2023-03-25 16:17:48 +00:00
|
|
|
# model/tokenizer
|
|
|
|
|
model_name: "zpn/llama-7b"
|
|
|
|
|
tokenizer_name: "zpn/llama-7b"
|
|
|
|
|
gradient_checkpointing: false
|
2023-03-27 17:33:13 +00:00
|
|
|
save_name: "nomic-ai/vicuna-lora-multi-turn"
|
2023-03-25 16:17:48 +00:00
|
|
|
|
|
|
|
|
# dataset
|
|
|
|
|
streaming: false
|
|
|
|
|
num_proc: 64
|
2023-03-27 17:33:13 +00:00
|
|
|
dataset_path: "data_multiturn"
|
2023-03-26 17:45:21 +00:00
|
|
|
max_length: 1024
|
|
|
|
|
batch_size: 4
|
2023-03-25 16:17:48 +00:00
|
|
|
|
|
|
|
|
# train dynamics
|
|
|
|
|
lr: 5.0e-5
|
|
|
|
|
eval_every: 2000
|
|
|
|
|
eval_steps: 100
|
|
|
|
|
save_every: 2000
|
2023-03-27 17:33:13 +00:00
|
|
|
output_dir: "ckpts/llama-7b-lora-multi"
|
2023-03-25 16:17:48 +00:00
|
|
|
checkpoint: null
|
|
|
|
|
lora: true
|
|
|
|
|
warmup_steps: 100
|
2023-03-27 17:33:13 +00:00
|
|
|
num_epochs: 2
|
2023-03-25 16:17:48 +00:00
|
|
|
|
|
|
|
|
# logging
|
2023-03-26 17:45:21 +00:00
|
|
|
wandb: true
|
|
|
|
|
wandb_entity: vicuna
|
|
|
|
|
wandb_project_name: vicuna
|
|
|
|
|
seed: 42
|
