctc=8.149, loss_att=3.453, acc=0.980, loss=4.861, backward_time=0.256, grad_norm=74.375, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.085e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 01:41:36,921 (trainer:779) INFO: 94epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.182, loss_att=3.485, acc=0.980, loss=4.894, backward_time=0.257, grad_norm=71.611, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.084e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 01:45:40,580 (trainer:779) INFO: 94epoch:train:5370-5782batch: iter_time=5.534e-04, forward_time=0.157, loss_ctc=8.230, loss_att=3.460, acc=0.982, loss=4.891, backward_time=0.256, grad_norm=67.727, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.083e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 01:49:45,415 (trainer:779) INFO: 94epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.168, loss_att=3.450, acc=0.978, loss=4.865, backward_time=0.256, grad_norm=67.011, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.081e-04, train_time=1.186
+[bmi2:0/4] 2024-07-18 01:53:48,165 (trainer:779) INFO: 94epoch:train:6196-6608batch: iter_time=2.352e-04, forward_time=0.157, loss_ctc=8.233, loss_att=3.471, acc=0.980, loss=4.899, backward_time=0.255, grad_norm=70.271, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.080e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 01:57:51,263 (trainer:779) INFO: 94epoch:train:6609-7021batch: iter_time=5.014e-04, forward_time=0.157, loss_ctc=8.299, loss_att=3.482, acc=0.977, loss=4.927, backward_time=0.256, grad_norm=75.224, clip=100.000, loss_scale=3.025e+33, optim_step_time=0.036, optim0_lr0=5.079e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 02:01:53,564 (trainer:779) INFO: 94epoch:train:7022-7434batch: iter_time=3.700e-04, forward_time=0.157, loss_ctc=8.277, loss_att=3.494, acc=0.979, loss=4.929, backward_time=0.255, grad_norm=65.586, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.077e-04, train_time=1.172
+[bmi2:0/4] 2024-07-18 02:05:58,769 (trainer:779) INFO: 94epoch:train:7435-7847batch: iter_time=0.003, forward_time=0.159, loss_ctc=8.266, loss_att=3.483, acc=0.976, loss=4.918, backward_time=0.256, grad_norm=69.004, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.076e-04, train_time=1.188
+[bmi2:0/4] 2024-07-18 02:07:38,435 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 02:10:01,570 (trainer:779) INFO: 94epoch:train:7848-8260batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.214, loss_att=3.433, acc=0.979, loss=4.868, backward_time=0.255, grad_norm=66.672, clip=100.000, loss_scale=3.655e+33, optim_step_time=0.036, optim0_lr0=5.075e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 02:12:42,199 (trainer:365) INFO: 94epoch results: [train] iter_time=9.162e-04, forward_time=0.157, loss_ctc=8.198, loss_att=3.463, acc=0.979, loss=4.883, backward_time=0.256, grad_norm=68.600, clip=100.000, loss_scale=3.857e+33, optim_step_time=0.036, optim0_lr0=5.087e-04, train_time=1.182, time=1 hour, 21 minutes and 27.32 seconds, total_count=777098, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.137, cer_ctc=0.039, loss_att=6.288, acc=0.948, cer=0.033, wer=0.478, loss=7.443, time=43.94 seconds, total_count=3196, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 51.7 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 02:12:47,752 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 02:12:47,831 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/93epoch.pth
+[bmi2:0/4] 2024-07-18 02:12:47,832 (trainer:299) INFO: 95/100epoch started. Estimated time to finish: 8 hours, 23 minutes and 53.97 seconds
+[bmi2:0/4] 2024-07-18 02:17:35,029 (trainer:779) INFO: 95epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.204, loss_att=3.455, acc=0.982, loss=4.880, backward_time=0.257, grad_norm=69.207, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.073e-04, train_time=1.391
+[bmi2:0/4] 2024-07-18 02:21:39,815 (trainer:779) INFO: 95epoch:train:414-826batch: iter_time=8.119e-04, forward_time=0.158, loss_ctc=8.204, loss_att=3.464, acc=0.978, loss=4.886, backward_time=0.256, grad_norm=71.477, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.072e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 02:25:43,226 (trainer:779) INFO: 95epoch:train:827-1239batch: iter_time=8.061e-04, forward_time=0.157, loss_ctc=8.198, loss_att=3.479, acc=0.977, loss=4.894, backward_time=0.255, grad_norm=68.529, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.071e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 02:29:47,601 (trainer:779) INFO: 95epoch:train:1240-1652batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.069, loss_att=3.409, acc=0.980, loss=4.807, backward_time=0.256, grad_norm=66.500, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.069e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 02:33:48,574 (trainer:779) INFO: 95epoch:train:1653-2065batch: iter_time=9.863e-04, forward_time=0.154, loss_ctc=8.273, loss_att=3.470, acc=0.980, loss=4.911, backward_time=0.254, grad_norm=66.009, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.068e-04, train_time=1.167
+[bmi2:0/4] 2024-07-18 02:37:50,622 (trainer:779) INFO: 95epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.154, loss_ctc=8.114, loss_att=3.451, acc=0.978, loss=4.850, backward_time=0.254, grad_norm=68.263, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.035, optim0_lr0=5.067e-04, train_time=1.171
+[bmi2:0/4] 2024-07-18 02:41:50,681 (trainer:779) INFO: 95epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.112, loss_att=3.412, acc=0.978, loss=4.822, backward_time=0.253, grad_norm=66.095, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.065e-04, train_time=1.163
+[bmi2:0/4] 2024-07-18 02:45:53,490 (trainer:779) INFO: 95epoch:train:2892-3304batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.254, loss_att=3.479, acc=0.978, loss=4.912, backward_time=0.255, grad_norm=69.642, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.064e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 02:49:57,304 (trainer:779) INFO: 95epoch:train:3305-3717batch: iter_time=1.705e-04, forward_time=0.158, loss_ctc=8.249, loss_att=3.465, acc=0.981, loss=4.900, backward_time=0.256, grad_norm=69.684, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.063e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 02:54:01,359 (trainer:779) INFO: 95epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.208, loss_att=3.471, acc=0.977, loss=4.892, backward_time=0.256, grad_norm=70.401, clip=100.000, loss_scale=4.979e+33, optim_step_time=0.036, optim0_lr0=5.061e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 02:58:04,635 (trainer:779) INFO: 95epoch:train:4131-4543batch: iter_time=6.547e-04, forward_time=0.157, loss_ctc=8.237, loss_att=3.452, acc=0.978, loss=4.887, backward_time=0.255, grad_norm=70.803, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.037, optim0_lr0=5.060e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 02:59:22,455 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 03:02:07,820 (trainer:779) INFO: 95epoch:train:4544-4956batch: iter_time=4.772e-04, forward_time=0.157, loss_ctc=8.035, loss_att=3.409, acc=0.978, loss=4.797, backward_time=0.255, grad_norm=67.850, clip=100.000, loss_scale=3.428e+33, optim_step_time=0.036, optim0_lr0=5.058e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 03:06:11,533 (trainer:779) INFO: 95epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.186, loss_att=3.447, acc=0.979, loss=4.869, backward_time=0.256, grad_norm=67.822, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.057e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 03:10:16,953 (trainer:779) INFO: 95epoch:train:5370-5782batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.050, loss_att=3.418, acc=0.979, loss=4.808, backward_time=0.257, grad_norm=65.162, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.056e-04, train_time=1.188
+[bmi2:0/4] 2024-07-18 03:14:20,669 (trainer:779) INFO: 95epoch:train:5783-6195batch: iter_time=1.763e-04, forward_time=0.158, loss_ctc=8.254, loss_att=3.480, acc=0.980, loss=4.913, backward_time=0.256, grad_norm=74.541, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.054e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 03:18:25,202 (trainer:779) INFO: 95epoch:train:6196-6608batch: iter_time=7.436e-04, forward_time=0.159, loss_ctc=8.072, loss_att=3.406, acc=0.978, loss=4.806, backward_time=0.256, grad_norm=69.468, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.053e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 03:22:28,200 (trainer:779) INFO: 95epoch:train:6609-7021batch: iter_time=2.347e-04, forward_time=0.156, loss_ctc=8.289, loss_att=3.496, acc=0.981, loss=4.934, backward_time=0.255, grad_norm=73.839, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.052e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 03:24:32,024 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 03:26:31,349 (trainer:779) INFO: 95epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.295, loss_att=3.517, acc=0.977, loss=4.951, backward_time=0.255, grad_norm=75.407, clip=100.000, loss_scale=1.960e+33, optim_step_time=0.036, optim0_lr0=5.051e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 03:30:33,348 (trainer:779) INFO: 95epoch:train:7435-7847batch: iter_time=2.998e-04, forward_time=0.155, loss_ctc=8.335, loss_att=3.481, acc=0.980, loss=4.937, backward_time=0.255, grad_norm=74.403, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.049e-04, train_time=1.172
+[bmi2:0/4] 2024-07-18 03:34:36,347 (trainer:779) INFO: 95epoch:train:7848-8260batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.221, loss_att=3.450, acc=0.977, loss=4.882, backward_time=0.255, grad_norm=71.894, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.048e-04, train_time=1.176
+[bmi2:0/4] 2024-07-18 03:37:19,489 (trainer:365) INFO: 95epoch results: [train] iter_time=0.001, forward_time=0.157, loss_ctc=8.193, loss_att=3.455, acc=0.979, loss=4.877, backward_time=0.255, grad_norm=69.857, clip=100.000, loss_scale=2.724e+33, optim_step_time=0.036, optim0_lr0=5.061e-04, train_time=1.188, time=1 hour, 21 minutes and 53.52 seconds, total_count=785365, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=9.932, cer_ctc=0.039, loss_att=5.961, acc=0.951, cer=0.032, wer=0.478, loss=7.152, time=44.2 seconds, total_count=3230, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 53.93 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 03:37:24,169 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 03:37:24,224 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/61epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/94epoch.pth
+[bmi2:0/4] 2024-07-18 03:37:24,224 (trainer:299) INFO: 96/100epoch started. Estimated time to finish: 7 hours and 3.47 seconds
+[bmi2:0/4] 2024-07-18 03:42:11,231 (trainer:779) INFO: 96epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.220, loss_att=3.436, acc=0.980, loss=4.871, backward_time=0.256, grad_norm=73.706, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.047e-04, train_time=1.390
+[bmi2:0/4] 2024-07-18 03:46:15,260 (trainer:779) INFO: 96epoch:train:414-826batch: iter_time=6.559e-04, forward_time=0.158, loss_ctc=8.300, loss_att=3.487, acc=0.980, loss=4.931, backward_time=0.256, grad_norm=71.802, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.045e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 03:50:20,074 (trainer:779) INFO: 96epoch:train:827-1239batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.178, loss_att=3.456, acc=0.977, loss=4.873, backward_time=0.257, grad_norm=72.833, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.044e-04, train_time=1.186
+[bmi2:0/4] 2024-07-18 03:54:23,687 (trainer:779) INFO: 96epoch:train:1240-1652batch: iter_time=8.587e-04, forward_time=0.158, loss_ctc=8.289, loss_att=3.478, acc=0.979, loss=4.921, backward_time=0.256, grad_norm=72.336, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.043e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 03:58:28,180 (trainer:779) INFO: 96epoch:train:1653-2065batch: iter_time=5.981e-04, forward_time=0.159, loss_ctc=8.199, loss_att=3.459, acc=0.981, loss=4.881, backward_time=0.257, grad_norm=69.244, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.041e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 04:02:32,517 (trainer:779) INFO: 96epoch:train:2066-2478batch: iter_time=6.446e-04, forward_time=0.158, loss_ctc=8.218, loss_att=3.455, acc=0.980, loss=4.884, backward_time=0.257, grad_norm=70.498, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.036, optim0_lr0=5.040e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 04:06:35,664 (trainer:779) INFO: 96epoch:train:2479-2891batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.176, loss_att=3.476, acc=0.973, loss=4.886, backward_time=0.255, grad_norm=68.423, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.037, optim0_lr0=5.039e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 04:10:39,371 (trainer:779) INFO: 96epoch:train:2892-3304batch: iter_time=2.832e-04, forward_time=0.158, loss_ctc=8.321, loss_att=3.483, acc=0.982, loss=4.934, backward_time=0.257, grad_norm=71.136, clip=100.000, loss_scale=2.358e+33, optim_step_time=0.036, optim0_lr0=5.037e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 04:14:42,667 (trainer:779) INFO: 96epoch:train:3305-3717batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.310, loss_att=3.495, acc=0.977, loss=4.939, backward_time=0.256, grad_norm=71.771, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.036e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 04:18:46,369 (trainer:779) INFO: 96epoch:train:3718-4130batch: iter_time=7.542e-04, forward_time=0.159, loss_ctc=8.194, loss_att=3.444, acc=0.981, loss=4.869, backward_time=0.256, grad_norm=67.340, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.035e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 04:22:47,777 (trainer:779) INFO: 96epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.106, loss_att=3.429, acc=0.979, loss=4.832, backward_time=0.255, grad_norm=68.297, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.033e-04, train_time=1.169
+[bmi2:0/4] 2024-07-18 04:26:50,156 (trainer:779) INFO: 96epoch:train:4544-4956batch: iter_time=4.242e-04, forward_time=0.157, loss_ctc=8.130, loss_att=3.402, acc=0.979, loss=4.821, backward_time=0.255, grad_norm=68.592, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.032e-04, train_time=1.173
+[bmi2:0/4] 2024-07-18 04:30:52,574 (trainer:779) INFO: 96epoch:train:4957-5369batch: iter_time=6.836e-04, forward_time=0.157, loss_ctc=8.255, loss_att=3.492, acc=0.977, loss=4.921, backward_time=0.255, grad_norm=72.525, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.031e-04, train_time=1.174
