mirror of
https://github.com/hanzoai/open-instruct.git
synced 2026-11-03 02:13:07 +00:00
35 lines
1.1 KiB
YAML
35 lines
1.1 KiB
YAML
model_name_or_path: meta-llama/Meta-Llama-3-8B
|
|
model_revision: main
|
|
use_flash_attn: true
|
|
tokenizer_name: meta-llama/Meta-Llama-3-8B
|
|
use_slow_tokenizer: true
|
|
dataset_mixer:
|
|
# Tulu V2 datasets
|
|
allenai/tulu-v2-sft-mixture: 326154
|
|
# Tulu V3 datasets (WIP)
|
|
HuggingFaceH4/no_robots: 9500 # all
|
|
ai2-adapt-dev/metamath-qa-reformat: 100000
|
|
ai2-adapt-dev/codefeedback-single-turn-reformat: 156526 # all
|
|
nvidia/Daring-Anteater: 99532 # all
|
|
# tbd:
|
|
# AI-MO/NuminaMath-TIR: 72441 # all
|
|
# ai2-adapt-dev/slim-orca-300k: 100000
|
|
max_seq_length: 4096 # Note, reduced from 8192 to fit on one GPU with DeepSpeed Stage3
|
|
preprocessing_num_workers: 128
|
|
per_device_train_batch_size: 1 # note, this is set up for 8 GPUs
|
|
gradient_accumulation_steps: 4 # effective batch size 128 with 4 nodes
|
|
learning_rate: 5.0e-06 # best LR so far
|
|
lr_scheduler_type: linear
|
|
warmup_ratio: 0.03
|
|
weight_decay: 0.0
|
|
num_train_epochs: 2
|
|
output_dir: /output/
|
|
with_tracking: true
|
|
report_to:
|
|
- wandb
|
|
logging_steps: 1
|
|
checkpointing_steps: epoch
|
|
dataset_mix_dir: /output/
|
|
checkpointing_steps: 1000
|
|
keep_last_n_checkpoints: 1
|