pursuitcurves

4 ms·

1. 1.▲
   Show HN: Finding Speaker Similarity Scores for Voice Cloning Models(github.com)

   1 points by pursuitcurves 3y ago|1 comments
2. 2.▲
   by pursuitcurves 3y ago

   ref blog: https://black.dubverse.ai/p/evals-are-all-we-need
3. 3.▲
   Why we need evals for audio generative AI models?(black.dubverse.ai)

   5 points by pursuitcurves 3y ago|1 comments
4. 4.▲
   by pursuitcurves 3y ago

   What criteria should we use to determine the best model when we have text-to-speech models such as ElevenLabs, Bark, etc? How do we scale this up when these audio models have their "stable diffusion moment" (thanks simonw for the
5. 5.▲
   by pursuitcurves 3y ago

   not sure if it's a custom model or some wrapper/fine tuning around Eleuther AI models (GPTJ, Neo etc) but they react like ChatGPT on SolidGoldMagikarp etc type tokens. https://www.lesswrong.com/posts/aPeJE8bSo
6. 6.▲
   GPT4 is a snitch, ChatGPT isn't(dubverseblack.substack.com)

   2 points by pursuitcurves 3y ago|2 comments
7. 7.▲
   by pursuitcurves 3y ago

   exploring prompt injections inside of GPT4 compressions.