@natsume_shokogami@mastodon.world
2025-03-04 10:00 UTC
@davidrevoy@framapiaf.org @simonzerafa@infosec.exchange @protonprivacy@mastodon.social @mozillaofficial@mastodon.social For example, the base HuBERT model (which can be used for speech recognition, speech-to-text, emotion recognition,...) is only around 360-370 MB while even the smallest LLM (the one that similar to the models powering ChatGPT) is LLaMA 3 8B is around 16 GB.
Replies (1)
-
@natsume_shokogami@mastodon.world 2025-03-04 10:03
@davidrevoy@framapiaf.org @simonzerafa@infosec.exchange @protonprivacy@mastodon.social @mozillaofficial@mastodon.social Even worse that unlike other companies and projects that build other types of models, generative AI models like LLM and image generation (usually stable diffusion models) models are often trying to improve their efficiency but making their models more complex and much much larger and large companies are pouring trillions of dollars in that, which helped create the AI bubble and worsening the environment issues.