I will fine tune a qwen model on your data for ollama


About this gig
I will fine-tune a Qwen 7-8B class model on YOUR data, and deliver it ready for Ollama.
What you get:
- Data cleaning and verification: dedup, format checks, arithmetic re-computation where applicable
- QLoRA fine-tuning on a Qwen 7-8B class model
- Evaluation report on your own test questions
- LoRA adapter, GGUF Q4_K_M, and Modelfile ready for Ollama
What I need from you:
- Your instruction data in JSONL format (please send a small sample first)
- A short description of the task and style you want
- 20-30 test questions for acceptance
- Confirmation that you have the rights to use the data
Please note:
- No guaranteed accuracy or benchmark scores; results depend on your data quality and size
- 7-8B class models only
- I do not train on data you do not have rights to
- For child-related use, the model is assistive only and requires adult supervision
- Training uses shared free GPU resources, so large orders may queue; one full run takes about 4.5 hours of GPU time plus preparation and evaluation
Get to know Lin
Automation Scripts and Data Processing Specialist
- FromCanada
- Member sinceOct 2026
Languages
Chinese, English
FAQ
What data format do you need?
JSONL instruction data. Please send a small sample first so I can check the format and size before the full run.
How long does training take?
One full run takes about 4.5 hours of GPU time plus preparation and evaluation. Training uses shared free GPU resources, so large orders may queue.
Do you guarantee accuracy or benchmark scores?
No. Results depend on your data quality and size. I report results on your own test questions so you can judge for yourself.
