ZhangShenao's picture
Update README.md
e2d859a verified
metadata
license: mit
base_model: ZhangShenao/SELM-Llama-3-8B-Instruct-iter-1
tags:
  - alignment-handbook
  - dpo
  - trl
  - selm
datasets:
  - HuggingFaceH4/ultrafeedback_binarized
model-index:
  - name: SELM-Llama-3-8B-Instruct-iter-2
    results: []

Self-Exploring Language Models: Active Preference Elicitation for Online Alignment.

SELM-Llama-3-8B-Instruct-iter-2

This model is a fine-tuned version of ZhangShenao/SELM-Llama-3-8B-Instruct-iter-1 using synthetic data based on on the HuggingFaceH4/ultrafeedback_binarized dataset.

Model description

  • Model type: A 8B parameter Llama3-instruct-based Self-Exploring Language Models (SELM).
  • License: MIT

Results

AlpacaEval 2.0 (LC WR) MT-Bench (Average)
SELM-Llama-3-8B-Instruct-iter-3                33.47               8.29
SELM-Llama-3-8B-Instruct-iter-2                35.65               8.09
SELM-Llama-3-8B-Instruct-iter-1                32.02               7.92
Meta-Llama-3-8B-Instruct                24.31               7.93

Training hyperparameters

The following hyperparameters were used during training:

  • alpha: 0.0001
  • beta: 0.01
  • train_batch_size: 4
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 8
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • num_epochs: 1

Framework versions

  • Transformers 4.40.2
  • Pytorch 2.1.2+cu121
  • Datasets 2.14.6
  • Tokenizers 0.19.1