-
GlaDOS
This is the Personality Core for GLaDOS, the first steps towards a real-life implementation of the AI from the Portal series by Valve. (by dnhkng)
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
Code is available at: https://github.com/dnhkng/GlaDOS
You can also run the Llama-3 8B GGUF, with the LLM, VAD, ASR and TTS models fitting on about 5 Gb of VRAM total, but it's not as good at following the conversation and being interesting.
The goals for the project are:
All local! No OpenAI or ElevenLabs, this should be fully open source.
Minimal latency - You should get a voice response within 500 ms (but no canned responses!)
Interruptible - You should be able to interrupt whenever you want, but GLaDOS also has the right to be annoyed if you do...
Interactive - GLaDOS should have multi-modality, and be able to proactively initiate conversations (not yet done, but in planning)
Lastly, the code base should be small and simple (no PyTorch etc), with minimal layers of abstraction.