Kategorien durchsuchen
Entdecken
Fiverr Pro
Deutsch
$
USD
I will finetune llms with lora for your custom use case
AI Product Engineer
AI Product Engineer focused on production-grade LLM systems, optimized inference, and scalable agent architectures.
I specialize in cost-efficient model deployment, quantized high-performance inferen...
Level 1
Hat bestimmte Leistungskriterien erfüllt und zeigt großes Potenzial auf dem Marktplatz.
Über diesen Service
I specialize in fine-tuning modern open-source LLMs (up to 70B) using LoRA & QLoRA delivering fast, cost-efficient, production-ready custom models tailored to your exact use case.
Whether you need a domain expert chatbot, structured JSON outputs, or a private on-prem assistant, I build clean, reproducible fine-tuning pipelines with full evaluation and deployment support.
MODELS I SUPPORT:
Llama 3 / 3.1 / 3.2 (1B to 70B)
Mistral 7B & Mixtral 8x7B
Qwen 2 / 2.5 (0.5B to 72B)
Gemma 2 (2B, 9B, 27B)
Phi-4 / Phi-3
DeepSeek v2 / v3
IBM Granite
WHAT'S INCLUDED IN EVERY ORDER:
Custom LoRA / QLoRA training config
Data cleaning, formatting & preprocessing
Full training run (Hugging Face + Unsloth)
Evaluation report (loss curves & benchmarks)
Merged model export (GGUF / safetensors)
Deployment-ready weights + setup instructions
IDEAL USE CASES:
Business chatbots & customer support agents
Domain Q&A (legal, medical, finance, HR)
Structured output generation (JSON, SQL, code)
RAG-augmented fine-tuned assistants
Private on-prem LLM deployment
Message me before ordering to discuss your dataset, goals, and the best approach for your project!
Programmiersprache:
Python
•
Pytorch
Datentyp:
Text
•
BILDER
KI-Engine:
GPT
•
Gemini
•
DeepSeek
•
Bert
•
RoBERTa
•
Llama
•
Falcon
•
PyTorch
Mein Portfolio
FAQ
What do you need from me to start?
Please provide your dataset (JSON, CSV, or text format), clear examples of desired outputs, and your target use case. If unsure about formatting, I can guide you.
Which models do you support?
I fine-tune modern open-source models up to 14B, including Qwen, Llama 3, Gemma, Mistral, Phi-4, IBM Granite, and selected DeepSeek variants.
Do you offer full fine-tuning?
By default, I use efficient LoRA/QLoRA methods. Full fine-tuning or larger models require a custom offer.
Can you deploy the model?
Yes. Premium packages include deployment guidance or API/Docker setup. I also support private/offline deployments.
What counts as a revision?
A revision means one additional training run using the same dataset size and scope. Major dataset changes require a new order.
Do you provide GPU compute?
Yes, limited GPU time is included per package. If additional training hours are required, I will inform you before proceeding.
Can you sign an NDA?
Yes, I’m open to NDAs for business or private datasets.
