File size: 2,778 Bytes
f16186b |
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 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 |
# Config for multi-device full finetuning in full_finetune_distributed.py
# using a Llama3 8B Instruct model
#
# This config assumes that you've run the following command before launching
# this run:
# tune download meta-llama/Meta-Llama-3-8B-Instruct --output-dir /tmp/Meta-Llama-3-8B-Instruct --hf-token <HF_TOKEN>
#
# To launch on 4 devices, run the following command from root:
# tune run --nproc_per_node 4 full_finetune_distributed --config llama3/8B_full
#
# You can add specific overrides through the command line. For example
# to override the checkpointer directory while launching training
# you can run:
# tune run --nproc_per_node 4 full_finetune_distributed --config llama3/8B_full checkpointer.checkpoint_dir=<YOUR_CHECKPOINT_DIR>
#
# This config works best when the model is being fine-tuned on 2+ GPUs.
# Single device full finetuning requires more memory optimizations. It's
# best to use 8B_full_single_device.yaml for those cases
# Tokenizer
tokenizer:
_component_: torchtune.models.llama3.llama3_s_tokenizer
path: ../model_zoo_llama3.2/tokenizer.model
max_seq_len: 4096
# Dataset
dataset:
_component_: torchtune.datasets.chat_dataset
source: homebrewltd/mixed-instruction-speech-whispervq-v3-full-phase2-3
conversation_column: conversations
conversation_style: openai
split: train
train_on_input: True
seed: 42
shuffle: False
# Model Arguments
model:
_component_: torchtune.models.llama3_2.llama3_2_s_3b
# path: model_zoo/Llama3.1_s_8b_init
checkpointer:
_component_: torchtune.training.FullModelHFCheckpointerSaveSteps
checkpoint_dir: ../model_zoo_llama3.2/llama3.2-s-3b-base
checkpoint_files: [
model-00001-of-00002.safetensors,
model-00002-of-00002.safetensors,
]
recipe_checkpoint: null
output_dir: ../model_zoo_llama3.2/llama3.2-3B-s-instruct
model_type: LLAMA3_2
resume_from_checkpoint: False
save_every_n_steps: 1000
max_checkpoints: 3
# Fine-tuning arguments
batch_size: 3
epochs: 1
max_steps_per_epoch: null
gradient_accumulation_steps: 12
compile: False
# Optimizer and Scheduler
optimizer:
_component_: torch.optim.AdamW #change this to use adam_mini: torchtune.modules.optimizer.Adam_mini
weight_decay: 0.005
lr: 7e-5
fused: True
lr_scheduler:
_component_: torchtune.modules.get_cosine_schedule_with_warmup
num_warmup_steps: 62
loss:
_component_: torch.nn.CrossEntropyLoss
fsdp:
cpu_offload: False
# Training env
device: cuda
dtype: bf16
# Memory management
enable_activation_checkpointing: True
memory_efficient_fsdp_wrap: True
ac_mode: 'selective'
# Logging
metric_logger:
_component_: torchtune.training.metric_logging.DiskLogger
log_dir: ${output_dir}
output_dir: ../model_zoo_llama3.2/llama3.2-3B-s-instruct-log/
log_every_n_steps: 1
log_peak_memory_stats: False |