Skip to content

Commit

Permalink
Fixing configs, adding two more for rerun on base and large robertas
Browse files Browse the repository at this point in the history
  • Loading branch information
dchaplinsky committed Oct 3, 2023
1 parent dbed625 commit a2b46e4
Show file tree
Hide file tree
Showing 3 changed files with 68 additions and 4 deletions.
Original file line number Diff line number Diff line change
@@ -0,0 +1,32 @@
{
"output_dir": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1.rerun1",
"model_type": "roberta",
"config_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1.rerun1",
"tokenizer_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1.rerun1",
"train_file": "data/malyuk.filter",
"max_seq_length": 512,
"do_train": true,
"do_eval": true,
"validation_file": "data/bruk_valid_data.txt",
"weight_decay": 1e-1,
"per_device_train_batch_size": 4,
"per_device_eval_batch_size": 4,
"gradient_accumulation_steps": 1,
"learning_rate": 1e-5,
"evaluation_strategy": "steps",
"warmup_steps": 9000,
"max_steps": 400000,
"eval_steps": 500,
"save_steps": 50000,
"logging_steps": 500,
"overwrite_output_dir": true,
"num_train_epochs": 0,
"adam_beta1": 0.965,
"adam_beta2": 0.99,
"adam_epsilon": 1e-6,
"preprocessing_num_workers": 96,
"model_name_or_path": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1.rerun1",
"bf16": true,
"optimizer": "sophia",
"lr_scheduler_type": "linear"
}
Original file line number Diff line number Diff line change
@@ -1,8 +1,8 @@
{
"output_dir": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1",
"output_dir": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun2",
"model_type": "roberta",
"config_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1",
"tokenizer_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1",
"config_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun2",
"tokenizer_name": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun2",
"train_file": "data/malyuk.filter",
"max_seq_length": 512,
"do_train": true,
Expand All @@ -25,7 +25,7 @@
"adam_beta2": 0.99,
"adam_epsilon": 1e-6,
"preprocessing_num_workers": 96,
"model_name_or_path": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun1",
"model_name_or_path": "exps/roberta.large.malyuk.filter.wechsel.largedict.tpu.run1.rerun2",
"bf16": true,
"optimizer": "sophia",
"lr_scheduler_type": "linear"
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,32 @@
{
"output_dir": "exps/roberta.malyuk.filter.wechsel.largedict.tpu.run7.rerun3.rerun1",
"model_type": "roberta",
"config_name": "exps/roberta.malyuk.filter.wechsel.largedict.tpu.run7.rerun3.rerun1",
"tokenizer_name": "exps/roberta.malyuk.filter.wechsel.largedict.tpu.run7.rerun3.rerun1",
"train_file": "data/malyuk.filter",
"max_seq_length": 512,
"do_train": true,
"do_eval": true,
"validation_file": "data/bruk_valid_data.txt",
"weight_decay": 1e-1,
"per_device_train_batch_size": 38,
"per_device_eval_batch_size": 38,
"gradient_accumulation_steps": 1,
"learning_rate": 1e-5,
"evaluation_strategy": "steps",
"warmup_steps": 9000,
"max_steps": 400000,
"eval_steps": 500,
"save_steps": 50000,
"logging_steps": 500,
"overwrite_output_dir": true,
"num_train_epochs": 0,
"adam_beta1": 0.965,
"adam_beta2": 0.99,
"adam_epsilon": 1e-6,
"preprocessing_num_workers": 96,
"model_name_or_path": "exps/roberta.malyuk.filter.wechsel.largedict.tpu.run7.rerun3.rerun1",
"bf16": true,
"optimizer": "sophia",
"lr_scheduler_type": "linear"
}

0 comments on commit a2b46e4

Please sign in to comment.