+[bmi2:0/4] 2024-07-18 04:34:56,610 (trainer:779) INFO: 96epoch:train:5370-5782batch: iter_time=2.524e-04, forward_time=0.159, loss_ctc=8.267, loss_att=3.470, acc=0.980, loss=4.909, backward_time=0.256, grad_norm=66.459, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.029e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 04:39:00,621 (trainer:779) INFO: 96epoch:train:5783-6195batch: iter_time=1.717e-04, forward_time=0.159, loss_ctc=8.335, loss_att=3.500, acc=0.981, loss=4.951, backward_time=0.257, grad_norm=74.025, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.028e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 04:43:03,746 (trainer:779) INFO: 96epoch:train:6196-6608batch: iter_time=8.351e-04, forward_time=0.157, loss_ctc=8.227, loss_att=3.441, acc=0.978, loss=4.877, backward_time=0.256, grad_norm=67.554, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.027e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 04:47:07,694 (trainer:779) INFO: 96epoch:train:6609-7021batch: iter_time=9.207e-04, forward_time=0.159, loss_ctc=8.133, loss_att=3.433, acc=0.979, loss=4.843, backward_time=0.256, grad_norm=70.531, clip=100.000, loss_scale=2.936e+33, optim_step_time=0.036, optim0_lr0=5.025e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 04:51:10,220 (trainer:779) INFO: 96epoch:train:7022-7434batch: iter_time=4.771e-04, forward_time=0.157, loss_ctc=8.208, loss_att=3.443, acc=0.978, loss=4.873, backward_time=0.255, grad_norm=67.450, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.024e-04, train_time=1.174
+[bmi2:0/4] 2024-07-18 04:55:14,545 (trainer:779) INFO: 96epoch:train:7435-7847batch: iter_time=5.789e-04, forward_time=0.159, loss_ctc=8.258, loss_att=3.462, acc=0.978, loss=4.901, backward_time=0.257, grad_norm=68.872, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.023e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 04:59:18,026 (trainer:779) INFO: 96epoch:train:7848-8260batch: iter_time=1.778e-04, forward_time=0.158, loss_ctc=8.304, loss_att=3.468, acc=0.981, loss=4.919, backward_time=0.256, grad_norm=72.096, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.021e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 05:02:00,793 (trainer:365) INFO: 96epoch results: [train] iter_time=7.746e-04, forward_time=0.158, loss_ctc=8.231, loss_att=3.460, acc=0.979, loss=4.891, backward_time=0.256, grad_norm=70.264, clip=100.000, loss_scale=2.539e+33, optim_step_time=0.036, optim0_lr0=5.034e-04, train_time=1.190, time=1 hour, 21 minutes and 59.11 seconds, total_count=793632, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.211, cer_ctc=0.039, loss_att=6.104, acc=0.949, cer=0.033, wer=0.483, loss=7.336, time=43.19 seconds, total_count=3264, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 54.27 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 05:02:05,818 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 05:02:05,819 (trainer:299) INFO: 97/100epoch started. Estimated time to finish: 5 hours, 36 minutes and 9.89 seconds
+[bmi2:0/4] 2024-07-18 05:06:49,174 (trainer:779) INFO: 97epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.194, loss_att=3.452, acc=0.976, loss=4.874, backward_time=0.255, grad_norm=71.633, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.020e-04, train_time=1.373
+[bmi2:0/4] 2024-07-18 05:10:52,621 (trainer:779) INFO: 97epoch:train:414-826batch: iter_time=5.725e-04, forward_time=0.158, loss_ctc=8.154, loss_att=3.433, acc=0.981, loss=4.849, backward_time=0.256, grad_norm=65.687, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.019e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 05:14:56,876 (trainer:779) INFO: 97epoch:train:827-1239batch: iter_time=0.004, forward_time=0.158, loss_ctc=8.180, loss_att=3.461, acc=0.976, loss=4.877, backward_time=0.257, grad_norm=72.806, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.018e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 05:19:00,259 (trainer:779) INFO: 97epoch:train:1240-1652batch: iter_time=5.582e-04, forward_time=0.158, loss_ctc=8.144, loss_att=3.430, acc=0.980, loss=4.844, backward_time=0.256, grad_norm=67.783, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.016e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 05:23:03,714 (trainer:779) INFO: 97epoch:train:1653-2065batch: iter_time=2.957e-04, forward_time=0.158, loss_ctc=8.136, loss_att=3.411, acc=0.978, loss=4.829, backward_time=0.255, grad_norm=70.513, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.015e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 05:27:07,356 (trainer:779) INFO: 97epoch:train:2066-2478batch: iter_time=5.569e-04, forward_time=0.158, loss_ctc=8.130, loss_att=3.453, acc=0.980, loss=4.856, backward_time=0.256, grad_norm=67.768, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.014e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 05:31:11,702 (trainer:779) INFO: 97epoch:train:2479-2891batch: iter_time=9.275e-04, forward_time=0.158, loss_ctc=8.194, loss_att=3.438, acc=0.980, loss=4.865, backward_time=0.257, grad_norm=87.834, clip=100.000, loss_scale=7.587e+33, optim_step_time=0.036, optim0_lr0=5.012e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 05:33:57,618 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 05:35:14,896 (trainer:779) INFO: 97epoch:train:2892-3304batch: iter_time=1.741e-04, forward_time=0.158, loss_ctc=8.197, loss_att=3.451, acc=0.980, loss=4.875, backward_time=0.256, grad_norm=71.204, clip=100.000, loss_scale=8.721e+33, optim_step_time=0.036, optim0_lr0=5.011e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 05:39:17,345 (trainer:779) INFO: 97epoch:train:3305-3717batch: iter_time=3.660e-04, forward_time=0.157, loss_ctc=8.169, loss_att=3.405, acc=0.979, loss=4.834, backward_time=0.255, grad_norm=70.189, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.010e-04, train_time=1.174
+[bmi2:0/4] 2024-07-18 05:43:20,415 (trainer:779) INFO: 97epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.208, loss_att=3.462, acc=0.977, loss=4.886, backward_time=0.256, grad_norm=71.793, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.008e-04, train_time=1.176
+[bmi2:0/4] 2024-07-18 05:47:23,823 (trainer:779) INFO: 97epoch:train:4131-4543batch: iter_time=5.586e-04, forward_time=0.157, loss_ctc=8.187, loss_att=3.456, acc=0.979, loss=4.875, backward_time=0.256, grad_norm=69.306, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.007e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 05:51:27,405 (trainer:779) INFO: 97epoch:train:4544-4956batch: iter_time=1.747e-04, forward_time=0.158, loss_ctc=8.173, loss_att=3.436, acc=0.979, loss=4.857, backward_time=0.256, grad_norm=69.420, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=5.006e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 05:53:28,247 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 05:55:31,281 (trainer:779) INFO: 97epoch:train:4957-5369batch: iter_time=2.379e-04, forward_time=0.158, loss_ctc=8.226, loss_att=3.445, acc=0.982, loss=4.879, backward_time=0.257, grad_norm=70.244, clip=100.000, loss_scale=3.875e+33, optim_step_time=0.036, optim0_lr0=5.005e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 05:59:36,099 (trainer:779) INFO: 97epoch:train:5370-5782batch: iter_time=9.439e-04, forward_time=0.158, loss_ctc=8.058, loss_att=3.403, acc=0.981, loss=4.799, backward_time=0.257, grad_norm=68.978, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.003e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 06:03:39,489 (trainer:779) INFO: 97epoch:train:5783-6195batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.260, loss_att=3.480, acc=0.979, loss=4.914, backward_time=0.256, grad_norm=70.628, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.002e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 06:07:44,155 (trainer:779) INFO: 97epoch:train:6196-6608batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.235, loss_att=3.438, acc=0.979, loss=4.877, backward_time=0.257, grad_norm=65.790, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=5.001e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 06:11:47,618 (trainer:779) INFO: 97epoch:train:6609-7021batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.166, loss_att=3.416, acc=0.978, loss=4.841, backward_time=0.256, grad_norm=65.330, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.999e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 06:15:51,578 (trainer:779) INFO: 97epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.037, loss_att=3.419, acc=0.975, loss=4.805, backward_time=0.256, grad_norm=74.749, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.998e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 06:19:55,056 (trainer:779) INFO: 97epoch:train:7435-7847batch: iter_time=1.704e-04, forward_time=0.159, loss_ctc=8.151, loss_att=3.415, acc=0.980, loss=4.836, backward_time=0.256, grad_norm=65.868, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.997e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 06:24:00,321 (trainer:779) INFO: 97epoch:train:7848-8260batch: iter_time=0.002, forward_time=0.159, loss_ctc=8.058, loss_att=3.410, acc=0.979, loss=4.805, backward_time=0.256, grad_norm=67.653, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.996e-04, train_time=1.187
+[bmi2:0/4] 2024-07-18 06:26:39,275 (trainer:365) INFO: 97epoch results: [train] iter_time=0.001, forward_time=0.158, loss_ctc=8.163, loss_att=3.436, acc=0.979, loss=4.854, backward_time=0.256, grad_norm=70.246, clip=100.000, loss_scale=4.512e+33, optim_step_time=0.036, optim0_lr0=5.008e-04, train_time=1.190, time=1 hour, 21 minutes and 59.57 seconds, total_count=801899, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.026, cer_ctc=0.039, loss_att=6.046, acc=0.950, cer=0.032, wer=0.482, loss=7.240, time=44.16 seconds, total_count=3298, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 49.72 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 06:26:44,580 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 06:26:44,627 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/96epoch.pth
+[bmi2:0/4] 2024-07-18 06:26:44,628 (trainer:299) INFO: 98/100epoch started. Estimated time to finish: 4 hours, 12 minutes and 11.96 seconds
+[bmi2:0/4] 2024-07-18 06:31:30,827 (trainer:779) INFO: 98epoch:train:1-413batch: iter_time=0.002, forward_time=0.155, loss_ctc=8.166, loss_att=3.445, acc=0.979, loss=4.861, backward_time=0.257, grad_norm=69.292, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.994e-04, train_time=1.387
+[bmi2:0/4] 2024-07-18 06:35:35,664 (trainer:779) INFO: 98epoch:train:414-826batch: iter_time=8.815e-04, forward_time=0.159, loss_ctc=8.102, loss_att=3.418, acc=0.977, loss=4.823, backward_time=0.257, grad_norm=66.140, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.993e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 06:39:40,825 (trainer:779) INFO: 98epoch:train:827-1239batch: iter_time=0.003, forward_time=0.159, loss_ctc=8.004, loss_att=3.405, acc=0.979, loss=4.785, backward_time=0.257, grad_norm=67.578, clip=100.000, loss_scale=4.764e+33, optim_step_time=0.036, optim0_lr0=4.992e-04, train_time=1.188
+[bmi2:0/4] 2024-07-18 06:43:44,759 (trainer:779) INFO: 98epoch:train:1240-1652batch: iter_time=2.881e-04, forward_time=0.158, loss_ctc=8.066, loss_att=3.387, acc=0.979, loss=4.791, backward_time=0.256, grad_norm=73.372, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.990e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 06:47:48,597 (trainer:779) INFO: 98epoch:train:1653-2065batch: iter_time=9.155e-04, forward_time=0.157, loss_ctc=8.066, loss_att=3.429, acc=0.979, loss=4.820, backward_time=0.257, grad_norm=64.748, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.989e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 06:51:52,554 (trainer:779) INFO: 98epoch:train:2066-2478batch: iter_time=8.366e-04, forward_time=0.158, loss_ctc=8.174, loss_att=3.458, acc=0.978, loss=4.873, backward_time=0.256, grad_norm=70.687, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.988e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 06:55:55,948 (trainer:779) INFO: 98epoch:train:2479-2891batch: iter_time=7.507e-04, forward_time=0.157, loss_ctc=8.155, loss_att=3.423, acc=0.981, loss=4.843, backward_time=0.257, grad_norm=66.567, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.987e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 06:58:10,609 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 07:00:00,029 (trainer:779) INFO: 98epoch:train:2892-3304batch: iter_time=1.664e-04, forward_time=0.158, loss_ctc=8.185, loss_att=3.450, acc=0.978, loss=4.871, backward_time=0.257, grad_norm=69.819, clip=100.000, loss_scale=4.020e+33, optim_step_time=0.036, optim0_lr0=4.985e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 07:04:03,684 (trainer:779) INFO: 98epoch:train:3305-3717batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.072, loss_att=3.383, acc=0.980, loss=4.789, backward_time=0.256, grad_norm=65.250, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.984e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 07:08:06,440 (trainer:779) INFO: 98epoch:train:3718-4130batch: iter_time=1.765e-04, forward_time=0.158, loss_ctc=8.200, loss_att=3.439, acc=0.981, loss=4.867, backward_time=0.255, grad_norm=70.257, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.983e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 07:12:09,848 (trainer:779) INFO: 98epoch:train:4131-4543batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.162, loss_att=3.428, acc=0.980, loss=4.848, backward_time=0.256, grad_norm=68.300, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.981e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 07:16:13,289 (trainer:779) INFO: 98epoch:train:4544-4956batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.210, loss_att=3.428, acc=0.977, loss=4.863, backward_time=0.255, grad_norm=73.238, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.980e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 07:20:18,139 (trainer:779) INFO: 98epoch:train:4957-5369batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.186, loss_att=3.428, acc=0.979, loss=4.855, backward_time=0.256, grad_norm=69.366, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.979e-04, train_time=1.186
+[bmi2:0/4] 2024-07-18 07:24:20,968 (trainer:779) INFO: 98epoch:train:5370-5782batch: iter_time=1.953e-04, forward_time=0.157, loss_ctc=8.136, loss_att=3.431, acc=0.980, loss=4.843, backward_time=0.255, grad_norm=68.985, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.978e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 07:28:24,009 (trainer:779) INFO: 98epoch:train:5783-6195batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.250, loss_att=3.466, acc=0.977, loss=4.901, backward_time=0.254, grad_norm=71.877, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.976e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 07:32:28,457 (trainer:779) INFO: 98epoch:train:6196-6608batch: iter_time=1.726e-04, forward_time=0.159, loss_ctc=8.191, loss_att=3.448, acc=0.980, loss=4.871, backward_time=0.257, grad_norm=73.606, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.975e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 07:36:31,521 (trainer:779) INFO: 98epoch:train:6609-7021batch: iter_time=1.811e-04, forward_time=0.157, loss_ctc=8.165, loss_att=3.413, acc=0.980, loss=4.839, backward_time=0.255, grad_norm=68.526, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.974e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 07:40:34,585 (trainer:779) INFO: 98epoch:train:7022-7434batch: iter_time=3.554e-04, forward_time=0.158, loss_ctc=8.190, loss_att=3.419, acc=0.981, loss=4.850, backward_time=0.255, grad_norm=67.569, clip=100.000, loss_scale=4.578e+33, optim_step_time=0.036, optim0_lr0=4.973e-04, train_time=1.176
