Spaces:

hebrew-llm-leaderboard

/

leaderboard

like 28

Running on CPU Upgrade

btw the main leaderboard updated to use a different categorization system
🟢 pretrained
🟩 continuously pretrained
🔶 fine-tuned on domain-specific datasets
💬 chat models (RLHF, DPO, IFT, ...
🤝 base merges and moerges

maybe also add a [category]:
🆎 : language adapted (FP, FT, ...)

the change was due to the fact that it: either wasn't always clear , or from a practical standpoint, which category to put them in

so it's
🟢 pretrained → 🟢 or 🟩
🔶 fine-tuned on domain-specific datasets → 🔶
⭕ instruction-tuned → 💬
🟦 RL-tuned → 💬

the following are direct merges/MoE's

• SicariusSicariiStuff/Zion_Alpha_Instruction_Tuned_SLERP
• uygarkurt/llama-3-merged-linear
• kekmodel/StopCarbon-10.7B-v5
• jeonsworld/CarbonVillain-en-10.7B-v4
• invalid-coder/Sakura-SOLAR-Instruct-CarbonVillain-en-10.7B-v2-slerp
• shadowml/BeagSake-7B
• zhengr/MixTAO-7Bx2-MoE-v8.1
• yunconglong/DARE_TIES_13B
• yunconglong/Truthful_DPO_TomGrc_FusionNet_7Bx2_MoE_13B
• shanchen/llama3-8B-slerp-med-chinese

🟩 continuously pretrained:

• yam-peleg/Hebrew-Mistral-7B
• yam-peleg/Hebrew-Gemma-11B-V2

🆎 : language adapted (FP, FT, ...) :

• ronigold/dictalm2.0-instruct-fine-tuned-alpaca-gpt4-hebrew
• SicariusSicariiStuff/Zion_Alpha_Instruction_Tuned
• ronigold/dictalm2.0-instruct-fine-tuned
• SicariusSicariiStuff/Zion_Alpha

maybe add these models?

• 🟩 yam-peleg/Hebrew-Mistral-7B-200K (FP32, BF16 is closest)
• 🟩 yam-peleg/Hebrew-Mixtral-8x22B (FP16)
• 💬 yam-peleg/Hebrew-Gemma-11B-Instruct (FP16)