👑 FALCON LLM beats LLAMA

Introducing Falcon-40B. A new language model trained on 1000B tokens.

What's included:

- 7B and 40B models made available by TII
- surpasses LLaMA 65B and other models like MPT and RedPajama on the Open LLM Leaderboard
- architecture is optimized for inference, with FlashAttention and multiquery
- Instruct model available
- license allows personal and research use and commercial use with limitations

https://huggingface.co/tiiuae/falcon-40b


❤️ If you want to support the channel ❤️
Support here:
Patreon - https://www.patreon.com/1littlecoder/
Ko-Fi - …

architecture inference language language model llama llm redpajama tokens