+[bmi2:0/4] 2024-07-18 07:44:37,873 (trainer:779) INFO: 98epoch:train:7435-7847batch: iter_time=1.816e-04, forward_time=0.157, loss_ctc=8.153, loss_att=3.418, acc=0.979, loss=4.839, backward_time=0.255, grad_norm=70.440, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.971e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 07:48:42,091 (trainer:779) INFO: 98epoch:train:7848-8260batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.314, loss_att=3.478, acc=0.976, loss=4.929, backward_time=0.256, grad_norm=68.064, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.970e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 07:51:23,271 (trainer:365) INFO: 98epoch results: [train] iter_time=9.426e-04, forward_time=0.158, loss_ctc=8.158, loss_att=3.430, acc=0.979, loss=4.848, backward_time=0.256, grad_norm=69.189, clip=100.000, loss_scale=3.655e+33, optim_step_time=0.036, optim0_lr0=4.982e-04, train_time=1.190, time=1 hour, 22 minutes and 2.35 seconds, total_count=810166, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.027, cer_ctc=0.038, loss_att=6.030, acc=0.951, cer=0.032, wer=0.479, loss=7.229, time=43.52 seconds, total_count=3332, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 52.76 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 07:51:28,541 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 07:51:28,605 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/52epoch.pth, exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/97epoch.pth
+[bmi2:0/4] 2024-07-18 07:51:28,605 (trainer:299) INFO: 99/100epoch started. Estimated time to finish: 2 hours, 48 minutes and 11.17 seconds
+[bmi2:0/4] 2024-07-18 07:56:03,351 (trainer:779) INFO: 99epoch:train:1-413batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.151, loss_att=3.434, acc=0.977, loss=4.849, backward_time=0.255, grad_norm=72.690, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.969e-04, train_time=1.331
+[bmi2:0/4] 2024-07-18 08:00:07,015 (trainer:779) INFO: 99epoch:train:414-826batch: iter_time=0.001, forward_time=0.158, loss_ctc=8.124, loss_att=3.428, acc=0.978, loss=4.837, backward_time=0.255, grad_norm=69.037, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.967e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 08:04:11,050 (trainer:779) INFO: 99epoch:train:827-1239batch: iter_time=7.016e-04, forward_time=0.158, loss_ctc=8.153, loss_att=3.426, acc=0.979, loss=4.844, backward_time=0.256, grad_norm=69.305, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.966e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 08:08:15,628 (trainer:779) INFO: 99epoch:train:1240-1652batch: iter_time=5.008e-04, forward_time=0.160, loss_ctc=8.181, loss_att=3.447, acc=0.981, loss=4.867, backward_time=0.257, grad_norm=68.653, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.965e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 08:12:19,177 (trainer:779) INFO: 99epoch:train:1653-2065batch: iter_time=0.001, forward_time=0.157, loss_ctc=8.110, loss_att=3.403, acc=0.977, loss=4.815, backward_time=0.256, grad_norm=76.922, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.964e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 08:16:22,870 (trainer:779) INFO: 99epoch:train:2066-2478batch: iter_time=9.577e-04, forward_time=0.157, loss_ctc=8.116, loss_att=3.427, acc=0.979, loss=4.834, backward_time=0.256, grad_norm=70.306, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.962e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 08:20:23,726 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 08:20:26,588 (trainer:779) INFO: 99epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.071, loss_att=3.382, acc=0.979, loss=4.788, backward_time=0.256, grad_norm=66.486, clip=100.000, loss_scale=5.598e+33, optim_step_time=0.036, optim0_lr0=4.961e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 08:24:29,767 (trainer:779) INFO: 99epoch:train:2892-3304batch: iter_time=1.762e-04, forward_time=0.158, loss_ctc=8.140, loss_att=3.404, acc=0.982, loss=4.825, backward_time=0.256, grad_norm=68.237, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.960e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 08:28:33,594 (trainer:779) INFO: 99epoch:train:3305-3717batch: iter_time=0.001, forward_time=0.159, loss_ctc=8.329, loss_att=3.507, acc=0.977, loss=4.954, backward_time=0.256, grad_norm=82.535, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.959e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 08:32:36,754 (trainer:779) INFO: 99epoch:train:3718-4130batch: iter_time=3.821e-04, forward_time=0.157, loss_ctc=8.340, loss_att=3.468, acc=0.979, loss=4.930, backward_time=0.256, grad_norm=71.386, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.957e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 08:36:40,289 (trainer:779) INFO: 99epoch:train:4131-4543batch: iter_time=4.476e-04, forward_time=0.158, loss_ctc=8.184, loss_att=3.436, acc=0.978, loss=4.861, backward_time=0.256, grad_norm=71.309, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.956e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 08:40:42,521 (trainer:779) INFO: 99epoch:train:4544-4956batch: iter_time=2.428e-04, forward_time=0.156, loss_ctc=8.230, loss_att=3.429, acc=0.980, loss=4.869, backward_time=0.256, grad_norm=67.800, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.955e-04, train_time=1.172
+[bmi2:0/4] 2024-07-18 08:44:47,028 (trainer:779) INFO: 99epoch:train:4957-5369batch: iter_time=8.620e-04, forward_time=0.159, loss_ctc=8.152, loss_att=3.428, acc=0.981, loss=4.845, backward_time=0.256, grad_norm=63.267, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.954e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 08:48:49,757 (trainer:779) INFO: 99epoch:train:5370-5782batch: iter_time=8.391e-04, forward_time=0.157, loss_ctc=8.188, loss_att=3.425, acc=0.980, loss=4.854, backward_time=0.255, grad_norm=69.071, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.952e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 08:52:53,193 (trainer:779) INFO: 99epoch:train:5783-6195batch: iter_time=8.327e-04, forward_time=0.157, loss_ctc=8.146, loss_att=3.431, acc=0.979, loss=4.845, backward_time=0.256, grad_norm=64.928, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.951e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 08:56:57,920 (trainer:779) INFO: 99epoch:train:6196-6608batch: iter_time=4.908e-04, forward_time=0.159, loss_ctc=8.099, loss_att=3.401, acc=0.979, loss=4.811, backward_time=0.257, grad_norm=64.027, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.950e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 09:01:02,597 (trainer:779) INFO: 99epoch:train:6609-7021batch: iter_time=1.760e-04, forward_time=0.159, loss_ctc=8.140, loss_att=3.409, acc=0.981, loss=4.829, backward_time=0.256, grad_norm=65.708, clip=100.000, loss_scale=6.881e+33, optim_step_time=0.036, optim0_lr0=4.949e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 09:01:13,925 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 09:05:06,648 (trainer:779) INFO: 99epoch:train:7022-7434batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.176, loss_att=3.423, acc=0.979, loss=4.849, backward_time=0.256, grad_norm=68.810, clip=100.000, loss_scale=5.419e+33, optim_step_time=0.036, optim0_lr0=4.947e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 09:09:10,860 (trainer:779) INFO: 99epoch:train:7435-7847batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.213, loss_att=3.434, acc=0.978, loss=4.868, backward_time=0.255, grad_norm=71.433, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.946e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 09:13:15,011 (trainer:779) INFO: 99epoch:train:7848-8260batch: iter_time=9.212e-04, forward_time=0.157, loss_ctc=8.046, loss_att=3.396, acc=0.978, loss=4.791, backward_time=0.256, grad_norm=68.995, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.945e-04, train_time=1.181
+[bmi2:0/4] 2024-07-18 09:16:00,461 (trainer:365) INFO: 99epoch results: [train] iter_time=9.160e-04, forward_time=0.158, loss_ctc=8.165, loss_att=3.427, acc=0.979, loss=4.848, backward_time=0.256, grad_norm=69.540, clip=100.000, loss_scale=5.308e+33, optim_step_time=0.036, optim0_lr0=4.957e-04, train_time=1.188, time=1 hour, 21 minutes and 51.25 seconds, total_count=818433, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.003, cer_ctc=0.039, loss_att=5.878, acc=0.951, cer=0.031, wer=0.484, loss=7.116, time=44.45 seconds, total_count=3366, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 56.15 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 09:16:05,696 (trainer:431) INFO: There are no improvements in this epoch
+[bmi2:0/4] 2024-07-18 09:16:05,706 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/77epoch.pth
+[bmi2:0/4] 2024-07-18 09:16:05,706 (trainer:299) INFO: 100/100epoch started. Estimated time to finish: 1 hour, 24 minutes and 6.8 seconds
+[bmi2:0/4] 2024-07-18 09:20:51,489 (trainer:779) INFO: 100epoch:train:1-413batch: iter_time=0.001, forward_time=0.155, loss_ctc=8.118, loss_att=3.397, acc=0.980, loss=4.813, backward_time=0.255, grad_norm=70.328, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.944e-04, train_time=1.385
+[bmi2:0/4] 2024-07-18 09:24:55,644 (trainer:779) INFO: 100epoch:train:414-826batch: iter_time=7.172e-04, forward_time=0.159, loss_ctc=8.046, loss_att=3.378, acc=0.979, loss=4.778, backward_time=0.256, grad_norm=65.428, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.942e-04, train_time=1.182
+[bmi2:0/4] 2024-07-18 09:28:57,269 (trainer:779) INFO: 100epoch:train:827-1239batch: iter_time=0.002, forward_time=0.156, loss_ctc=8.036, loss_att=3.378, acc=0.978, loss=4.776, backward_time=0.254, grad_norm=74.908, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.941e-04, train_time=1.170
+[bmi2:0/4] 2024-07-18 09:33:00,045 (trainer:779) INFO: 100epoch:train:1240-1652batch: iter_time=2.024e-04, forward_time=0.156, loss_ctc=8.110, loss_att=3.422, acc=0.980, loss=4.828, backward_time=0.255, grad_norm=71.082, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.940e-04, train_time=1.175
+[bmi2:0/4] 2024-07-18 09:37:04,606 (trainer:779) INFO: 100epoch:train:1653-2065batch: iter_time=7.720e-04, forward_time=0.158, loss_ctc=7.988, loss_att=3.363, acc=0.980, loss=4.751, backward_time=0.257, grad_norm=63.096, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.036, optim0_lr0=4.939e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 09:41:09,545 (trainer:779) INFO: 100epoch:train:2066-2478batch: iter_time=0.002, forward_time=0.158, loss_ctc=8.076, loss_att=3.382, acc=0.978, loss=4.790, backward_time=0.256, grad_norm=67.335, clip=100.000, loss_scale=5.192e+33, optim_step_time=0.037, optim0_lr0=4.937e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 09:45:13,788 (trainer:779) INFO: 100epoch:train:2479-2891batch: iter_time=0.002, forward_time=0.157, loss_ctc=8.108, loss_att=3.399, acc=0.978, loss=4.812, backward_time=0.256, grad_norm=68.137, clip=100.000, loss_scale=6.654e+33, optim_step_time=0.036, optim0_lr0=4.936e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 09:48:43,303 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 09:49:17,493 (trainer:779) INFO: 100epoch:train:2892-3304batch: iter_time=1.735e-04, forward_time=0.158, loss_ctc=8.039, loss_att=3.374, acc=0.980, loss=4.773, backward_time=0.256, grad_norm=70.508, clip=100.000, loss_scale=9.654e+33, optim_step_time=0.036, optim0_lr0=4.935e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 09:52:00,224 (trainer:710) WARNING: The grad norm is inf. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 09:53:22,029 (trainer:779) INFO: 100epoch:train:3305-3717batch: iter_time=7.859e-04, forward_time=0.158, loss_ctc=8.105, loss_att=3.403, acc=0.980, loss=4.813, backward_time=0.257, grad_norm=71.349, clip=100.000, loss_scale=4.318e+33, optim_step_time=0.036, optim0_lr0=4.934e-04, train_time=1.184
+[bmi2:0/4] 2024-07-18 09:57:25,285 (trainer:779) INFO: 100epoch:train:3718-4130batch: iter_time=0.001, forward_time=0.156, loss_ctc=8.219, loss_att=3.425, acc=0.980, loss=4.863, backward_time=0.255, grad_norm=67.477, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.932e-04, train_time=1.177
+[bmi2:0/4] 2024-07-18 10:01:28,909 (trainer:779) INFO: 100epoch:train:4131-4543batch: iter_time=5.993e-04, forward_time=0.157, loss_ctc=8.032, loss_att=3.367, acc=0.977, loss=4.767, backward_time=0.255, grad_norm=64.015, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.931e-04, train_time=1.180
+[bmi2:0/4] 2024-07-18 10:05:32,528 (trainer:779) INFO: 100epoch:train:4544-4956batch: iter_time=4.924e-04, forward_time=0.157, loss_ctc=8.074, loss_att=3.386, acc=0.980, loss=4.792, backward_time=0.256, grad_norm=66.722, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.930e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 10:09:38,048 (trainer:779) INFO: 100epoch:train:4957-5369batch: iter_time=0.002, forward_time=0.159, loss_ctc=7.900, loss_att=3.348, acc=0.979, loss=4.714, backward_time=0.256, grad_norm=65.935, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.929e-04, train_time=1.189
+[bmi2:0/4] 2024-07-18 10:13:41,428 (trainer:779) INFO: 100epoch:train:5370-5782batch: iter_time=7.350e-04, forward_time=0.157, loss_ctc=8.076, loss_att=3.387, acc=0.981, loss=4.794, backward_time=0.256, grad_norm=65.966, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.927e-04, train_time=1.178
+[bmi2:0/4] 2024-07-18 10:17:45,528 (trainer:779) INFO: 100epoch:train:5783-6195batch: iter_time=8.243e-04, forward_time=0.157, loss_ctc=8.038, loss_att=3.353, acc=0.979, loss=4.758, backward_time=0.256, grad_norm=65.565, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.926e-04, train_time=1.183
+[bmi2:0/4] 2024-07-18 10:21:49,183 (trainer:779) INFO: 100epoch:train:6196-6608batch: iter_time=1.855e-04, forward_time=0.157, loss_ctc=8.118, loss_att=3.402, acc=0.982, loss=4.817, backward_time=0.257, grad_norm=65.717, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.925e-04, train_time=1.179
+[bmi2:0/4] 2024-07-18 10:25:49,757 (trainer:779) INFO: 100epoch:train:6609-7021batch: iter_time=0.001, forward_time=0.154, loss_ctc=8.122, loss_att=3.380, acc=0.975, loss=4.803, backward_time=0.253, grad_norm=65.760, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.924e-04, train_time=1.165
+[bmi2:0/4] 2024-07-18 10:29:54,849 (trainer:779) INFO: 100epoch:train:7022-7434batch: iter_time=9.670e-04, forward_time=0.159, loss_ctc=8.071, loss_att=3.386, acc=0.980, loss=4.792, backward_time=0.256, grad_norm=70.021, clip=100.000, loss_scale=2.596e+33, optim_step_time=0.036, optim0_lr0=4.922e-04, train_time=1.186
+[bmi2:0/4] 2024-07-18 10:30:55,767 (trainer:710) WARNING: The grad norm is nan. Skipping updating the model.
