File size: 1,433 Bytes
d9f57c5 |
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 53 54 55 56 57 |
model:
_component_: torchtune.models.llama2.qlora_llama2_7b
lora_attn_modules:
- q_proj
- v_proj
- k_proj
- output_proj
apply_lora_to_mlp: true
apply_lora_to_output: false
lora_rank: 8
lora_alpha: 16
tokenizer:
_component_: torchtune.models.llama2.llama2_tokenizer
path: ./checkpoint/tokenizer.model
checkpointer:
_component_: torchtune.utils.FullModelHFCheckpointer
checkpoint_dir: ./checkpoint
checkpoint_files:
- pytorch_model-00001-of-00002.bin
- pytorch_model-00002-of-00002.bin
adapter_checkpoint: null
recipe_checkpoint: null
output_dir: ./checkpoint
model_type: LLAMA2
resume_from_checkpoint: false
dataset:
_component_: torchtune.datasets.alpaca_cleaned_dataset
train_on_input: true
seed: null
shuffle: true
batch_size: 2
optimizer:
_component_: torch.optim.AdamW
weight_decay: 0.01
lr: 0.0003
lr_scheduler:
_component_: torchtune.modules.get_cosine_schedule_with_warmup
num_warmup_steps: 100
loss:
_component_: torch.nn.CrossEntropyLoss
epochs: 1
max_steps_per_epoch: null
gradient_accumulation_steps: 16
compile: false
output_dir: /tmp/qlora_finetune_output/
metric_logger:
_component_: torchtune.utils.metric_logging.WandBLogger
log_dir: ${output_dir}
log_every_n_steps: 1
device: cuda
dtype: bf16
enable_activation_checkpointing: true
profiler:
_component_: torchtune.utils.profiler
enabled: false
output_dir: ${output_dir}/torchtune_perf_tracing.json
|