5 ms·
PaLM was basically a larger version of GPT-3. It was not similar to ChatGPT. It was a pure token predictor. It was not fine-tuned for chat.
by cubefox 3y ago
PaLM was basically a larger version of GPT-3. It was not similar to ChatGPT. It was a pure token predictor. It was not fine-tuned for chat.
- sinuhe69 3y agoPaLM use a pathway architect, not a decoder only transformer as ChatGPT. At least you can look it up on Wikipedia before making such wild assumptions.