+[bmi2:0/4] 2024-07-18 10:33:59,522 (trainer:779) INFO: 100epoch:train:7435-7847batch: iter_time=3.087e-04, forward_time=0.159, loss_ctc=8.145, loss_att=3.406, acc=0.980, loss=4.828, backward_time=0.257, grad_norm=68.251, clip=100.000, loss_scale=1.615e+33, optim_step_time=0.036, optim0_lr0=4.921e-04, train_time=1.185
+[bmi2:0/4] 2024-07-18 10:38:04,385 (trainer:779) INFO: 100epoch:train:7848-8260batch: iter_time=1.812e-04, forward_time=0.158, loss_ctc=8.210, loss_att=3.428, acc=0.981, loss=4.863, backward_time=0.257, grad_norm=73.054, clip=100.000, loss_scale=1.298e+33, optim_step_time=0.037, optim0_lr0=4.920e-04, train_time=1.185
+O Channel 03 : 2[56000] -> 3[57000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO Connected all rings
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 3. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 3(=57000)
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 00 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 01 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 02 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO P2P is disabled between connected GPUs 2 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856045:1859958 [2] NCCL INFO Could not enable P2P between dev 2(=56000) and dev 1(=52000)
+bmi2:1856045:1859958 [2] NCCL INFO Channel 03 : 2[56000] -> 1[52000] via SHM/direct/direct
+bmi2:1856045:1859958 [2] NCCL INFO Connected all trees
+bmi2:1856045:1859958 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1856045:1859958 [2] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1856045:1859958 [2] NCCL INFO comm 0xab4fd10 rank 2 nranks 4 cudaDev 2 busId 56000 - Init COMPLETE
+ INFO Channel 03 : 3[57000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO Connected all rings
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 00 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 01 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 02 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO P2P is disabled between connected GPUs 3 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856046:1859959 [3] NCCL INFO Could not enable P2P between dev 3(=57000) and dev 2(=56000)
+bmi2:1856046:1859959 [3] NCCL INFO Channel 03 : 3[57000] -> 2[56000] via SHM/direct/direct
+bmi2:1856046:1859959 [3] NCCL INFO Connected all trees
+bmi2:1856046:1859959 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1856046:1859959 [3] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1856046:1859959 [3] NCCL INFO comm 0xb2bc150 rank 3 nranks 4 cudaDev 3 busId 57000 - Init COMPLETE
+O Channel 03 : 1[52000] -> 2[56000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO Connected all rings
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 2. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 2(=56000)
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 00 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 01 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 02 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO P2P is disabled between connected GPUs 1 and 0. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856044:1859957 [1] NCCL INFO Could not enable P2P between dev 1(=52000) and dev 0(=4f000)
+bmi2:1856044:1859957 [1] NCCL INFO Channel 03 : 1[52000] -> 0[4f000] via SHM/direct/direct
+bmi2:1856044:1859957 [1] NCCL INFO Connected all trees
+bmi2:1856044:1859957 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1856044:1859957 [1] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1856044:1859957 [1] NCCL INFO comm 0x7be5e20 rank 1 nranks 4 cudaDev 1 busId 52000 - Init COMPLETE
+bmi2:1856046:1859979 [3] NCCL INFO [Service thread] Connection closed by localRank 3
+bmi2:1856046:1856046 [3] NCCL INFO comm 0xb2bc150 rank 3 nranks 4 cudaDev 3 busId 57000 - Abort COMPLETE
+bmi2:1856045:1859980 [2] NCCL INFO [Service thread] Connection closed by localRank 2
+bmi2:1856045:1856045 [2] NCCL INFO comm 0xab4fd10 rank 2 nranks 4 cudaDev 2 busId 56000 - Abort COMPLETE
+bmi2:1856044:1859976 [1] NCCL INFO [Service thread] Connection closed by localRank 1
+bmi2:1856044:1856044 [1] NCCL INFO comm 0x7be5e20 rank 1 nranks 4 cudaDev 1 busId 52000 - Abort COMPLETE
+[bmi2:0/4] 2024-07-18 10:40:52,420 (trainer:365) INFO: 100epoch results: [train] iter_time=9.499e-04, forward_time=0.157, loss_ctc=8.081, loss_att=3.388, acc=0.979, loss=4.796, backward_time=0.256, grad_norm=68.012, clip=100.000, loss_scale=3.900e+33, optim_step_time=0.036, optim0_lr0=4.932e-04, train_time=1.191, time=1 hour, 22 minutes and 3.77 seconds, total_count=826700, gpu_max_cached_mem_GB=22.631, [valid] loss_ctc=10.100, cer_ctc=0.039, loss_att=5.896, acc=0.951, cer=0.031, wer=0.478, loss=7.157, time=44.3 seconds, total_count=3400, gpu_max_cached_mem_GB=22.631, [att_plot] time=1 minute and 58.64 seconds, total_count=0, gpu_max_cached_mem_GB=22.631
+[bmi2:0/4] 2024-07-18 10:40:56,563 (trainer:433) INFO: The best model has been updated: valid.acc
+[bmi2:0/4] 2024-07-18 10:40:56,570 (trainer:487) INFO: The model files were removed: exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/80epoch.pth
+[bmi2:0/4] 2024-07-18 10:40:56,570 (trainer:505) INFO: The training was finished at 100 epochs 
+[bmi2:0/4] 2024-07-18 10:40:56,571 (average_nbest_models:69) INFO: Averaging 10best models: criterion="valid.acc": exp/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave_10best.pth
+bmi2:1856035:1859955 [0] NCCL INFO Channel 03 : 0[4f000] -> 1[52000] via SHM/direct/direct
+bmi2:1856035:1859955 [0] NCCL INFO Connected all rings
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO P2P is disabled between connected GPUs 0 and 1. You can repress this message with NCCL_IGNORE_DISABLED_P2P=1.
+bmi2:1856035:1859955 [0] NCCL INFO Could not enable P2P between dev 0(=4f000) and dev 1(=52000)
+bmi2:1856035:1859955 [0] NCCL INFO Connected all trees
+bmi2:1856035:1859955 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+bmi2:1856035:1859955 [0] NCCL INFO 4 coll channels, 4 p2p channels, 2 p2p channels per peer
+bmi2:1856035:1859955 [0] NCCL INFO comm 0x9983420 rank 0 nranks 4 cudaDev 0 busId 4f000 - Init COMPLETE
+bmi2:1856035:1859978 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+bmi2:1856035:1856035 [0] NCCL INFO comm 0x9983420 rank 0 nranks 4 cudaDev 0 busId 4f000 - Abort COMPLETE
+# Accounting: time=136342 threads=1
+# Ended (code 0) at Thu Jul 18 10:41:02 HKT 2024, elapsed time 136342 seconds
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave.pth
new file mode 100644
index 0000000000000000000000000000000000000000..36460577cd03eb622dbd03a6560dc8f49d18ffa5
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1454a34a2d05ee8c95bfe0335072731efe31608febed0aad6b2b80e633b9bc0d
+size 398011411
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave_10best.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave_10best.pth
new file mode 100644
index 0000000000000000000000000000000000000000..36460577cd03eb622dbd03a6560dc8f49d18ffa5
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.ave_10best.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1454a34a2d05ee8c95bfe0335072731efe31608febed0aad6b2b80e633b9bc0d
+size 398011411
diff --git a/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.best.pth b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.best.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2ff0ee5449b9b8ee13277d3889f14fa2c370a5a3
--- /dev/null
+++ b/checkpoints/asr_transformer_HierDecayv2_woBias_XavierInit_Qv7/valid.acc.best.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dd0daaefe2355e6b0151b39c8f53b350fc4f531973a9cff2cb5dfb6fffeaddf
+size 398012507
diff --git a/src/Spike_driven/Q_transformer_encoder.py b/src/Spike_driven/Q_transformer_encoder.py
new file mode 100644
index 0000000000000000000000000000000000000000..e54403bcd158c121a920d86336fcd30fa5d379e4
--- /dev/null
+++ b/src/Spike_driven/Q_transformer_encoder.py
@@ -0,0 +1,397 @@
+# Copyright 2019 Shigeki Karita
+#  Apache 2.0  (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""Transformer encoder definition."""
+
+from typing import List, Optional, Tuple
+
+import torch
+import torch.nn as nn
+from typeguard import typechecked
+
+from espnet2.asr.ctc import CTC
+from espnet2.asr.encoder.abs_encoder import AbsEncoder
+from espnet.nets.pytorch_backend.nets_utils import make_pad_mask
+from espnet2.asr.encoder.Spike_driven.Spike_driven_modules.Q_attention import *
+
+from espnet.nets.pytorch_backend.transformer.embedding import PositionalEncoding
+from espnet.nets.pytorch_backend.transformer.layer_norm import LayerNorm
+# from espnet2.asr_transducer.normalization import RMSNorm
+from espnet.nets.pytorch_backend.transformer.multi_layer_conv import (
+    Conv1dLinear,
+    MultiLayeredConv1d,
+)
+from espnet2.asr.encoder.Spike_driven.Spike_driven_modules.Q_positionwise_feed_forward import Q_PositionwiseFeedForward, Q_GLU
+from espnet.nets.pytorch_backend.transformer.repeat import repeat
+from espnet.nets.pytorch_backend.transformer.subsampling import (
+    Conv1dSubsampling2,
+    Conv2dSubsampling,
+    Conv2dSubsampling1,
+    Conv2dSubsampling2,
+    Conv2dSubsampling6,
+    Conv2dSubsampling8,
+    TooShortUttError,
+    check_short_utt,
+)
+from espnet2.asr.encoder.Spike_driven.Q_trick import MultiSpike
+
+class Q_Transformer_EncoderLayer(nn.Module):
+    """Encoder layer module.
+
+    Args:
+        size (int): Input dimension.
+        self_attn (torch.nn.Module): Self-attention module instance.
+            `MultiHeadedAttention` or `RelPositionMultiHeadedAttention` instance
+            can be used as the argument.
+        feed_forward (torch.nn.Module): Feed-forward module instance.
+            `PositionwiseFeedForward`, `MultiLayeredConv1d`, or `Conv1dLinear` instance
+            can be used as the argument.
+        dropout_rate (float): Dropout rate.
+        normalize_before (bool): Whether to use layer_norm before the first block.
+        concat_after (bool): Whether to concat attention layer's input and output.
+            if True, additional linear will be applied.
+            i.e. x -> x + linear(concat(x, att(x)))
+            if False, no additional linear will be applied. i.e. x -> x + att(x)
+        stochastic_depth_rate (float): Proability to skip this layer.
+            During training, the layer may skip residual computation and return input
+            as-is with given probability.
+    """
+
+    def __init__(
+        self,
+        size,
+        self_attn,
+        feed_forward,
+        dropout_rate,
+        normalize_before=True,
+        concat_after=False,
+        stochastic_depth_rate=0.0,
+    ):
+        """Construct an EncoderLayer object."""
+        super(Q_Transformer_EncoderLayer, self).__init__()
+        self.self_attn = self_attn
+        self.feed_forward = feed_forward
+        self.norm1 = LayerNorm(size)
+        self.norm2 = LayerNorm(size)
+        self.dropout = nn.Dropout(dropout_rate)
+        self.size = size
+        self.normalize_before = normalize_before
+        self.concat_after = concat_after
+        if self.concat_after:
+            self.concat_linear = nn.Linear(size + size, size)
+        self.stochastic_depth_rate = stochastic_depth_rate
+        self.ATT_sn = MultiSpike(size)
+        self.FFN_sn = MultiSpike(size)
+
+    def forward(self, x, mask, iiter=None, cache=None):
+        """Compute encoded features.
+
+        Args:
+            x_input (torch.Tensor): Input tensor (#batch, time, size).
+            mask (torch.Tensor): Mask tensor for the input (#batch, 1, time).
+            cache (torch.Tensor): Cache tensor of the input (#batch, time - 1, size).
+
+        Returns:
+            torch.Tensor: Output tensor (#batch, time, size).
+            torch.Tensor: Mask tensor (#batch, 1, time).
+
+        """
+        skip_layer = False
+        # with stochastic depth, residual connection `x + f(x)` becomes
+        # `x <- x + 1 / (1 - p) * f(x)` at training time.
+        stoch_layer_coeff = 1.0
+        if self.training and self.stochastic_depth_rate > 0:
+            skip_layer = torch.rand(1).item() < self.stochastic_depth_rate
+            stoch_layer_coeff = 1.0 / (1 - self.stochastic_depth_rate)
+
+        if skip_layer:
+            if cache is not None:
+                x = torch.cat([cache, x], dim=1)
+            return x, mask
+
+        residual = x
+        if self.normalize_before:
+            x = self.norm1(x)
+
+        if cache is None:
+            x_q = x
+        else:
+            assert cache.shape == (x.shape[0], x.shape[1] - 1, self.size)
+            x_q = x[:, -1:, :]
+            residual = residual[:, -1:, :]
+            mask = None if mask is None else mask[:, -1:, :]
+
+        x_q = self.ATT_sn(x_q, iiter)
+        x = self.ATT_sn(x, iiter)
+        if self.concat_after:
+            x_concat = torch.cat((x, self.self_attn(x_q, x, x, mask, iiter)), dim=-1)
+            x = residual + stoch_layer_coeff * self.concat_linear(x_concat)
+        else:
+            x = residual + stoch_layer_coeff * self.dropout(
+                self.self_attn(x_q, x, x, mask, iiter)
+            )
+        if not self.normalize_before:
+            x = self.norm1(x)
+
+        residual = x
+        x = self.FFN_sn(x, iiter)
+        if self.normalize_before:
+            x = self.norm2(x)
+        x = residual + stoch_layer_coeff * self.dropout(self.feed_forward(x, iiter))
+        if not self.normalize_before:
+            x = self.norm2(x)
+
+        if cache is not None:
+            x = torch.cat([cache, x], dim=1)
+
+        return x, mask
+
+
+class Q_TransformerEncoder(AbsEncoder):
+    """Transformer encoder module.
