gpt-code-clippy/run_clm_streaming_filter.sh
arampacha 92e00c6855
Dataloaders (#55)
* adds online data filtering

* Add new lang extensions to filter by

* Switch script to use filtering

* Add C++ headers also `lang_exts` should be a set

* script for filtered run

Co-authored-by: ncoop57 <nacooper01@email.wm.edu>
Co-authored-by: Santiago Hincapie <shpotes3@gmail.com>
2021-07-17 03:07:39 +03:00

38 lines
1.2 KiB
Bash

#! /bin/bash
./run_clm_streaming_filter_flax.py \
--output_dir $HOME/gpt-code-clippy-lr1e-4-bs1024-f \
--model_name_or_path EleutherAI/gpt-neo-125M \
--dataset_name $HOME/gpt-code-clippy/code_clippy_filter.py \
--data_dir /home/shared/code_clippy_dedup_data \
--text_column_name="text" \
--do_train --do_eval \
--block_size="2048" \
--per_device_train_batch_size="8" \
--per_device_eval_batch_size="16" \
--preprocessing_num_workers="8" \
--learning_rate="1e-4" \
--max_steps 100000 \
--warmup_steps 4000 \
--decay_steps 50000 \
--adam_beta1="0.9" \
--adam_beta2="0.95" \
--weight_decay="0.1" \
--overwrite_output_dir \
--logging_steps="50" \
--eval_steps="1000" \
--push_to_hub="True" \
--report_to="all" \
--dtype="bfloat16" \
--skip_memory_metrics="False" \
--save_steps="1000" \
--save_total_limit 5 \
--gradient_accumulation_steps 16 \
--report_to="wandb" \
--run_name="testing" \
--max_eval_samples 2000 \
--save_optimizer true \
# --adafactor \
# --resume_from_checkpoint $HOME/gpt-neo-125M-code-clippy/ \
# --max_train_samples="10000" \