---
language:
- en
tags:
- pytorch
- causal-lm
- pythia
license: apache-2.0
datasets:
- Anthropic/hh-rlhf
---
[Pythia-410m](https://huggingface.co/EleutherAI/pythia-410m) supervised finetuned with [Anthropic-hh-rlhf dataset](https://huggingface.co/datasets/Anthropic/hh-rlhf) for 1 epoch [(sft-model)](https://huggingface.co/lomahony/eleuther-pythia70m-hh-sft), before DPO [(paper)](https://arxiv.org/abs/2305.18290) with same dataset for 1 epoch. 

[wandb log](https://wandb.ai/pythia_dpo/Pythia_LOM/runs/b3mh4jba?workspace=user-lauraomahony)

[Benchmark evaluations](https://github.com/lomahony/direct-preference-optimization/blob/main/model_eval_files/eleuther-pythia410m-hh-dpo.txt) done using [lm-evaluation-harness](https://github.com/EleutherAI/lm-evaluation-harness/tree/big-refactor). 

See [Pythia-410m](https://huggingface.co/EleutherAI/pythia-410m) for original model details [(paper)](https://arxiv.org/abs/2101.00027).