+
+    Args:
+        input_size: input dim
+        output_size: dimension of attention
+        attention_heads: the number of heads of multi head attention
+        linear_units: the number of units of position-wise feed forward
+        num_blocks: the number of decoder blocks
+        dropout_rate: dropout rate
+        attention_dropout_rate: dropout rate in attention
+        positional_dropout_rate: dropout rate after adding positional encoding
+        input_layer: input layer type
+        pos_enc_class: PositionalEncoding or ScaledPositionalEncoding
+        normalize_before: whether to use layer_norm before the first block
+        concat_after: whether to concat attention layer's input and output
+            if True, additional linear will be applied.
+            i.e. x -> x + linear(concat(x, att(x)))
+            if False, no additional linear will be applied.
+            i.e. x -> x + att(x)
+        positionwise_layer_type: linear of conv1d
+        positionwise_conv_kernel_size: kernel size of positionwise conv1d layer
+        padding_idx: padding_idx for input_layer=embed
+    """
+
+    @typechecked
+    def __init__(
+        self,
+        input_size: int,
+        output_size: int = 256,
+        attention_heads: int = 4,
+        attention_layer_type: str = "selfattn",
+        linear_units: int = 2048,
+        num_blocks: int = 6,
+        dropout_rate: float = 0.1,
+        positional_dropout_rate: float = 0.1,
+        attention_dropout_rate: float = 0.0,
+        input_layer: Optional[str] = "conv2d",
+        pos_enc_class=PositionalEncoding,
+        normalize_before: bool = True,
+        concat_after: bool = False,
+        positionwise_layer_type: str = "FFN",
+        padding_idx: int = -1,
+        interctc_layer_idx: List[int] = [],
+        interctc_use_conditioning: bool = False,
+        layer_drop_rate: float = 0.0,
+    ):
+        super().__init__()
+        self._output_size = output_size
+
+        if input_layer == "linear":
+            self.embed = torch.nn.Sequential(
+                torch.nn.Linear(input_size, output_size),
+                torch.nn.LayerNorm(output_size),
+                torch.nn.Dropout(dropout_rate),
+                torch.nn.ReLU(),
+                pos_enc_class(output_size, positional_dropout_rate),
+            )
+        elif input_layer == "conv1d2":
+            self.embed = Conv1dSubsampling2(
+                input_size,
+                output_size,
+                dropout_rate,
+                pos_enc_class(output_size, positional_dropout_rate),
+            )
+        elif input_layer == "conv2d":
+            self.embed = Conv2dSubsampling(input_size, output_size, dropout_rate)
+        elif input_layer == "conv2d1":
+            self.embed = Conv2dSubsampling1(input_size, output_size, dropout_rate)
+        elif input_layer == "conv2d2":
+            self.embed = Conv2dSubsampling2(input_size, output_size, dropout_rate)
+        elif input_layer == "conv2d6":
+            self.embed = Conv2dSubsampling6(input_size, output_size, dropout_rate)
+        elif input_layer == "conv2d8":
+            self.embed = Conv2dSubsampling8(input_size, output_size, dropout_rate)
+        elif input_layer == "embed":
+            self.embed = torch.nn.Sequential(
+                torch.nn.Embedding(input_size, output_size, padding_idx=padding_idx),
+                pos_enc_class(output_size, positional_dropout_rate),
+            )
+        elif input_layer is None:
+            if input_size == output_size:
+                self.embed = None
+            else:
+                self.embed = torch.nn.Linear(input_size, output_size)
+        else:
+            raise ValueError("unknown input_layer: " + input_layer)
+        self.normalize_before = normalize_before
+        if attention_layer_type == "selfattn":
+            encoder_selfattn_layer = Q_MultiHeadedAttention
+            encoder_selfattn_layer_args = (
+                attention_heads, 
+                output_size, 
+                attention_dropout_rate
+            )
+        elif attention_layer_type == "selfattn_woSoftMax":
+            encoder_selfattn_layer = Q_MultiHeadedAttention_woSoftMax
+            encoder_selfattn_layer_args = (
+                attention_heads, 
+                output_size, 
+                attention_dropout_rate
+            ) 
+        elif attention_layer_type == "HierDecayv2":
+            encoder_selfattn_layer = Q_MultiHeadedAttention_HierDecay
+            encoder_selfattn_layer_args = (
+                attention_heads,
+                output_size,
+                attention_dropout_rate,
+            )
+        elif attention_layer_type == "HierDecay_woSoftMax":
+            encoder_selfattn_layer = Q_MultiHeadedAttention_HierDecay_woSoftMax
+            encoder_selfattn_layer_args = (
+                attention_heads,
+                output_size,
+                attention_dropout_rate,
+            )
+        
+        else:
+            raise ValueError("unknown encoder_attn_layer: " + attention_layer_type)
+
+        positionwise_layer = Q_PositionwiseFeedForward
+        positionwise_layer_args = (
+            output_size,
+            linear_units,
+            dropout_rate,
+        )
+        
+
+        if "HierDecay" in attention_layer_type:
+            self.encoders = repeat(
+                num_blocks,
+                lambda lnum: Q_Transformer_EncoderLayer(
+                    output_size,
+                    encoder_selfattn_layer(*encoder_selfattn_layer_args, lnum),
+                    positionwise_layer(*positionwise_layer_args),
+                    dropout_rate,
+                    normalize_before,
+                    concat_after,
+                ),
+                layer_drop_rate,
+            )
+        else:
+            self.encoders = repeat(
+                num_blocks,
+                lambda lnum: Q_Transformer_EncoderLayer(
+                    output_size,
+                    encoder_selfattn_layer(*encoder_selfattn_layer_args),
+                    positionwise_layer(*positionwise_layer_args),
+                    dropout_rate,
+                    normalize_before,
+                    concat_after,
+                ),
+                layer_drop_rate,
+            )
+        
+        if self.normalize_before:
+            self.after_norm = LayerNorm(output_size)
+
+        self.interctc_layer_idx = interctc_layer_idx
+        if len(interctc_layer_idx) > 0:
+            assert 0 < min(interctc_layer_idx) and max(interctc_layer_idx) < num_blocks
+        self.interctc_use_conditioning = interctc_use_conditioning
+        self.conditioning_layer = None
+
+    def output_size(self) -> int:
+        return self._output_size
+
+    def forward(
+        self,
+        xs_pad: torch.Tensor,
+        ilens: torch.Tensor,
+        iiter: int = 0,
+        prev_states: torch.Tensor = None,
+        ctc: CTC = None,
+        return_all_hs: bool = False,
+    ) -> Tuple[torch.Tensor, torch.Tensor, Optional[torch.Tensor]]:
+        """Embed positions in tensor.
+
+        Args:
+            xs_pad: input tensor (B, L, D)
+            ilens: input length (B)
+            prev_states: Not to be used now.
+            ctc (CTC): ctc module for intermediate CTC loss
+            return_all_hs (bool): whether to return all hidden states
+
+        Returns:
+            position embedded tensor and mask
+        """
+        masks = (~make_pad_mask(ilens)[:, None, :]).to(xs_pad.device)
+        # print('iiter:{}'.format(iiter))
+        if self.embed is None:
+            xs_pad = xs_pad
+        elif (
+            isinstance(self.embed, Conv2dSubsampling)
+            or isinstance(self.embed, Conv1dSubsampling2)
+            or isinstance(self.embed, Conv2dSubsampling1)
+            or isinstance(self.embed, Conv2dSubsampling2)
+            or isinstance(self.embed, Conv2dSubsampling6)
+            or isinstance(self.embed, Conv2dSubsampling8)
+        ):
+            short_status, limit_size = check_short_utt(self.embed, xs_pad.size(1))
+            if short_status:
+                raise TooShortUttError(
+                    f"has {xs_pad.size(1)} frames and is too short for subsampling "
+                    + f"(it needs more than {limit_size} frames), return empty results",
+                    xs_pad.size(1),
+                    limit_size,
+                )
+            xs_pad, masks = self.embed(xs_pad, masks)
+        else:
+            xs_pad = self.embed(xs_pad)
+        intermediate_outs = []
+        if len(self.interctc_layer_idx) == 0:
+            for layer_idx, encoder_layer in enumerate(self.encoders):
+                xs_pad, masks = encoder_layer(xs_pad, masks, iiter)
+                if return_all_hs:
+                    if isinstance(xs_pad, tuple):
+                        intermediate_outs.append(xs_pad[0])
+                    else:
+                        intermediate_outs.append(xs_pad)
+
+                
+        else:
+            for layer_idx, encoder_layer in enumerate(self.encoders):
+                xs_pad, masks = encoder_layer(xs_pad, masks, iiter)
+
+                if layer_idx + 1 in self.interctc_layer_idx:
+                    encoder_out = xs_pad
+
+                    # intermediate outputs are also normalized
+                    if self.normalize_before:
+                        encoder_out = self.after_norm(encoder_out)
+
+                    intermediate_outs.append((layer_idx + 1, encoder_out))
+
+                    if self.interctc_use_conditioning:
+                        ctc_out = ctc.softmax(encoder_out)
+                        xs_pad = xs_pad + self.conditioning_layer(ctc_out)
+
+        if self.normalize_before:
+            xs_pad = self.after_norm(xs_pad)
+
+        olens = masks.squeeze(1).sum(1)
+        # from IPython import embed; embed()
+        if len(intermediate_outs) > 0:
+            return (xs_pad, intermediate_outs), olens, None
+        
+        return xs_pad, olens, None
diff --git a/src/Spike_driven/Q_trick.py b/src/Spike_driven/Q_trick.py
new file mode 100644
index 0000000000000000000000000000000000000000..3bda801ee03dadafb17f02a02ab5144f9baafd49
--- /dev/null
+++ b/src/Spike_driven/Q_trick.py
@@ -0,0 +1,46 @@
+import torch
+import torch.nn as nn
+
+class quant(torch.autograd.Function):
+    @staticmethod
+    def forward(ctx, input, T):
+        ctx.save_for_backward(input)
+        ctx.T = T
+        return torch.round(torch.clamp(input, min=0, max=T))
+        
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, = ctx.saved_tensors
+        grad_input = grad_output.clone()
+        grad_input[input < 0] = 0
+        grad_input[input > ctx.T] = 0
+        return grad_input, None
+
+
+
+class MultiSpike(torch.nn.Module):
+    def __init__(self, dim: int, T=4):
+        super().__init__()
+        self.T = T
+        self.spike = quant()
+        self.momentum = 0.1
+        self.eps = 1e-5
+        self.register_buffer("running_stats", torch.zeros(dim))
+    
+    def __repr__(self):
+        return f"MultiSpike(T={self.T})"    
+    
+    def forward(self, x, iiter=0):
+        #v7
+        # print('iiter:{}'.format(iiter))
+        if self.training:
+            Stats = x.max(dim=0).values.max(dim=0).values
+            # Stats = x.abs().mean(dim=[0,1])
+            with torch.no_grad():
+                self.running_stats = self.momentum * Stats + (1-self.momentum) * self.running_stats
+        else:
+            Stats = self.running_stats
+
+        scale = self.T / (Stats[None, None, :] + self.eps)
+
+        return self.spike.apply(scale* x, self.T) / scale
\ No newline at end of file
diff --git a/src/Spike_driven/Spike_driven_modules/Q_attention.py b/src/Spike_driven/Spike_driven_modules/Q_attention.py
new file mode 100644
index 0000000000000000000000000000000000000000..ff3d922e8ca4949e5ccfcda4c2a8c41a6ba33610
--- /dev/null
+++ b/src/Spike_driven/Spike_driven_modules/Q_attention.py
@@ -0,0 +1,455 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+# Copyright 2019 Shigeki Karita
+#  Apache 2.0  (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""Multi-Head Attention layer definition."""
+
+import math
+import numpy
+import torch
+from torch import nn
+from espnet2.asr.encoder.Spike_driven.Q_trick import MultiSpike
+
+
+
+class Q_MultiHeadedAttention_HierDecay(nn.Module):
+    """Implementation of HD-RepSSA_S
+
+    Args:
+        n_head (int): The number of heads.
+        n_feat (int): The number of features.
+        dropout_rate (float): Dropout rate.
+        layer_id (int): Layer ID for decay calculation.
+
+    """
+
+    def __init__(self, n_head, n_feat, dropout_rate, layer_id):
+        """Construct an MultiHeadedAttention object."""
+        super(Q_MultiHeadedAttention_HierDecay_v2, self).__init__()
+        assert n_feat % n_head == 0
+        # We assume d_v always equals d_k
+        self.d_k = n_feat // n_head
+        self.h = n_head
+        self.linear_q = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_k = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_v = nn.Linear(n_feat, n_feat, bias=False)
+        self.v_sn = MultiSpike(n_feat)
+        self.output_sn = MultiSpike(n_feat)
+        self.linear_out = nn.Linear(n_feat, n_feat)
+        self.attn = None
+        self.dropout = nn.Dropout(p=dropout_rate)
+        
+        # Hierarchical decay calculation
+        layer_decay = 1 - 2 ** (-5 - layer_id)
+        decay = torch.log(torch.tensor(layer_decay).repeat(n_head))
+        self.register_buffer("decay", decay)
+
+    def forward_qkv(self, query, key, value, iiter):
+        """Transform query, key and value.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+
+        Returns:
+            torch.Tensor: Transformed query tensor (#batch, n_head, time1, d_k).
+            torch.Tensor: Transformed key tensor (#batch, n_head, time2, d_k).
+            torch.Tensor: Transformed value tensor (#batch, n_head, time2, d_k).
+
+        """
+        n_batch = query.size(0)
+        q = self.linear_q(query).view(n_batch, -1, self.h, self.d_k)
+        k = self.linear_k(key).view(n_batch, -1, self.h, self.d_k)
+        v = self.v_sn(self.linear_v(value)).view(n_batch, -1, self.h, self.d_k)
+        q = q.transpose(1, 2)  # (batch, head, time1, d_k)
+        k = k.transpose(1, 2)  # (batch, head, time2, d_k)
+        v = v.transpose(1, 2)  # (batch, head, time2, d_k)
+        
+        return q, k, v
+
+    def forward_attention(self, value, scores, mask, inner_mask, iiter):
+        """Compute attention context vector.
