Dataset · flytech
Python instructions for 25k tasks
flytech/python-codes-25k
Cleaned set of ~25k Python instructions with input/output for training and evaluating code models.
What's inside
A dataset of 24,813 records with instruction, input, output, and text fields; 24,580 unique instructions and 24,581 unique outputs. It covers Python tasks and behavioral aspects with an average length of about 508 tokens. MIT license, language en.
What you can do with it
- fine-tune a Python assistant
- run SFT or LoRA
- evaluate code generation and summarization
- Tasks
- Text classification, Question answering, Token classification, Summarization, Text generation
- Size
- 10K–100K rows
- Downloads this month
- 10.3K
- Likes
- 193
Models trained on it
- haydarkadioglu/Qwen3-0.6B-lora-python-expert-fine-tunedText generation 1.2K
- mradermacher/Atem-v1-1.5B-i1-GGUFText generation 739
- mradermacher/Atem-0.6B-i1-GGUF 529
- FlameF0X/Qwen2-0.2B-itText generation 409
- 56m/Dumb-1.2 365
- EphAsad/Atem-v1-1.5BText generation 305
- mradermacher/gpt2_pycode-i1-GGUF 302
- ShyamSaran-18/Python-wizard-Llama-3.1-8bText generation 281
How to load it
from datasets import load_dataset
ds = load_dataset("flytech/python-codes-25k")