Start with an IntraLLM model, a supported open-source model or your own compatible model. Prepare your data, configure the training run and use your available GPU infrastructure to create AI better suited to your organisation’s language, knowledge and objectives.
Every organisation has its own terminology, expertise and ways of working. Model Training gives your team a structured environment for adapting supported AI models using relevant examples and datasets.
From initial configuration to the completed result, the workflow remains connected inside IntraLLM. Your team can prepare the run, follow its progress, review the outcome and make the resulting model available for use—all without piecing together a separate training environment.
Fine-tune supported models for more specialised tasks, subject areas and patterns of communication.
Manage model selection, datasets, training settings, compute and results through one integrated experience.
Run training using the GPU resources supporting your IntraLLM deployment, with models and data remaining within your managed environment.
Move through one connected workflow for selecting a model, preparing data,
running GPU training and putting the result to work.
PHASE 01
Choose a suitable starting model, organise your datasets and confirm the inputs before allocating GPU resources.
Select from IntraLLM Model Hub, supported open-source models, or compatible models developed by your organisation. This flexibility allows each training project to begin with the right balance of capability, size and resource requirements.
Upload individual dataset files or use containers to organise related files for a particular model, project or use case. Source documents can be parsed and transformed into structured outputs, with controls for preserving document structure, extracting images and retaining relevant formatting.
Choose a prepared dataset file or supported dataset container for training, then add separate validation data when required. Clear selection and status information helps your team confirm the inputs before committing compute resources.
PHASE 02
Configure the run, launch it on available GPU resources and monitor progress from the IntraLLM workspace.
Adjust the training parameters, runtime options and compute configuration according to the model, dataset and intended outcome. The workspace presents the essential controls in one place, making each run easier to prepare, understand and repeat.
Track the run from the platform, review its current status and inspect the available training information as the job progresses. Training history also provides a useful record of previous runs and configurations.
PHASE 03
Review the completed run, save the resulting model to Model Hub and make it available in IntraLLM Chat.
Once training is complete, review the training details and recorded outcomes. Save the resulting model to the Model Hub, make it available within the appropriate workspace and select it for use in IntraLLM Chat.
Model Training is part of the wider IntraLLM experience.
The outcome is not left isolated in a training environment:
it becomes a model that can be managed,
accessed and used through the platform.
Work with IntraLLM-provided models, supported open-source models and compatible models from your own organisation.
Adapt general-purpose models for more focused organisational requirements and use cases.
Move from model selection and training to repository management and practical use within one platform.
Use the compute resources supporting your deployment while maintaining oversight of models, data and training activity.
Keep training configurations and outcomes visible, making it easier to learn from previous runs and plan the next iteration.
Continue developing your model portfolio as your data, use cases and preferred technologies change.
IntraLLM brings model selection, configuration, GPU compute and monitoring into one connected workspace.
Model Training supports the people responsible for turning AI strategy into something useful and sustainable.
Prepare fine-tuning runs, evaluate outcomes and develop models for more specialised requirements.
Connect trained models with the applications, assistants and experiences being developed across the platform.
Provide model-training capability through an environment connected to the organisation’s existing deployment and compute resources.
Develop internal AI capability with greater flexibility over model choice, infrastructure and future direction.