+
+        Args:
+            value (torch.Tensor): Transformed value (#batch, n_head, time2, d_k).
+            scores (torch.Tensor): Attention score (#batch, n_head, time1, time2).
+            mask (torch.Tensor): Mask (#batch, 1, time2) or (#batch, time1, time2).
+            inner_mask (torch.Tensor): Inner mask for hierarchical decay.
+
+        Returns:
+            torch.Tensor: Transformed value (#batch, time1, d_model)
+                weighted by the attention score (#batch, time1, time2).
+
+        """
+        n_batch = value.size(0)
+        scores += inner_mask
+        if mask is not None:
+            mask = mask.unsqueeze(1).eq(0)  # (batch, 1, *, time2)
+            min_value = torch.finfo(scores.dtype).min
+            scores = scores.masked_fill(mask, min_value)
+            self.attn = torch.softmax(scores, dim=-1).masked_fill(
+                mask, 0.0
+            )  # (batch, head, time1, time2)
+        else:
+            self.attn = torch.softmax(scores, dim=-1)  # (batch, head, time1, time2)
+        p_attn = self.dropout(self.attn)
+
+        x = torch.matmul(p_attn, value)  # (batch, head, time1, d_k)
+        x = (
+            x.transpose(1, 2).contiguous().view(n_batch, -1, self.h * self.d_k)
+        )  # (batch, time1, d_model)
+        x = self.output_sn(x)
+        return self.linear_out(x)  # (batch, time1, d_model)
+    
+    def forward(self, query, key, value, mask, iiter):
+        """Compute scaled dot product attention.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+            mask (torch.Tensor): Mask tensor (#batch, 1, time2) or
+                (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Output tensor (#batch, time1, d_model).
+
+        """
+        slen = query.shape[1] if query.shape[1]==key.shape[1] else mask.shape[1]
+        index = torch.arange(slen).to(self.decay)
+        inner_mask = torch.abs(index.view(slen,1) - index.view(1, slen))
+        inner_mask = inner_mask * self.decay[:, None, None]
+
+        q, k, v = self.forward_qkv(query, key, value, iiter)
+        scores = torch.matmul(q, k.transpose(-2, -1))/ math.sqrt(self.d_k)
+        return self.forward_attention(v, scores, mask, inner_mask, iiter)
+
+class Q_MultiHeadedAttention_HierDecay_woSoftMax(Q_MultiHeadedAttention):
+    """Implementation of HD-RepSSA_S
+
+    Args:
+        n_head (int): The number of heads.
+        n_feat (int): The number of features.
+        dropout_rate (float): Dropout rate.
+        layer_id (int): Layer ID for decay calculation.
+
+    """
+
+    def __init__(self, n_head, n_feat, dropout_rate, layer_id):
+        """Construct an MultiHeadedAttention object."""
+        
+        super().__init__(n_head, n_feat, dropout_rate)
+        assert n_feat % n_head == 0
+        # We assume d_v always equals d_k
+        self.d_k = n_feat // n_head
+        self.h = n_head
+        self.linear_q = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_k = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_v = nn.Linear(n_feat, n_feat, bias=False)
+        self.q_sn = MultiSpike(n_feat)
+        self.k_sn = MultiSpike(n_feat)
+        self.v_sn = MultiSpike(n_feat)
+        self.output_sn = MultiSpike(n_feat)
+        self.linear_out = nn.Linear(n_feat, n_feat)
+        self.attn = None
+        self.dropout = nn.Dropout(p=dropout_rate)
+        
+        layer_decay = 1 - 2 ** (-5 - layer_id)
+        
+        decay = torch.log(torch.tensor(layer_decay).repeat(n_head))
+        self.register_buffer("decay", decay)
+
+        self.ln = torch.nn.LayerNorm(self.d_k)
+
+
+    def forward_qkv(self, query, key, value, iiter):
+        """Transform query, key and value.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+
+        Returns:
+            torch.Tensor: Transformed query tensor (#batch, n_head, time1, d_k).
+            torch.Tensor: Transformed key tensor (#batch, n_head, time2, d_k).
+            torch.Tensor: Transformed value tensor (#batch, n_head, time2, d_k).
+
+        """
+        n_batch = query.size(0)
+        q = self.q_sn(self.linear_q(query)).view(n_batch, -1, self.h, self.d_k)
+        k = self.k_sn(self.linear_k(key)).view(n_batch, -1, self.h, self.d_k)
+        v = self.v_sn(self.linear_v(value)).view(n_batch, -1, self.h, self.d_k)
+        q = q.transpose(1, 2)  # (batch, head, time1, d_k)
+        k = k.transpose(1, 2)  # (batch, head, time2, d_k)
+        v = v.transpose(1, 2)  # (batch, head, time2, d_k)
+        
+        return q, k, v
+
+    def forward_attention(self, value, scores, mask, inner_mask, iiter):
+        """Compute attention context vector.
+
+        Args:
+            value (torch.Tensor): Transformed value (#batch, n_head, time2, d_k).
+            scores (torch.Tensor): Attention score (#batch, n_head, time1, time2).
+            mask (torch.Tensor): Mask (#batch, 1, time2) or (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Transformed value (#batch, time1, d_model)
+                weighted by the attention score (#batch, time1, time2).
+
+        """
+        n_batch = value.size(0)
+        scores = scores / scores.detach().abs().sum(dim=-1, keepdim=True).clamp(min=1, max=5e4)
+        if mask is not None:
+            mask = mask.unsqueeze(1).eq(0)  # (batch, 1, *, time2)
+            min_value = torch.finfo(scores.dtype).min
+            scores = scores.masked_fill(mask, min_value)
+            self.attn = scores.masked_fill(
+                mask, 0.0
+            )  # (batch, head, time1, time2)
+        else:
+            self.attn = scores  # (batch, head, time1, time2)
+        self.attn = inner_mask * self.attn
+        p_attn = self.dropout(self.attn)
+
+        x = self.ln(torch.matmul(p_attn, value))  # (batch, head, time1, d_k)
+        x = (
+            x.transpose(1, 2).contiguous().view(n_batch, -1, self.h * self.d_k)
+        )  # (batch, time1, d_model)
+        x = self.output_sn(x)
+        return self.linear_out(x)  # (batch, time1, d_model)
+
+    def forward(self, query, key, value, mask, iiter):
+        """Compute scaled dot product attention.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+            mask (torch.Tensor): Mask tensor (#batch, 1, time2) or
+                (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Output tensor (#batch, time1, d_model).
+
+        """
+        slen = query.shape[1] if query.shape[1]==key.shape[1] else mask.shape[1]
+        index = torch.arange(slen).to(self.decay)
+        inner_mask = torch.abs(index.view(slen,1) - index.view(1, slen))
+        inner_mask = torch.exp(inner_mask * self.decay[:, None, None])
+
+        q, k, v = self.forward_qkv(query, key, value, iiter)
+        scores = torch.matmul(q, k.transpose(-2, -1))
+        return self.forward_attention(v, scores, mask, inner_mask, iiter)
+
+class Q_MultiHeadedAttention(nn.Module):
+    """Multi-Head Attention layer.
+
+    Args:
+        n_head (int): The number of heads.
+        n_feat (int): The number of features.
+        dropout_rate (float): Dropout rate.
+
+    """
+
+    def __init__(self, n_head, n_feat, dropout_rate):
+        """Construct an MultiHeadedAttention object."""
+        super(Q_MultiHeadedAttention, self).__init__()
+        assert n_feat % n_head == 0
+        # We assume d_v always equals d_k
+        self.d_k = n_feat // n_head
+        self.h = n_head
+        self.linear_q = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_k = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_v = nn.Linear(n_feat, n_feat, bias=False)
+        self.v_sn = MultiSpike(n_feat)
+        self.output_sn = MultiSpike(n_feat)
+        self.linear_out = nn.Linear(n_feat, n_feat)
+        self.attn = None
+        self.dropout = nn.Dropout(p=dropout_rate)
+
+    def forward_qkv(self, query, key, value, iiter):
+        """Transform query, key and value.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+
+        Returns:
+            torch.Tensor: Transformed query tensor (#batch, n_head, time1, d_k).
+            torch.Tensor: Transformed key tensor (#batch, n_head, time2, d_k).
+            torch.Tensor: Transformed value tensor (#batch, n_head, time2, d_k).
+
+        """
+        n_batch = query.size(0)
+        q = self.linear_q(query).view(n_batch, -1, self.h, self.d_k)
+        k = self.linear_k(key).view(n_batch, -1, self.h, self.d_k)
+        v = self.v_sn(self.linear_v(value)).view(n_batch, -1, self.h, self.d_k)
+        q = q.transpose(1, 2)  # (batch, head, time1, d_k)
+        k = k.transpose(1, 2)  # (batch, head, time2, d_k)
+        v = v.transpose(1, 2)  # (batch, head, time2, d_k)
+
+        return q, k, v
+
+    def forward_attention(self, value, scores, mask, iiter):
+        """Compute attention context vector.
+
+        Args:
+            value (torch.Tensor): Transformed value (#batch, n_head, time2, d_k).
+            scores (torch.Tensor): Attention score (#batch, n_head, time1, time2).
+            mask (torch.Tensor): Mask (#batch, 1, time2) or (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Transformed value (#batch, time1, d_model)
+                weighted by the attention score (#batch, time1, time2).
+
+        """
+        n_batch = value.size(0)
+        if mask is not None:
+            mask = mask.unsqueeze(1).eq(0)  # (batch, 1, *, time2)
+            min_value = torch.finfo(scores.dtype).min
+            scores = scores.masked_fill(mask, min_value)
+            self.attn = torch.softmax(scores, dim=-1).masked_fill(
+                mask, 0.0
+            )  # (batch, head, time1, time2)
+        else:
+            self.attn = torch.softmax(scores, dim=-1)  # (batch, head, time1, time2)
+
+        p_attn = self.dropout(self.attn)
+        x = torch.matmul(p_attn, value)  # (batch, head, time1, d_k)
+        
+        x = (
+            x.transpose(1, 2).contiguous().view(n_batch, -1, self.h * self.d_k)
+        )  # (batch, time1, d_model)
+        x = self.output_sn(x)
+        return self.linear_out(x)  # (batch, time1, d_model)
+
+    def forward(self, query, key, value, mask, iiter):
+        """Compute scaled dot product attention.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+            mask (torch.Tensor): Mask tensor (#batch, 1, time2) or
+                (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Output tensor (#batch, time1, d_model).
+
+        """
+        q, k, v = self.forward_qkv(query, key, value, iiter)
+        scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(self.d_k)
+        return self.forward_attention(v, scores, mask, iiter)
+
+
+
+class Q_MultiHeadedAttention_woSoftMax(Q_MultiHeadedAttention):
+    """Multi-Head Attention layer without SoftMax.
+
+    Args:
+        n_head (int): The number of heads.
+        n_feat (int): The number of features.
+        dropout_rate (float): Dropout rate.
+
+    """
+
+    def __init__(self, n_head, n_feat, dropout_rate):
+        """Construct an MultiHeadedAttention object."""
+        super().__init__(n_head, n_feat, dropout_rate)
+        assert n_feat % n_head == 0
+        # We assume d_v always equals d_k
+        self.d_k = n_feat // n_head
+        self.h = n_head
+        self.linear_q = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_k = nn.Linear(n_feat, n_feat, bias=False)
+        self.linear_v = nn.Linear(n_feat, n_feat, bias=False)
+        self.q_sn = MultiSpike(n_feat)
+        self.k_sn = MultiSpike(n_feat)
+        self.v_sn = MultiSpike(n_feat)
+        self.output_sn = MultiSpike(n_feat)
+        self.linear_out = nn.Linear(n_feat, n_feat)
+        self.attn = None
+        # self.dropout = nn.Dropout(p=dropout_rate)
+        self.ln = torch.nn.LayerNorm(self.d_k)
+        # self.scale = self.d_k ** -0.5
+    def forward_qkv(self, query, key, value, iiter):
+        """Transform query, key and value.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+
+        Returns:
+            torch.Tensor: Transformed query tensor (#batch, n_head, time1, d_k).
+            torch.Tensor: Transformed key tensor (#batch, n_head, time2, d_k).
+            torch.Tensor: Transformed value tensor (#batch, n_head, time2, d_k).
+
+        """
+        n_batch = query.size(0)
+        k = self.linear_k(key).view(n_batch, -1, self.h, self.d_k)
+        v = self.linear_v(value).view(n_batch, -1, self.h, self.d_k)
+        q = self.q_sn(self.linear_q(query)).view(n_batch, -1, self.h, self.d_k)
+        q = q.transpose(1, 2)  # (batch, head, time1, d_k)
+        k = k.transpose(1, 2)  # (batch, head, time2, d_k)
+        v = v.transpose(1, 2)  # (batch, head, time2, d_k)
+
+        return q, k, v
+
+    def forward_attention(self, value, scores, mask, iiter):
+        """Compute attention context vector.
+
+        Args:
+            value (torch.Tensor): Transformed value (#batch, n_head, time2, d_k).
+            scores (torch.Tensor): Attention score (#batch, n_head, time1, time2).
+            mask (torch.Tensor): Mask (#batch, 1, time2) or (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Transformed value (#batch, time1, d_model)
+                weighted by the attention score (#batch, time1, time2).
+
+        """
+        n_batch = value.size(0)
+        scores = scores / scores.detach().abs().sum(dim=-1, keepdim=True).clamp(min=1, max=5e4)
+        if mask is not None:
+            mask = mask.unsqueeze(1).eq(0)  # (batch, 1, *, time2)
+            min_value = torch.finfo(scores.dtype).min
+            scores = scores.masked_fill(mask, min_value)
+            self.attn = scores.masked_fill(
+                mask, 0.0
+            )  # (batch, head, time1, time2)
+        else:
+            self.attn = scores  # (batch, head, time1, time2)
+        p_attn = self.dropout(self.attn)
+        x = self.ln(torch.matmul(p_attn, value))# (batch, head, time1, d_k)
+        # x = torch.matmul(p_attn, value) * self.scale# (batch, head, time1, d_k)
+        x = (
+            x.transpose(1, 2).contiguous().view(n_batch, -1, self.h * self.d_k)
+        )  # (batch, time1, d_model)
+        x = self.output_sn(x)
+        return self.linear_out(x)  # (batch, time1, d_model)
+
+    def forward(self, query, key, value, mask, iiter):
+        """Compute scaled dot product attention.
