---
language:
- en
tags:
- pytorch
- causal-lm
- pythia
license: apache-2.0
datasets:
- Anthropic/hh-rlhf
---

[Pythia-410m](https://huggingface.co/EleutherAI/pythia-410m) DPO finetuned using original DPO code with the helpful subset of [Anthropic-hh-rlhf dataset](https://huggingface.co/datasets/Anthropic/hh-rlhf) for 1 epoch. 

Checkpoints are also uploaded. 

Fully reproducible finetuning code is available on [GitHub](https://github.com/lauraaisling/direct-preference-optimization/tree/main)

[wandb log](https://wandb.ai/lauraomahony999/pythia-dpo/runs/sb6r4wt7)

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

See further details of these models in the paper [Attributing Mode Collapse in the Fine-Tuning of Large Language Models](https://openreview.net/pdf?id=3pDMYjpOxk).

You can cite these models if they are helpful as follows: 

<pre>
@inproceedings{o2024attributing,
  title={Attributing Mode Collapse in the Fine-Tuning of Large Language Models},
  author={O’Mahony, Laura and Grinsztajn, Leo and Schoelkopf, Hailey and Biderman, Stella},
  booktitle={ICLR 2024, Mathematical and Empirical Understanding of Foundation Models (ME-FoMo) workshop},
  year={2024}
}
</pre>

hf (pretrained=lomahony/pythia-410m-helpful-dpo), gen_kwargs: (None), limit: None, num_fewshot: 0, batch_size: 16
|    Tasks     |Version|Filter|n-shot|    Metric     | Value |   |Stderr|
|--------------|------:|------|-----:|---------------|------:|---|------|
|arc_challenge |      1|none  |     0|acc            | 0.2338|±  |0.0124|
|              |       |none  |     0|acc_norm       | 0.2602|±  |0.0128|
|arc_easy      |      1|none  |     0|acc            | 0.5185|±  |0.0103|
|              |       |none  |     0|acc_norm       | 0.4609|±  |0.0102|
|boolq         |      2|none  |     0|acc            | 0.6214|±  |0.0085|
|hellaswag     |      1|none  |     0|acc            | 0.3447|±  |0.0047|
|              |       |none  |     0|acc_norm       | 0.4074|±  |0.0049|
|lambada_openai|      1|none  |     0|perplexity     |19.0431|±  |0.7027|
|              |       |none  |     0|acc            | 0.3978|±  |0.0068|
|openbookqa    |      1|none  |     0|acc            | 0.2000|±  |0.0179|
|              |       |none  |     0|acc_norm       | 0.3100|±  |0.0207|
|piqa          |      1|none  |     0|acc            | 0.6779|±  |0.0109|
|              |       |none  |     0|acc_norm       | 0.6757|±  |0.0109|
|sciq          |      1|none  |     0|acc            | 0.7760|±  |0.0132|
|              |       |none  |     0|acc_norm       | 0.6690|±  |0.0149|
|wikitext      |      2|none  |     0|word_perplexity|24.3807|±  |N/A   |
|              |       |none  |     0|byte_perplexity| 1.8171|±  |N/A   |
|              |       |none  |     0|bits_per_byte  | 0.8617|±  |N/A   |
|winogrande    |      1|none  |     0|acc            | 0.5343|±  |0.0140|

hf (pretrained=lomahony/pythia-410m-helpful-dpo), gen_kwargs: (None), limit: None, num_fewshot: 5, batch_size: 16
|    Tasks     |Version|Filter|n-shot|    Metric     | Value |   |Stderr|
|--------------|------:|------|-----:|---------------|------:|---|------|
|arc_challenge |      1|none  |     5|acc            | 0.2346|±  |0.0124|
|              |       |none  |     5|acc_norm       | 0.2747|±  |0.0130|
|arc_easy      |      1|none  |     5|acc            | 0.5509|±  |0.0102|
|              |       |none  |     5|acc_norm       | 0.5198|±  |0.0103|
|boolq         |      2|none  |     5|acc            | 0.5982|±  |0.0086|
|hellaswag     |      1|none  |     5|acc            | 0.3437|±  |0.0047|
|              |       |none  |     5|acc_norm       | 0.4059|±  |0.0049|
|lambada_openai|      1|none  |     5|perplexity     |34.3002|±  |1.3044|
|              |       |none  |     5|acc            | 0.3148|±  |0.0065|
|openbookqa    |      1|none  |     5|acc            | 0.1740|±  |0.0170|
|              |       |none  |     5|acc_norm       | 0.2880|±  |0.0203|
|piqa          |      1|none  |     5|acc            | 0.6741|±  |0.0109|
|              |       |none  |     5|acc_norm       | 0.6670|±  |0.0110|
|sciq          |      1|none  |     5|acc            | 0.8520|±  |0.0112|
|              |       |none  |     5|acc_norm       | 0.8350|±  |0.0117|
|wikitext      |      2|none  |     5|word_perplexity|24.3807|±  |N/A   |
|              |       |none  |     5|byte_perplexity| 1.8171|±  |N/A   |
|              |       |none  |     5|bits_per_byte  | 0.8617|±  |N/A   |
|winogrande    |      1|none  |     5|acc            | 0.5162|±  |0.0140|