pere commited on
Commit
0c8ac33
·
1 Parent(s): a05c270

Saving weights and logs of epoch 6

Browse files
config.json CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a1fa5adbbe6a5cbeedcab79760dd7611a074c1905f4f79d8d386f835eee730a9
3
- size 701
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba2f2141e020e11d5a8378398ab356436905f898315d8ffe70b26dc91addc441
3
+ size 700
events.out.tfevents.1630919923.t1v-n-1a0a7c50-w-0.3421924.3.v2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78b1af55372d9d5e839b4800985716d13bd89ce660e256adc83a1d5d0d898750
3
+ size 40
events.out.tfevents.1630921213.t1v-n-1a0a7c50-w-0.3431412.3.v2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:442420b6e0a7477c4764844291689e084536f4a835cb79e27fea6333ee3931cb
3
+ size 2865513
flax_model.msgpack CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5fe3762aac25de79ac89d75581bfcbc3bfd701c3f616afa4b7341e97c6414497
3
- size 1100762015
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5167e3435eacd785c79013187f090ebc1538ea5a04d3f649692cd929c470268c
3
+ size 2329617315
run_summarization_t5.sh CHANGED
@@ -1,18 +1,20 @@
1
 
2
  python run_summarization_t5_flax.py \
3
  --output_dir "." \
4
- --model_name_or_path pere/norwegian-t5-base-NCC-fast \
5
- --tokenizer_name pere/norwegian-t5-base-NCC-fast \
6
  --dataset_name xsum \
7
  --do_train --do_eval --do_predict \
8
  --predict_with_generate \
9
  --num_train_epochs 6 \
10
  --learning_rate 5e-5 \
11
  --warmup_steps 0 \
12
- --per_device_train_batch_size 32 \
13
- --per_device_eval_batch_size 32 \
14
  --overwrite_output_dir \
15
  --max_source_length 512 \
16
  --max_target_length 64 \
17
- --revision $1 \
18
  --push_to_hub
 
 
 
 
1
 
2
  python run_summarization_t5_flax.py \
3
  --output_dir "." \
4
+ --model_name_or_path google/mt5-base \
5
+ --tokenizer_name google/mt5-base \
6
  --dataset_name xsum \
7
  --do_train --do_eval --do_predict \
8
  --predict_with_generate \
9
  --num_train_epochs 6 \
10
  --learning_rate 5e-5 \
11
  --warmup_steps 0 \
12
+ --per_device_train_batch_size 8 \
13
+ --per_device_eval_batch_size 8 \
14
  --overwrite_output_dir \
15
  --max_source_length 512 \
16
  --max_target_length 64 \
 
17
  --push_to_hub
18
+
19
+
20
+ #--revision None \
run_summarization_t5_flax.py CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4e2782269c57143454f219ab7563eaed06ab04782ddb7c7d50906b1f1302b4b3
3
- size 34784
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7bb91fa2da0e303cda754fe06548c3710f06dcddf575795da18eca4e67976d5
3
+ size 35053