Magpie-Pro Datasets (Llama-3)
Collection
Dataset built with Meta Llama 3 70B. Models are fine-tuned from Llama 3 8B.
β’
5 items
β’
Updated
β’
15
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on Magpie-Align/Magpie-Pro-300K-Filtered dataset.
It achieves performance comparable with the official Llama-3-8B-Instruct Model with SFT only!
License: Please follow Meta Llama 3 Community License.
Conversation Template: Please use Llama 3 official chat template for the best performance.
If you find the model, data, or code useful, please cite our paper:
@misc{xu2024magpie,
title={Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing},
author={Zhangchen Xu and Fengqing Jiang and Luyao Niu and Yuntian Deng and Radha Poovendran and Yejin Choi and Bill Yuchen Lin},
year={2024},
eprint={2406.08464},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
The following hyperparameters were used during training:
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.8664 | 0.0012 | 1 | 0.8860 |
0.4038 | 0.9989 | 825 | 0.4250 |
0.327 | 1.9830 | 1650 | 0.4219 |
axolotl version: 0.4.0
base_model: meta-llama/Meta-Llama-3-8B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
- path: Magpie-Align/Magpie-Pro-300K-Filtered
type: sharegpt
conversation: llama3
dataset_prepared_path: last_run_prepared
val_set_size: 0.001
output_dir: ./out_Llama-3-8B-Magpie-Pro-300K-FilteredL
sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 100
evals_per_epoch: 1
eval_table_size:
saves_per_epoch: 3
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
pad_token: <|end_of_text|>
Base model
meta-llama/Meta-Llama-3-8B