|
|
|
@ -1,8 +1,8 @@
|
|
|
|
|
# model/tokenizer
|
|
|
|
|
model_name: "zpn/llama-7b"
|
|
|
|
|
tokenizer_name: "zpn/llama-7b"
|
|
|
|
|
model_name: # update
|
|
|
|
|
tokenizer_name: # update
|
|
|
|
|
gradient_checkpointing: false
|
|
|
|
|
save_name: "nomic-ai/vicuna-lora-multi-turn"
|
|
|
|
|
save_name: "nomic-ai/gpt4all-lora-multi-turn"
|
|
|
|
|
|
|
|
|
|
# dataset
|
|
|
|
|
streaming: false
|
|
|
|
@ -16,7 +16,7 @@ lr: 5.0e-5
|
|
|
|
|
eval_every: 2000
|
|
|
|
|
eval_steps: 100
|
|
|
|
|
save_every: 2000
|
|
|
|
|
output_dir: "ckpts/llama-7b-lora-multi"
|
|
|
|
|
output_dir: "ckpts/gpt4all-lora-multi"
|
|
|
|
|
checkpoint: null
|
|
|
|
|
lora: true
|
|
|
|
|
warmup_steps: 100
|
|
|
|
@ -24,6 +24,6 @@ num_epochs: 2
|
|
|
|
|
|
|
|
|
|
# logging
|
|
|
|
|
wandb: true
|
|
|
|
|
wandb_entity: vicuna
|
|
|
|
|
wandb_project_name: vicuna
|
|
|
|
|
seed: 42
|
|
|
|
|
wandb_entity: # update
|
|
|
|
|
wandb_project_name: # update
|
|
|
|
|
seed: 42
|
|
|
|
|