+
+        Args:
+            query (torch.Tensor): Query tensor (#batch, time1, size).
+            key (torch.Tensor): Key tensor (#batch, time2, size).
+            value (torch.Tensor): Value tensor (#batch, time2, size).
+            mask (torch.Tensor): Mask tensor (#batch, 1, time2) or
+                (#batch, time1, time2).
+
+        Returns:
+            torch.Tensor: Output tensor (#batch, time1, d_model).
+
+        """
+        q, k, v = self.forward_qkv(query, key, value, iiter)
+        scores = torch.matmul(q, k.transpose(-2, -1))
+        return self.forward_attention(v, scores, mask, iiter)
+
diff --git a/src/Spike_driven/Spike_driven_modules/Q_positionwise_feed_forward.py b/src/Spike_driven/Spike_driven_modules/Q_positionwise_feed_forward.py
new file mode 100644
index 0000000000000000000000000000000000000000..8536be1f498117dbeb25d583abc1a38c860439f2
--- /dev/null
+++ b/src/Spike_driven/Spike_driven_modules/Q_positionwise_feed_forward.py
@@ -0,0 +1,35 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+# Copyright 2019 Shigeki Karita
+#  Apache 2.0  (http://www.apache.org/licenses/LICENSE-2.0)
+
+"""Positionwise feed forward layer definition."""
+
+import torch
+from torch import nn
+from espnet2.asr.encoder.Spike_driven.Q_trick import MultiSpike,
+
+class Q_PositionwiseFeedForward(torch.nn.Module):
+    """Positionwise feed forward layer.
+
+    Args:
+        idim (int): Input dimenstion.
+        hidden_units (int): The number of hidden units.
+        dropout_rate (float): Dropout rate.
+
+    """
+
+    def __init__(self, idim, hidden_units, dropout_rate):
+        """Construct an PositionwiseFeedForward object."""
+        super(Q_PositionwiseFeedForward, self).__init__()
+        self.w_1 = torch.nn.Linear(idim, hidden_units)
+        self.w_2 = torch.nn.Linear(hidden_units, idim)
+        self.activation = MultiSpike(hidden_units)
+
+    def forward(self, x, iiter):
+        """Forward function."""
+        x = self.w_1(x)
+        x = self.activation(x)
+        x = self.w_2(x)
+        return x
\ No newline at end of file
diff --git a/src/asr.py b/src/asr.py
new file mode 100644
index 0000000000000000000000000000000000000000..7252ca971710f89449d30ef48c22755f8100f95a
--- /dev/null
+++ b/src/asr.py
@@ -0,0 +1,635 @@
+import argparse
+import logging
+from typing import Callable, Collection, Dict, List, Optional, Tuple
+
+import numpy as np
+import torch
+from typeguard import typechecked
+
+from espnet2.asr.ctc import CTC
+from espnet2.asr.decoder.abs_decoder import AbsDecoder
+from espnet2.asr.decoder.hugging_face_transformers_decoder import (  # noqa: H301
+    HuggingFaceTransformersDecoder,
+)
+from espnet2.asr.decoder.mlm_decoder import MLMDecoder
+from espnet2.asr.decoder.rnn_decoder import RNNDecoder
+from espnet2.asr.decoder.s4_decoder import S4Decoder
+from espnet2.asr.decoder.transducer_decoder import TransducerDecoder
+from espnet2.asr.decoder.transformer_decoder import (
+    DynamicConvolution2DTransformerDecoder,
+    DynamicConvolutionTransformerDecoder,
+    LightweightConvolution2DTransformerDecoder,
+    LightweightConvolutionTransformerDecoder,
+    TransformerDecoder,
+)
+from espnet2.asr.decoder.whisper_decoder import OpenAIWhisperDecoder
+from espnet2.asr.encoder.abs_encoder import AbsEncoder
+from espnet2.asr.encoder.avhubert_encoder import FairseqAVHubertEncoder
+from espnet2.asr.encoder.branchformer_encoder import BranchformerEncoder
+from espnet2.asr.encoder.conformer_encoder import ConformerEncoder
+from espnet2.asr.encoder.contextual_block_conformer_encoder import (
+    ContextualBlockConformerEncoder,
+)
+from espnet2.asr.encoder.contextual_block_transformer_encoder import (
+    ContextualBlockTransformerEncoder,
+)
+from espnet2.asr.encoder.e_branchformer_encoder import EBranchformerEncoder
+from espnet2.asr.encoder.hubert_encoder import (
+    FairseqHubertEncoder,
+    FairseqHubertPretrainEncoder,
+    TorchAudioHuBERTPretrainEncoder,
+)
+from espnet2.asr.encoder.longformer_encoder import LongformerEncoder
+from espnet2.asr.encoder.rnn_encoder import RNNEncoder
+from espnet2.asr.encoder.transformer_encoder import TransformerEncoder
+from espnet2.asr.encoder.Spike_driven.Q_transformer_encoder import Q_TransformerEncoder
+from espnet2.asr.encoder.transformer_encoder_multispkr import (
+    TransformerEncoder as TransformerEncoderMultiSpkr,
+)
+from espnet2.asr.encoder.vgg_rnn_encoder import VGGRNNEncoder
+from espnet2.asr.encoder.wav2vec2_encoder import FairSeqWav2Vec2Encoder
+from espnet2.asr.encoder.whisper_encoder import OpenAIWhisperEncoder
+from espnet2.asr.espnet_model import ESPnetASRModel
+from espnet2.asr.frontend.abs_frontend import AbsFrontend
+from espnet2.asr.frontend.default import DefaultFrontend
+from espnet2.asr.frontend.fused import FusedFrontends
+from espnet2.asr.frontend.s3prl import S3prlFrontend
+from espnet2.asr.frontend.whisper import WhisperFrontend
+from espnet2.asr.frontend.windowing import SlidingWindow
+from espnet2.asr.maskctc_model import MaskCTCModel
+from espnet2.asr.pit_espnet_model import ESPnetASRModel as PITESPnetModel
+from espnet2.asr.postencoder.abs_postencoder import AbsPostEncoder
+from espnet2.asr.postencoder.hugging_face_transformers_postencoder import (
+    HuggingFaceTransformersPostEncoder,
+)
+from espnet2.asr.postencoder.length_adaptor_postencoder import LengthAdaptorPostEncoder
+from espnet2.asr.preencoder.abs_preencoder import AbsPreEncoder
+from espnet2.asr.preencoder.linear import LinearProjection
+from espnet2.asr.preencoder.sinc import LightweightSincConvs
+from espnet2.asr.specaug.abs_specaug import AbsSpecAug
+from espnet2.asr.specaug.specaug import SpecAug
+from espnet2.asr_transducer.joint_network import JointNetwork
+from espnet2.layers.abs_normalize import AbsNormalize
+from espnet2.layers.global_mvn import GlobalMVN
+from espnet2.layers.utterance_mvn import UtteranceMVN
+from espnet2.tasks.abs_task import AbsTask
+from espnet2.text.phoneme_tokenizer import g2p_choices
+from espnet2.torch_utils.initialize import initialize
+from espnet2.train.abs_espnet_model import AbsESPnetModel
+from espnet2.train.class_choices import ClassChoices
+from espnet2.train.collate_fn import CommonCollateFn
+from espnet2.train.preprocessor import (
+    AbsPreprocessor,
+    CommonPreprocessor,
+    CommonPreprocessor_multi,
+)
+from espnet2.train.trainer import Trainer
+from espnet2.utils.get_default_kwargs import get_default_kwargs
+from espnet2.utils.nested_dict_action import NestedDictAction
+from espnet2.utils.types import float_or_none, int_or_none, str2bool, str_or_none
+
+frontend_choices = ClassChoices(
+    name="frontend",
+    classes=dict(
+        default=DefaultFrontend,
+        sliding_window=SlidingWindow,
+        s3prl=S3prlFrontend,
+        fused=FusedFrontends,
+        whisper=WhisperFrontend,
+    ),
+    type_check=AbsFrontend,
+    default="default",
+)
+specaug_choices = ClassChoices(
+    name="specaug",
+    classes=dict(
+        specaug=SpecAug,
+    ),
+    type_check=AbsSpecAug,
+    default=None,
+    optional=True,
+)
+normalize_choices = ClassChoices(
+    "normalize",
+    classes=dict(
+        global_mvn=GlobalMVN,
+        utterance_mvn=UtteranceMVN,
+    ),
+    type_check=AbsNormalize,
+    default="utterance_mvn",
+    optional=True,
+)
+model_choices = ClassChoices(
+    "model",
+    classes=dict(
+        espnet=ESPnetASRModel,
+        maskctc=MaskCTCModel,
+        pit_espnet=PITESPnetModel,
+    ),
+    type_check=AbsESPnetModel,
+    default="espnet",
+)
+preencoder_choices = ClassChoices(
+    name="preencoder",
+    classes=dict(
+        sinc=LightweightSincConvs,
+        linear=LinearProjection,
+    ),
+    type_check=AbsPreEncoder,
+    default=None,
+    optional=True,
+)
+encoder_choices = ClassChoices(
+    "encoder",
+    classes=dict(
+        conformer=ConformerEncoder,
+        transformer=TransformerEncoder,
+        Q_transformer=Q_TransformerEncoder,
+        transformer_multispkr=TransformerEncoderMultiSpkr,
+        contextual_block_transformer=ContextualBlockTransformerEncoder,
+        contextual_block_conformer=ContextualBlockConformerEncoder,
+        vgg_rnn=VGGRNNEncoder,
+        rnn=RNNEncoder,
+        wav2vec2=FairSeqWav2Vec2Encoder,
+        hubert=FairseqHubertEncoder,
+        hubert_pretrain=FairseqHubertPretrainEncoder,
+        torchaudiohubert=TorchAudioHuBERTPretrainEncoder,
+        longformer=LongformerEncoder,
+        branchformer=BranchformerEncoder,
+        whisper=OpenAIWhisperEncoder,
+        e_branchformer=EBranchformerEncoder,
+        avhubert=FairseqAVHubertEncoder,
+    ),
+    type_check=AbsEncoder,
+    default="rnn",
+)
+postencoder_choices = ClassChoices(
+    name="postencoder",
+    classes=dict(
+        hugging_face_transformers=HuggingFaceTransformersPostEncoder,
+        length_adaptor=LengthAdaptorPostEncoder,
+    ),
+    type_check=AbsPostEncoder,
+    default=None,
+    optional=True,
+)
+decoder_choices = ClassChoices(
+    "decoder",
+    classes=dict(
+        transformer=TransformerDecoder,
+        lightweight_conv=LightweightConvolutionTransformerDecoder,
+        lightweight_conv2d=LightweightConvolution2DTransformerDecoder,
+        dynamic_conv=DynamicConvolutionTransformerDecoder,
+        dynamic_conv2d=DynamicConvolution2DTransformerDecoder,
+        rnn=RNNDecoder,
+        transducer=TransducerDecoder,
+        mlm=MLMDecoder,
+        whisper=OpenAIWhisperDecoder,
+        hugging_face_transformers=HuggingFaceTransformersDecoder,
+        s4=S4Decoder,
+    ),
+    type_check=AbsDecoder,
+    default=None,
+    optional=True,
+)
+preprocessor_choices = ClassChoices(
+    "preprocessor",
+    classes=dict(
+        default=CommonPreprocessor,
+        multi=CommonPreprocessor_multi,
+    ),
+    type_check=AbsPreprocessor,
+    default="default",
+)
+
+
+class ASRTask(AbsTask):
+    # If you need more than one optimizers, change this value
+    num_optimizers: int = 1
+
+    # Add variable objects configurations
+    class_choices_list = [
+        # --frontend and --frontend_conf
+        frontend_choices,
+        # --specaug and --specaug_conf
+        specaug_choices,
+        # --normalize and --normalize_conf
+        normalize_choices,
+        # --model and --model_conf
+        model_choices,
+        # --preencoder and --preencoder_conf
+        preencoder_choices,
+        # --encoder and --encoder_conf
+        encoder_choices,
+        # --postencoder and --postencoder_conf
+        postencoder_choices,
+        # --decoder and --decoder_conf
+        decoder_choices,
+        # --preprocessor and --preprocessor_conf
+        preprocessor_choices,
+    ]
+
+    # If you need to modify train() or eval() procedures, change Trainer class here
+    trainer = Trainer
+
+    @classmethod
+    def add_task_arguments(cls, parser: argparse.ArgumentParser):
+        group = parser.add_argument_group(description="Task related")
+
+        # NOTE(kamo): add_arguments(..., required=True) can't be used
+        # to provide --print_config mode. Instead of it, do as
+        required = parser.get_default("required")
+        required += ["token_list"]
+
+        group.add_argument(
+            "--token_list",
+            type=str_or_none,
+            default=None,
+            help="A text mapping int-id to token",
+        )
+        group.add_argument(
+            "--init",
+            type=lambda x: str_or_none(x.lower()),
+            default=None,
+            help="The initialization method",
+            choices=[
+                "chainer",
+                "xavier_uniform",
+                "xavier_normal",
+                "kaiming_uniform",
+                "kaiming_normal",
+                None,
+            ],
+        )
+
+        group.add_argument(
+            "--input_size",
+            type=int_or_none,
+            default=None,
+            help="The number of input dimension of the feature",
+        )
+
+        group.add_argument(
+            "--ctc_conf",
+            action=NestedDictAction,
+            default=get_default_kwargs(CTC),
+            help="The keyword arguments for CTC class.",
+        )
+        group.add_argument(
+            "--joint_net_conf",
+            action=NestedDictAction,
+            default=None,
+            help="The keyword arguments for joint network class.",
+        )
+
+        group = parser.add_argument_group(description="Preprocess related")
+        group.add_argument(
+            "--use_preprocessor",
+            type=str2bool,
+            default=True,
+            help="Apply preprocessing to data or not",
+        )
+        group.add_argument(
+            "--use_lang_prompt",
+            type=str2bool,
+            default=False,
+            help="Use language id as prompt",
+        )
+        group.add_argument(
+            "--use_nlp_prompt",
+            type=str2bool,
+            default=False,
+            help="Use natural language phrases as prompt",
+        )
+        group.add_argument(
+            "--token_type",
+            type=str,
+            default="bpe",
+            choices=[
+                "bpe",
+                "char",
+                "word",
+                "phn",
+                "hugging_face",
+                "whisper_en",
+                "whisper_multilingual",
+            ],
+            help="The text will be tokenized " "in the specified level token",
+        )
+        group.add_argument(
+            "--bpemodel",
+            type=str_or_none,
+            default=None,
+            help="The model file of sentencepiece",
+        )
+        parser.add_argument(
+            "--non_linguistic_symbols",
+            type=str_or_none,
+            help="non_linguistic_symbols file path",
+        )
+        group.add_argument(
+            "--cleaner",
+            type=str_or_none,
+            choices=[
+                None,
+                "tacotron",
+                "jaconv",
+                "vietnamese",
+                "whisper_en",
+                "whisper_basic",
+            ],
+            default=None,
+            help="Apply text cleaning",
+        )
+        group.add_argument(
+            "--g2p",
+            type=str_or_none,
+            choices=g2p_choices,
+            default=None,
+            help="Specify g2p method if --token_type=phn",
+        )
+        group.add_argument(
+            "--speech_volume_normalize",
+            type=float_or_none,
+            default=None,
+            help="Scale the maximum amplitude to the given value.",
+        )
+        group.add_argument(
+            "--rir_scp",
+            type=str_or_none,
+            default=None,
+            help="The file path of rir scp file.",
+        )
+        group.add_argument(
+            "--rir_apply_prob",
+            type=float,
+            default=1.0,
+            help="THe probability for applying RIR convolution.",
+        )
+        group.add_argument(
+            "--noise_scp",
+            type=str_or_none,
+            default=None,
+            help="The file path of noise scp file.",
+        )
+        group.add_argument(
+            "--noise_apply_prob",
+            type=float,
+            default=1.0,
+            help="The probability applying Noise adding.",
+        )
+        group.add_argument(
+            "--noise_db_range",
+            type=str,
+            default="13_15",
+            help="The range of noise decibel level.",
+        )
+        group.add_argument(
+            "--short_noise_thres",
+            type=float,
+            default=0.5,
+            help="If len(noise) / len(speech) is smaller than this threshold during "
+            "dynamic mixing, a warning will be displayed.",
+        )
+        group.add_argument(
+            "--aux_ctc_tasks",
+            type=str,
+            nargs="+",
+            default=[],
+            help="Auxillary tasks to train on using CTC loss. ",
+        )
+
+        for class_choices in cls.class_choices_list:
+            # Append --<name> and --<name>_conf.
