README.md

---
language:
- el
- en
- fr
- de
- es
- it
- pt
pretty_name: Greek Apertus 8B, post-training data
license: other
license_name: mixed-per-part
license_link: https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/blob/main/README.md
extra_gated_prompt: 'The post-training data of Greek Apertus 8B: second-stage SFT, chat SFT and preference pairs. Contains third-party data under its own licences alongside the project''s own Greek sets. By requesting access you agree to use the data for research and to respect the licences of each part.'
extra_gated_fields:
  Affiliation: text
  Intended use: text
tags:
- apertus
- greek
- glossapi
- sft
- preference-data
- chat
---

# Greek Apertus 8B: post-training data

Second-stage SFT, chat SFT and preference pairs.

## Contents

| Stage | Data |
|---|---|
| Greek maths | 5,482 maths rows + 4,496 replay rows |
| Conversation | 3,185 chat turns + 568 replay rows |
| Preference datasets | `dpo/` |

## Repositories and checkpoints

| Resource | Links |
|---|---|
| Instruction models | [Repository](https://huggingface.co/glossAPI/greek-apertus-8b-instruct), [Greek maths](https://huggingface.co/glossAPI/greek-apertus-8b-instruct/tree/stage2-sft-recipe3), [Stage 1](https://huggingface.co/glossAPI/greek-apertus-8b-instruct/tree/stage1-sft-S1), [Conversation](https://huggingface.co/glossAPI/greek-apertus-8b-instruct/tree/stage3-sft-chat-P), [03-dpo01-r4-BALs43](https://huggingface.co/glossAPI/greek-apertus-8b-instruct), [02-sft-r4-1-G4F6P1](https://huggingface.co/glossAPI/greek-apertus-8b-instruct), [07-sftv2-S1-ag0x3-wpdpo-s43](https://huggingface.co/glossAPI/greek-apertus-8b-instruct), [01-sft-r2-idB](https://huggingface.co/glossAPI/greek-apertus-8b-instruct), [08-sftv2-S1-armA-lrv3clst-s43](https://huggingface.co/glossAPI/greek-apertus-8b-instruct) |
| SFT data | [Repository](https://huggingface.co/datasets/glossAPI/greek-apertus-sft-data), [Stage 1](https://huggingface.co/datasets/glossAPI/greek-apertus-sft-data/tree/main/stage1-sft-S1), [01-sft-r2-idB](https://huggingface.co/datasets/glossAPI/greek-apertus-sft-data/tree/main/01-sft-r2-idB) |
| Pre-training data | [Repository](https://huggingface.co/datasets/glossAPI/apertus-8b-greek-cpt-modern-greek-train) |
| Greek base | [base-avg30B-50B](https://huggingface.co/glossAPI/apertus-8b-greek-cpt/tree/base-avg30B-50B) |
| Post-training data | [second_stage_sft/stage2-sft-recipe3/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/second_stage_sft/stage2-sft-recipe3), [second_stage_sft/stage3-sft-chat-P/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/second_stage_sft/stage3-sft-chat-P), [dpo/03-dpo01-r4-BALs43/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/dpo/03-dpo01-r4-BALs43), [dpo/07-sftv2-S1-ag0x3-wpdpo-s43/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/dpo/07-sftv2-S1-ag0x3-wpdpo-s43), [second_stage_sft/01-sft-r2-idB/stage2_R2_idB/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/second_stage_sft/01-sft-r2-idB), [second_stage_sft/08-sftv2-S1-armA-lrv3clst-s43/](https://huggingface.co/datasets/glossAPI/greek-apertus-post-training-data/tree/main/second_stage_sft/08-sftv2-S1-armA-lrv3clst-s43) |
| Tokenizer | [Repository](https://huggingface.co/fffoivos/apertus-tokenizer-extension) |

## Licences

Mixed, per part; each part's original card states its own terms.

## Acknowledgements

This work was implemented thanks to a grant by Swiss AI for compute on CSCS.

Collection: [Greek Apertus 8B](https://huggingface.co/collections/glossAPI/greek-apertus-8b-v05-preview-6abecae3e0362dff3324b647).