Files
pythia410m-sft-tldr/code/configs/gptrm1b_pythia160m_r8_alpha32.yml
ModelHub XC 4bb3a5553f 初始化项目,由ModelHub XC社区提供模型
Model: mnoukhov/pythia410m-sft-tldr
Source: Original Platform
2026-05-30 14:21:21 +08:00

17 lines
429 B
YAML

dataset_name: mnoukhov/openai_summarize_comparisons_relabel_pythia1b
model_name: /home/toolkit/huggingface/tldr_sft_pythia160m_fp32
load_in_8bit: False
bf16: False
fp16: True
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
gradient_accumulation_steps: 2
gradient_checkpointing: False
learning_rate: 1e-5
optimizer_type: adamw_torch
num_train_epochs: 3
use_lora: True
lora_all_linear: True
lora_r: 8
lora_alpha: 32