our second generation language model
Dialogue optimised, trained on a vast amount of data and with a leading feature set.
Thanks to our groundbreaking Infinity Context and Verum Engine technologies NLM 2 generates responses you can rely on, while considering all your company knowledge.
Driven by research
We are passionate about solving computer science challenges - inspired by the capabilites that make a noticable difference for our users, we're at the forefront of AI research to engineer breakthroughs that enable new AI capabilites and areas of application.
With our strong research focus, we are proud to showcase three of our most important results that enable the NLM 2 model:
Most modern large language models have a limitation in the amount of text they can consider when generating a response. With Infinity Context, we're not restricted by the context window. This means, NLM 2 can interact with prompts of any length, get additional information from your data sources, interact with scientific papers, print articles and other long texts.
To use an AI assistant in the enterprise, you need to be able to trust its generated responses. Our Verum Engine is not a traditional guardrail solution, but instead a part of the model itself, that ensures factual responses. To further increase the usefulness, it even provides a confidence score and, if applicable, cites sourses for the response. Like this, you know that the responses are reliable.
Your data changes constantly - and so can our model. Thanks to Flash Retraining we can fine-tune in near real time, so NLM-Dialogue's answers will never be out of date.
Ready for enterprise - ready for you
NLM 2 is not just a language model - it's a platform for building the next generation of AI applications.
Explore additonal features
No matter what your organisational goals are, we designed NLM 2 for you: Explore the features that make our second generational model truly enterprise grade.
- More than 15 Languages
- NLM 2 supports English, German, Ukrainian, Danish, Dutch, Estonian, Finnish, French, Greek, Italian, Norwegian, Polish, Portuguese, Spanish, Swedish and more to be announced
- Fine-Tuning with Low Rank Adaptation
- Fine-tuning allows for the model to adopt your organisations tone and communication style, learn to interact with organisation and sector specific vocabulary and to adopt your organisations knowledge to expand base training.
- Conversational Model
- The NLM 2 base model is fine-tuned on question/answer pairs as a conversational model. This makes prompting the model very easy.
- Specialised Data Sets
- A lot of problems plaguing many language models stem from the fact that they have been trained on large, but low quality data sets. For NLM 2 we are utilising a purpose build data set assambled exclusively for NLM 2.
- Optionally On-Premise
- The power of NLM 2 is not just limited to the cloud: we licence the model for commercial use, in your data centre - including the same high level of enterprise support.
- Multiple Scalings
- NLM 2 will be available in multiple scalings, helping you find the optimal balance of inference cost and capabiltities.
We piqued your interest?
NLM 2 is currently in development. Register your interest here, to be the first to gain access: