Skip to main content

Service

LLM Finetuning

A foundation model trained further on your task and your data.

Talk to usSee OpenThai 2.0

OpenThai 2.0 was built this way: three LoRA finetuning stages on about 143,000 verified rows.

What we do​

  • Task-specific modelsClassification, extraction, generation, summarisation, translation and question answering.
  • Your languagesThai, English, Chinese, Japanese and other Asian languages.
  • Full or LoRAFull-parameter or LoRA finetuning, chosen for your data and budget.
  • Faster servingQuantisation, pruning and serving settings that cut latency and cost.

How we work​

  1. Prepare the data. We assess your data, set quality criteria, then clean and format the training set.
  2. Train. Finetuning with a hyperparameter search.
  3. Evaluate and iterate. We test against baselines and your criteria, and train again until the target is met.

What you get​

  • The finetuned model, optimised for serving.
  • Its evaluation against baselines and your quality criteria.
  • Help deploying it.

Start​

Tell us the task, the data you have and where the model has to run, and we reply with a scope and a quote. Email sale@iapp.co.th or call 02-124-4041.