T5JEP

config.json

@@ -0,0 +1,31 @@
+{
+  "_name_or_path": "t5je",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 2048,
+  "d_kv": 64,
+  "d_model": 768,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "gelu_new",
+  "dropout_rate": 0.1,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-gelu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 12,
+  "num_heads": 12,
+  "num_layers": 12,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 32,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.26.1",
+  "use_cache": true,
+  "vocab_size": 32128
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "decoder_start_token_id": 0,
+  "eos_token_id": 1,
+  "pad_token_id": 0,
+  "transformers_version": "4.26.1"
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88b0a6108a40a82b8eb3fe375da56393b26705c85ce5291ce880c6c9ba61141e
+size 990404917

special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "eos_token": "</s>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:108ea5dbb232558d744aff5011d29b92a76751c210ad8560e6a65738c9630bdf
+size 775057

tokenizer_config.json

@@ -0,0 +1,12 @@
+{
+  "additional_special_tokens": [],
+  "eos_token": "</s>",
+  "extra_ids": 0,
+  "model_max_length": 1000000000000000019884624838656,
+  "name_or_path": "t5je",
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "special_tokens_map_file": "t5snoje/special_tokens_map.json",
+  "tokenizer_class": "T5Tokenizer",
+  "unk_token": "<unk>"
+}

train_log.txt

@@ -0,0 +1,41 @@
+[script] /home/acc12952oa/cuda11_6/work/bin/t5train --model_path=t5je --output_path=t5je_mlm2 --max_hours=12 --batch_size=1024 --solver=adafactor --float32_matmul_precision=medium mlm.jsonl.gz
+[hparams] Namespace(adam_epsilon=1e-08, batch_size=1024, checkpoint_path=None, early_stopping=False, fast_dev_run=False, files=['mlm.jsonl.gz'], float32_matmul_precision='medium', gradient_accumulation_steps=1, learning_rate=0.0003, max_epochs=10, max_grad_norm=1.0, max_hours=12.0, max_length=128, model_path='t5je', n_gpus=1, num_workers=4, output_path='t5je_mlm2', seed=42, solver='adafactor', source_max_length=128, step_batch_size=0, target_max_length=128, tokenizer_path='t5je', warmup_steps=1, weight_decay=0.0)
+[train] ['mlm.jsonl.gz']
+val epoch=1 loss=4.90121 PPL=134.45273
+train epoch=1 loss=4.86353 PPL=129.48067
+val epoch=1 loss=3.63718 PPL=37.98471
+train epoch=1 loss=4.40872 PPL=82.16438
+val epoch=1 loss=3.36813 PPL=29.02423
+train epoch=1 loss=3.56357 PPL=35.28907
+val epoch=1 loss=3.09257 PPL=22.03372
+train epoch=1 loss=3.43388 PPL=30.99654
+val epoch=1 loss=2.74062 PPL=15.49665
+train epoch=1 loss=3.79245 PPL=44.36494
+val epoch=1 loss=2.70695 PPL=14.98345
+train epoch=1 loss=3.07763 PPL=21.70693
+val epoch=1 loss=2.44397 PPL=11.51874
+[auto_batch_size] 64
+[batch_size] 1024
+[accumulate_grad_batches] 16
+[max_time] {'days': 0, 'hours': 12, 'minutes': 0}
+val epoch=1 loss=5.18878 PPL=179.25021
+val epoch=1 loss=0.77304 PPL=2.16635
+train epoch=1 loss=0.08330 PPL=1.08687
+val epoch=2 loss=0.63744 PPL=1.89163
+train epoch=2 loss=0.04870 PPL=1.04990
+val epoch=3 loss=0.56578 PPL=1.76082
+train epoch=3 loss=0.04121 PPL=1.04207
+val epoch=4 loss=0.51813 PPL=1.67889
+train epoch=4 loss=0.03711 PPL=1.03780
+val epoch=5 loss=0.48163 PPL=1.61872
+train epoch=5 loss=0.03429 PPL=1.03489
+val epoch=6 loss=0.44883 PPL=1.56647
+train epoch=6 loss=0.03223 PPL=1.03276
+val epoch=7 loss=0.42527 PPL=1.53001
+train epoch=7 loss=0.03055 PPL=1.03102
+val epoch=8 loss=0.40439 PPL=1.49838
+train epoch=8 loss=0.02917 PPL=1.02960
+val epoch=9 loss=0.38409 PPL=1.46827
+train epoch=9 loss=0.02796 PPL=1.02835
+val epoch=10 loss=0.37744 PPL=1.45855
+train epoch=10 loss=0.02662 PPL=1.02698