File size: 1,128 Bytes
881ca1d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
alpha: 0.2
base_model: meta-llama/Llama-3.2-1B-Instruct
custom_name: d4-a0.2-v4
dtype: bfloat16
lambdas:
- 1.0
- 1.0
- 1.0
lora_config: null
loss_types:
- anti-watermark
- anti-watermark
meta_learning: false
meta_learning_config: null
n_wm_tokens: 0
proportions:
- 0.5
- 0.2
- 0.3
regularization_datasets:
- !!python/object/apply:finetuning.dataset.DatasetType
- AlpacaGPT4
- !!python/object/apply:finetuning.dataset.DatasetType
- WO_NoHealth
sequence_length: 512
streaming: true
training_args:
bf16: false
do_train: true
fp16: false
gradient_accumulation_steps: 16
gradient_checkpointing: false
hub_strategy: all_checkpoints
learning_rate: 2.0e-05
logging_steps: 10
lr_scheduler_type: cosine
max_steps: 2500
num_train_epochs: 1
optim: adafactor
output_dir: Grogros/dmWM-llama-3.2-1B-Instruct-WOHealth-Al4-NH-WO-d4-a0.2-v4
overwrite_output_dir: true
per_device_train_batch_size: 4
push_to_hub: true
report_to: none
save_steps: 500
save_strategy: steps
warmup_ratio: 0.1
watermark_datasets:
- !!python/object/apply:finetuning.dataset.DatasetType
- WOHealth
watermark_eval_config: []
|