yam-peleg/Hebrew-Mistral-7B

Text Generation • 8B • Updated Apr 26, 2024 • 5.12k • 73

Note Current state-of-the-art base model trained in Hebrew. Continuously pre-trained from Mistral-7B, vocab extended with additional 32,000 hebrew tokens.

yam-peleg/Hebrew-Mistral-7B-200K

Text Generation • 8B • Updated May 6, 2024 • 1.3k • 15

Note Hebrew-Mistral-7B Continuously pre-trained with 200K context window.

yam-peleg/Hebrew-Mixtral-8x22B

Text Generation • 141B • Updated Apr 21, 2024 • 7 • 24

Note The largest Hebrew base model at the moment. Important note: Under-trained comparing to the others.

yam-peleg/Hebrew-Gemma-11B

Text Generation • 10B • Updated Mar 16, 2024 • 42 • 38

Note Previous state-of-the-art base model trained in Hebrew. Continuously pre-trained from Gemma-7B and extended to 11B parameters.

yam-peleg/Hebrew-Gemma-11B-V2

Text Generation • 10B • Updated Mar 16, 2024 • 5.17k • 12

Note Updated: V2! Previous state-of-the-art base model trained in Hebrew. Continuously pre-trained from Gemma-7B and extended to 11B parameters.

yam-peleg/Hebrew-Gemma-11B-Instruct

Text Generation • 10B • Updated Mar 16, 2024 • 1.41k • 24

Note Instruct fine tune of Hebrew-Gemma-11B. Quickly trained for demonstration purposes.