Door categorieën bladeren
Ontdekken
Fiverr Pro
Nederlands
$
USD
Our agency will fine tune llm gpt and llama on your data with python
Your Web Data Partner for Extraction, Automation and Scale
BigiByte is a data and technology agency focused on turning complex challenges into practical, scalable solutions. We specialize in web development, web scraping, data extraction, automation, data pro...
Gescreend door Fiverr Pro
BigiByte is geselecteerd door het team van Fiverr Pro vanwege diens expertise.
Gescreend voor
Data scraping
Gegevensverwerking
Over deze dienst
Generic GPT answers miss your tone, domain, and format. You need a model that learns from your data, not another wrapper.
That is the work I do. I am Kawsar, a Python developer who fine-tunes LLMs (GPT-style APIs, Llama, and other open models) and trains custom ML models on your dataset.
What I deliver:
- Dataset check and training format for your use case
- OpenAI and Anthropic API fine-tuning setup
- Dataset cleaning, dedupe, and JSONL conversion
- LoRA or QLoRA fine-tuning on open models when it fits
- GPT-style fine-tune or custom training path when your stack needs it
- Eval summary so you see what improved vs the base model
- Adapters or model files plus a notebook you can rerun
- Docker container for your own GPU server
- Optional FastAPI demo on Premium so you can call the model from an app
- FastAPI or Flask inference endpoint
- Colab or Jupyter notebook you can rerun
I work with Hugging Face, Python, PyTorch, and clear handoff docs.
Message me your goal and a data sample before you order.
NOTE:
- You provide the dataset or access for training
- GPU or API costs may apply and are confirmed in chat
- Custom work per order
- No school or university assignments
Programmeertaal:
Python
•
JavaScript
•
TypeScript
•
Tensorflow
Datatype:
Tekst
•
Afbeeldingen
•
Multimodal
AI engine:
GPT
•
Gemini
•
DALL-E
•
DeepSeek
•
tensorflow
•
Llama
•
Langchain
•
Grok
Klanten waar we mee hebben gewerkt
MyHeritage
Internet Software & Services
I've been hired to carry out certain web extraction-related jobs in order to build a database for user reviews, which will be used to develop the product and make changes in certain instances.
mei 2022-mei 2023
Fiverr
Internet Software & Services
The project was to gather a detailed list of 5000 keywords from various profession types and list their Google Rank, Link, and even Local Search Results, etc. Information and make a Report with these!
jul 2024
Portfolio
Andere Data science en ML diensten die we aanbieden
Veelgestelde vragen
What data do you need from me?
A sample of your training examples (CSV, JSON, JSONL, or chat pairs). Message me a small sample before ordering so we confirm format and package size.
Which models can you fine-tune?
Open models such as Llama-family and similar Hugging Face models via LoRA/QLoRA, plus GPT-style API fine-tunes when that path fits. We pick the model in chat.
Do you also train classic ML models?
Yes when the job is a custom ML model on your tabular or text labels. Many orders are LLM fine-tunes; say which you need up front.
What is LoRA or QLoRA?
Efficient fine-tuning methods that adapt a base model with smaller adapters instead of full retraining. Good for domain tone and format on limited budgets.
Who pays for GPU or OpenAI fine-tune costs?
Cloud GPU or provider API training costs are confirmed before work starts. They may be buyer-paid or included by agreement. Nothing silent is charged.
Will I get code and model files?
Yes. You get a notebook or scripts plus adapters/model artifacts as scoped. Premium adds a FastAPI demo for app calls.
Is this the same as a RAG chatbot?
Fine-tuning teaches the model from examples. RAG retrieves your docs at answer time. If you need RAG instead, message me and we can scope that separately.
Is this for school or university homework?
No. I do not take school or university assignments.
