remove old exp&model and lower steps to save

fix script

make it resumable training

Saving weights and logs of step 10000

update event

update tfevent

update event

remove old event

add experiment_2 tfevents

update script

update script regarding to https://github.com/huggingface/transformers/pull/12608

lower lr regarding explode gradient. also new seed number

Merge branch 'main' of https://huggingface.co/flax-community/roberta-base-thai into main

update tfevents

Update .gitattributes

update event

update event

update event

update tfevents

update tfevents

update tfevents

add tfevents

switch back to batch_size = 32

change saving and eval steps

probe tfevents

adjust batch_size and update run_mlm_flax script

check tfevents

adjust running script

reduce batch size

change to small batch size

update tfevent

change dtype to bfloat16

probe first tfevents

adjust script

update setting and batch size

fix run_mlm_flax script

change batch size and command

add runing script

add pre-generated config.json and trained Thai tokenizer.json

initial commit