David comments

Results 204 comments of


                                            David

Responses ending in "<|eot_id|><|start_header_id|>user<|end_header_id|>" or "<!--end_eot}}"

To see that, change the logging level. I have the current context and lots of other information logged at the 'info' level, so you can track that easily.

Responses ending in "<|eot_id|><|start_header_id|>user<|end_header_id|>" or "<!--end_eot}}"

The template *should* be the official chat template used to train Llama-3. I also need to check the default context length for the llama.cpp[server]! I will look into that today.

Responses ending in "<|eot_id|><|start_header_id|>user<|end_header_id|>" or "<!--end_eot}}"

Yep. the context length was at the default length of 512! I have made that a parameter, and extended the length to 8192, and is seems to be much better...

whisper error

Yep. Make generates Libwhisper.so which must be in the GLaDOS/submodules/whisper.cpp directory, or you get this error.

Continuous TTS voice interruptions

I get this on my laptop. I think to run GLaDOS, you need a high quality conference microphone. These have omnidirectional microphones and build in self-cancelling, so they don't hear...

Ubuntu Serveur - Install service Glados

There is a pull request for a Docker version. Maybe have a look there? It's still quite technical though.

Ubuntu Serveur - Install service Glados

Closing this, as I think the Docke solution should be acceptable here.

CPU usage during idle listening 50%

Seems like its inferencing the VAD on CPU, but that's still too much CPU...

CPU usage during idle listening 50%

Make sure you have uninstalled onnxruntime before installing onnxruntime-gpu.

CPU usage during idle listening 50%

Seems this is CPU only. https://github.com/snakers4/silero-vad/discussions/388