FLAN-UL2 vs. Koala
LLM Comparison
FLAN-UL2
Overview
Similar to FLAN-T5, FLAN-UL2 is a model based on Google's popular T5 architecture with an upgraded pre-training procedure dubbed UL2. On most NLU benchmarks, FLAN-UL2 outperforms FLAN-T5 by a significant margin.
Initial release: 2023-03-03
Reference
Further Reading
Koala
Overview
Released alongside Vicuna, Koala is one of many descendants of the Meta LLaMA model trained on dialogue data collected from the web. On the developers' benchmarks, Koala outperforms its sibling Alpaca, though its adoption has been significantly less than that of its other sibling, Vicuna. Due to its use of LLaMA, only research use is permitted.
Initial release: 2023-04-03
Further Reading
Looking for an LLM API/SDK that works out of the box? No prompts or ad hoc guardrails.
Sapling APIMore Comparisons
FLAN-UL2 |
Koala |
|
|---|---|---|
| Products & Features | ||
| Instruct Models | ||
| Coding Capability | ||
| Customization | ||
| Finetuning | ||
| Open Source | ||
| License | Apache 2.0 | Noncommercial |
| Model Sizes | 20B | 7B, 13B |