Starting from 512b for 10epochs

config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "./",
+  "architectures": [
+    "RobertaForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "transformers_version": "4.16.0.dev0",
+  "type_vocab_size": 1,
+  "use_cache": true,
+  "vocab_size": 50265
+}

eval_results.json

@@ -0,0 +1,5 @@
+{
+    "eval_accuracy": 0.732703958770086,
+    "eval_loss": 1.2010057153335871,
+    "eval_perplexity": 3.3234576948863452
+}

flax_model.msgpack

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b973b8f46e005ee02f6ba580dbba6b3800f61749eae1eea4a455000847dc0578
+size 498796983

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "sep_token": "</s>", "pad_token": "<pad>", "cls_token": "<s>", "mask_token": {"content": "<mask>", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "add_prefix_space": false, "errors": "replace", "sep_token": "</s>", "cls_token": "<s>", "pad_token": "<pad>", "mask_token": "<mask>", "trim_offsets": true, "special_tokens_map_file": null, "name_or_path": "./", "tokenizer_class": "RobertaTokenizer"}

train.128.sh

@@ -0,0 +1,25 @@
+python run_mlm_flax.py \
+    --output_dir="./" \
+    --model_type="roberta" \
+    --config_name="roberta-base" \
+    --tokenizer_name="NbAiLab/nb-roberta-base" \
+    --dataset_name="NbAiLab/NCC" \
+    --max_seq_length="128" \
+    --weight_decay="0.01" \
+    --per_device_train_batch_size="232" \
+    --per_device_eval_batch_size="232" \
+    --pad_to_max_length \
+    --learning_rate="6e-4" \
+    --warmup_steps="10000" \
+    --overwrite_output_dir \
+    --num_train_epochs="3" \
+    --adam_beta1="0.9" \
+    --adam_beta2="0.98" \
+    --adam_epsilon="1e-6" \
+    --logging_steps="1000" \
+    --save_steps="1000" \
+    --eval_steps="1000" \
+    --do_train \
+    --do_eval \
+    --dtype="bfloat16" \
+    --push_to_hub

train.512.sh

@@ -0,0 +1,26 @@
+python run_mlm_flax.py \
+    --output_dir="./" \
+    --model_type="roberta" \
+    --model_name_or_path="./" \
+    --config_name="./" \
+    --tokenizer_name="./" \
+    --dataset_name="NbAiLab/NCC" \
+    --max_seq_length="512" \
+    --weight_decay="0.01" \
+    --per_device_train_batch_size="46" \
+    --per_device_eval_batch_size="46" \
+    --pad_to_max_length \
+    --learning_rate="0.00015" \
+    --warmup_steps="1000" \
+    --overwrite_output_dir \
+    --num_train_epochs="10" \
+    --adam_beta1="0.9" \
+    --adam_beta2="0.98" \
+    --adam_epsilon="1e-6" \
+    --logging_steps="1000" \
+    --save_steps="1000" \
+    --eval_steps="1000" \
+    --do_train \
+    --do_eval \
+    --dtype="bfloat16" \
+    --push_to_hub