DISTIL LABS

50x smaller size and matching quality compared to cutting-edge LLMs like LLama3 405b.

NLP is our game: Streamline NLP tasks like classification, extractive QA, NER, summarisation, function calling, and ranking.

Upload your data

We need a “task description”, a few hundred examples of what the model is supposed to do (training data), and any additional data about the problem (documentation, unlabelled examples, …)

Evaluate the model

We train the model and share benchmarks based on a test dataset we isolate from the training data. You can either iterate (change the task description, upload more data) or choose to Deploy the model

Access and integrate the model

We can either share the model binaries with you or handle the deployment and provide an API endpoint that you can query & integrate with your product.

Key benefits

Best of both worlds – LLMs & conventional Machine Learning

Accelerating your development process

You need a much smaller labeled training dataset than ever before; there is no need to train, manage and pay for subject matter expert human annotators. 100-1000 labeled examples instead of 10 000s.

Reducing latency & costs of your AI pipeline

A smaller specialized model enables deployment on cheaper & faster infrastructure. You don't need to pass on as much context in the prompt further reducing your token usage.

Local deployment

Small models can be easily deployed directly on mobile hardware as a part of the application and give the developers total control. This means your applications are not reliant on a strong network connection and helps ensure data privacy compliance within your organization.

10x

Less data needed to train a performant model

Same accuracy using significantly less data thanks to model distillation

50x

Smaller model with the same accuracy

Specialized Llama3 8b compared to instruction tuned Llama3 405b

10x

Faster model inference

Specialized Llama3 8b compared to instruction tuned Llama3 405b

100x

Less time needed to deploy specialized model

Comparing to a standard 7-week annotation & training project