+            # e.g. --encoder and --encoder_conf
+            class_choices.add_arguments(group)
+
+    @classmethod
+    @typechecked
+    def build_collate_fn(cls, args: argparse.Namespace, train: bool) -> Callable[
+        [Collection[Tuple[str, Dict[str, np.ndarray]]]],
+        Tuple[List[str], Dict[str, torch.Tensor]],
+    ]:
+        # NOTE(kamo): int value = 0 is reserved by CTC-blank symbol
+        return CommonCollateFn(float_pad_value=0.0, int_pad_value=-1)
+
+    @classmethod
+    @typechecked
+    def build_preprocess_fn(
+        cls, args: argparse.Namespace, train: bool
+    ) -> Optional[Callable[[str, Dict[str, np.array]], Dict[str, np.ndarray]]]:
+        if args.use_preprocessor:
+            try:
+                _ = getattr(args, "preprocessor")
+            except AttributeError:
+                setattr(args, "preprocessor", "default")
+                setattr(args, "preprocessor_conf", dict())
+            except Exception as e:
+                raise e
+
+            preprocessor_class = preprocessor_choices.get_class(args.preprocessor)
+            retval = preprocessor_class(
+                train=train,
+                token_type=args.token_type,
+                token_list=args.token_list,
+                bpemodel=args.bpemodel,
+                non_linguistic_symbols=args.non_linguistic_symbols,
+                text_cleaner=args.cleaner,
+                g2p_type=args.g2p,
+                # NOTE(kamo): Check attribute existence for backward compatibility
+                rir_scp=args.rir_scp if hasattr(args, "rir_scp") else None,
+                rir_apply_prob=(
+                    args.rir_apply_prob if hasattr(args, "rir_apply_prob") else 1.0
+                ),
+                noise_scp=args.noise_scp if hasattr(args, "noise_scp") else None,
+                noise_apply_prob=(
+                    args.noise_apply_prob if hasattr(args, "noise_apply_prob") else 1.0
+                ),
+                noise_db_range=(
+                    args.noise_db_range if hasattr(args, "noise_db_range") else "13_15"
+                ),
+                short_noise_thres=(
+                    args.short_noise_thres
+                    if hasattr(args, "short_noise_thres")
+                    else 0.5
+                ),
+                speech_volume_normalize=(
+                    args.speech_volume_normalize if hasattr(args, "rir_scp") else None
+                ),
+                aux_task_names=(
+                    args.aux_ctc_tasks if hasattr(args, "aux_ctc_tasks") else None
+                ),
+                use_lang_prompt=(
+                    args.use_lang_prompt if hasattr(args, "use_lang_prompt") else None
+                ),
+                **args.preprocessor_conf,
+                use_nlp_prompt=(
+                    args.use_nlp_prompt if hasattr(args, "use_nlp_prompt") else None
+                ),
+            )
+        else:
+            retval = None
+        return retval
+
+    @classmethod
+    def required_data_names(
+        cls, train: bool = True, inference: bool = False
+    ) -> Tuple[str, ...]:
+        if not inference:
+            retval = ("speech", "text")
+        else:
+            # Recognition mode
+            retval = ("speech",)
+        return retval
+
+    @classmethod
+    def optional_data_names(
+        cls, train: bool = True, inference: bool = False
+    ) -> Tuple[str, ...]:
+        MAX_REFERENCE_NUM = 4
+
+        retval = ["text_spk{}".format(n) for n in range(2, MAX_REFERENCE_NUM + 1)]
+        retval = retval + ["prompt"]
+        retval = tuple(retval)
+
+        logging.info(f"Optional Data Names: {retval }")
+        return retval
+
+    @classmethod
+    @typechecked
+    def build_model(cls, args: argparse.Namespace) -> ESPnetASRModel:
+        if isinstance(args.token_list, str):
+            with open(args.token_list, encoding="utf-8") as f:
+                token_list = [line.rstrip() for line in f]
+
+            # Overwriting token_list to keep it as "portable".
+            args.token_list = list(token_list)
+        elif isinstance(args.token_list, (tuple, list)):
+            token_list = list(args.token_list)
+        else:
+            raise RuntimeError("token_list must be str or list")
+
+        # If use multi-blank transducer criterion,
+        # big blank symbols are added just before the standard blank
+        if args.model_conf.get("transducer_multi_blank_durations", None) is not None:
+            sym_blank = args.model_conf.get("sym_blank", "<blank>")
+            blank_idx = token_list.index(sym_blank)
+            for dur in args.model_conf.get("transducer_multi_blank_durations"):
+                if f"<blank{dur}>" not in token_list:  # avoid this during inference
+                    token_list.insert(blank_idx, f"<blank{dur}>")
+            args.token_list = token_list
+
+        vocab_size = len(token_list)
+        logging.info(f"Vocabulary size: {vocab_size }")
+
+        # 1. frontend
+        if args.input_size is None:
+            # Extract features in the model
+            frontend_class = frontend_choices.get_class(args.frontend)
+            frontend = frontend_class(**args.frontend_conf)
+            input_size = frontend.output_size()
+        else:
+            # Give features from data-loader
+            args.frontend = None
+            args.frontend_conf = {}
+            frontend = None
+            input_size = args.input_size
+
+        # 2. Data augmentation for spectrogram
+        if args.specaug is not None:
+            specaug_class = specaug_choices.get_class(args.specaug)
+            specaug = specaug_class(**args.specaug_conf)
+        else:
+            specaug = None
+
+        # 3. Normalization layer
+        if args.normalize is not None:
+            normalize_class = normalize_choices.get_class(args.normalize)
+            normalize = normalize_class(**args.normalize_conf)
+        else:
+            normalize = None
+
+        # 4. Pre-encoder input block
+        # NOTE(kan-bayashi): Use getattr to keep the compatibility
+        if getattr(args, "preencoder", None) is not None:
+            preencoder_class = preencoder_choices.get_class(args.preencoder)
+            preencoder = preencoder_class(**args.preencoder_conf)
+            input_size = preencoder.output_size()
+        else:
+            preencoder = None
+
+        # 4. Encoder
+        encoder_class = encoder_choices.get_class(args.encoder)
+        encoder = encoder_class(input_size=input_size, **args.encoder_conf)
+
+        # 5. Post-encoder block
+        # NOTE(kan-bayashi): Use getattr to keep the compatibility
+        encoder_output_size = encoder.output_size()
+        if getattr(args, "postencoder", None) is not None:
+            postencoder_class = postencoder_choices.get_class(args.postencoder)
+            postencoder = postencoder_class(
+                input_size=encoder_output_size, **args.postencoder_conf
+            )
+            encoder_output_size = postencoder.output_size()
+        else:
+            postencoder = None
+
+        # 5. Decoder
+        if getattr(args, "decoder", None) is not None:
+            decoder_class = decoder_choices.get_class(args.decoder)
+
+            if args.decoder == "transducer":
+                decoder = decoder_class(
+                    vocab_size,
+                    embed_pad=0,
+                    **args.decoder_conf,
+                )
+
+                joint_network = JointNetwork(
+                    vocab_size,
+                    encoder.output_size(),
+                    decoder.dunits,
+                    **args.joint_net_conf,
+                )
+            else:
+                decoder = decoder_class(
+                    vocab_size=vocab_size,
+                    encoder_output_size=encoder_output_size,
+                    **args.decoder_conf,
+                )
+                joint_network = None
+        else:
+            decoder = None
+            joint_network = None
+
+        # 6. CTC
+        ctc = CTC(
+            odim=vocab_size, encoder_output_size=encoder_output_size, **args.ctc_conf
+        )
+
+        # 7. Build model
+        try:
+            model_class = model_choices.get_class(args.model)
+        except AttributeError:
+            model_class = model_choices.get_class("espnet")
+        model = model_class(
+            vocab_size=vocab_size,
+            frontend=frontend,
+            specaug=specaug,
+            normalize=normalize,
+            preencoder=preencoder,
+            encoder=encoder,
+            postencoder=postencoder,
+            decoder=decoder,
+            ctc=ctc,
+            joint_network=joint_network,
+            token_list=token_list,
+            **args.model_conf,
+        )
+
+        # FIXME(kamo): Should be done in model?
+        # 8. Initialize
+        if args.init is not None:
+            initialize(model, args.init)
+
+        return model
diff --git a/src/conf/Librispeech/train_asr_Q_transformer3_HierDecayv2.yaml b/src/conf/Librispeech/train_asr_Q_transformer3_HierDecayv2.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..91c5c2aef5a6161d550e79ec4aa870bedb468517
--- /dev/null
+++ b/src/conf/Librispeech/train_asr_Q_transformer3_HierDecayv2.yaml
@@ -0,0 +1,64 @@
+batch_type: numel
+batch_bins: 45000000
+accum_grad: 2
+max_epoch: 100
+patience: none
+# The initialization method for model parameters
+init: xavier_uniform
+best_model_criterion:
+-   - valid
+    - acc
+    - max
+keep_nbest_models: 10
+
+encoder: Q_transformer
+encoder_conf:
+    output_size: 512
+    attention_heads: 8
+    attention_layer_type: HierDecayv2
+    linear_units: 2048
+    num_blocks: 18
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.1
+    input_layer: conv2d6
+    normalize_before: true
+
+decoder: transformer
+decoder_conf:
+    attention_heads: 8
+    linear_units: 2048
+    num_blocks: 6
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.1
+    src_attention_dropout_rate: 0.1
+
+model_conf:
+    ctc_weight: 0.3
+    lsm_weight: 0.1
+    length_normalized_loss: false
+
+use_amp: true
+optim: adam
+optim_conf:
+    lr: 0.002
+scheduler: warmuplr
+scheduler_conf:
+    warmup_steps: 25000
+
+specaug: specaug
+specaug_conf:
+    apply_time_warp: true
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    num_freq_mask: 2
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 40
+    num_time_mask: 2
diff --git a/src/run.sh b/src/run.sh
new file mode 100644
index 0000000000000000000000000000000000000000..80620b3155bc72799bf6b058209fc6cc2a569102
--- /dev/null
+++ b/src/run.sh
@@ -0,0 +1,42 @@
+#!/usr/bin/env bash
+# Set bash to 'debug' mode, it will exit on :
+# -e 'error', -u 'undefined variable', -o ... 'error in pipeline', -x 'print commands',
+set -e
+set -u
+set -o pipefail
+
+train_set="train_960"
+valid_set="dev"
+test_sets="dev_clean"
+
+asr_config=conf/tuning/SNN/train_asr_Q_transformer3_HierDecayv2.yaml
+lm_config=conf/tuning/train_lm_transformer2.yaml
+inference_config=conf/decode_asr.yaml
+
+asr_tag=transformer_HierDecayv2_woBias_XavierInit_Qv7_test
+
+use_lm=false
+skip_data_prep=true
+skip_train=true
+
+./asr.sh \
+    --skip_data_prep ${skip_data_prep}\
+    --skip_train ${skip_train}\
+    --stage 12 \
+    --stop_stage 12 \
+    --lang en \
+    --ngpu 1 \
+    --use_lm ${use_lm} \
+    --nbpe 5000 \
+    --max_wav_duration 30 \
+    --asr_tag "${asr_tag}" \
+    --gpu_inference false \
+    --speed_perturb_factors "0.9 1.0 1.1" \
+    --asr_config "${asr_config}" \
+    --lm_config "${lm_config}" \
+    --inference_config "${inference_config}" \
+    --train_set "${train_set}" \
+    --valid_set "${valid_set}" \
+    --test_sets "${test_sets}" \
+    --lm_train_text "data/${train_set}/text data/local/other_text/text" \
+    --bpe_train_text "data/${train_set}/text" "$@"