Datasets:
File size: 1,278 Bytes
e5db5a0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 | ---
license: apache-2.0
language:
- ar
pretty_name: Arabic Instruct Curated
tags:
- instruction-tuning
---
# Arabic Instruct Curated
Curated Arabic instruction-following dataset for SFT of Qwen2.5 LoRAs
(`RASHID778/qwen2.5-3b-arabic-lora-v2` and `-7b-`).
Format: JSON Lines, one record per line:
```json
{"messages": [{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}], "source": "alpaca"}
```
## Composition (17,313 rows)
| source | kept | raw | notes |
|---|---:|---:|---|
| `Yasbok/Alpaca_arabic_instruct` | 6,000 | 52,002 | cap reached; filtered short/repetitive/parroting |
| `AhmedBou/Arabic_instruction_dataset_for_llm_ft` (train) | 3,966 | 10,000 | dropped ar->en translation tasks (Latin-dominated) |
| `yrrhall/Arabic_Aya14200` (`aya_collection-aya_dataset`) | 7,347 | 13,960 | dedup removed 3,135 near-duplicates |
Excluded entirely: `Mohamed-Sami/instruction-fine-tuning-arabic-dataset` (864K rows,
template noise + garbled OCR content).
## Filtering
- user >= 10 chars, assistant >= 15 chars
- no 8+ repeated-character runs
- no control characters
- Latin/letter ratio > 0.5 dropped (keeps Arabic-dominant samples)
- exact + normalized near-duplicate removal (Alef/Ya/Taa-Marbuta normalization, sha256)
- shuffled with seed 42
|