File size: 3,578 Bytes
839d7d0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
Model,MT-bench (score),Arena Elo rating,MMLU,License,Link
GPT-4,8.99,1210,0.864,Proprietary,https://openai.com/research/gpt-4
Claude-2,8.06,1142,0.785,Proprietary,https://www.anthropic.com/index/claude-2
GPT-3.5-turbo,7.94,1124,0.700,Proprietary,https://openai.com/blog/chatgpt
Claude-1,7.90,1167,0.770,Proprietary,https://www.anthropic.com/index/introducing-claude
Claude-instant-1,7.85,1140,0.734,Proprietary,https://www.anthropic.com/index/introducing-claude
Vicuna-33B,7.12,1097,0.592,Non-commercial,https://huggingface.co/lmsys/vicuna-33b-v1.3
WizardLM-30B,7.01,-,0.587,Non-commercial,https://huggingface.co/WizardLM/WizardLM-30B-V1.0
Llama-2-70b-chat,6.86,-,0.630,Llama 2 Community,https://huggingface.co/meta-llama/Llama-2-70b-hf
WizardLM-13B-v1.1,6.76,1039,0.500,Non-commercial,https://huggingface.co/WizardLM/WizardLM-13B-V1.1
Llama-2-13b-chat,6.65,989,0.536,Llama 2 Community,https://huggingface.co/meta-llama/Llama-2-13b-hf
Vicuna-13B,6.57,1052,0.558,Llama 2 Community,https://huggingface.co/lmsys/vicuna-13b-v1.5
Guanaco-33B,6.53,1039,0.576,Non-commercial,https://huggingface.co/timdettmers/guanaco-33b-merged
Tulu-30B,6.43,-,0.581,Non-commercial,https://huggingface.co/allenai/tulu-30b
Guanaco-65B,6.41,-,0.621,Non-commercial,https://huggingface.co/timdettmers/guanaco-65b-merged
OpenAssistant-LLaMA-30B,6.41,-,0.560,Non-commercial,https://huggingface.co/OpenAssistant/oasst-sft-6-llama-30b-xor
PaLM-Chat-Bison-001,6.40,1015,-,Proprietary,https://cloud.google.com/vertex-ai/docs/generative-ai/learn/models#foundation_models
MPT-30B-chat,6.39,1052,0.504,CC-BY-NC-SA-4.0,https://huggingface.co/mosaicml/mpt-30b-chat
WizardLM-13B,6.35,,0.523,Non-commercial,https://huggingface.co/WizardLM/WizardLM-13B-V1.0
Llama-2-7b-chat,6.27,970,0.458,Llama 2 Community,https://huggingface.co/meta-llama/Llama-2-7b-chat-hf
Vicuna-7B,6.17,1001,0.498,Llama 2 Community,https://huggingface.co/lmsys/vicuna-7b-v1.5
Baize-v2-13B,5.75,-,0.489,Non-commercial,https://huggingface.co/project-baize/baize-v2-13b
XGen-7B-8K-Inst,5.55,-,0.421,Non-commercial,https://huggingface.co/Salesforce/xgen-7b-8k-inst
Nous-Hermes-13B,5.51,-,0.493,Non-commercial,https://huggingface.co/NousResearch/Nous-Hermes-13b
MPT-7B-Chat,5.42,943,0.320,CC-BY-NC-SA-4.0,https://huggingface.co/mosaicml/mpt-7b-chat
GPT4All-13B-Snoozy,5.41,964,0.430,Non-commercial,https://huggingface.co/nomic-ai/gpt4all-13b-snoozy
Koala-13B,5.35,981,0.447,Non-commercial,https://bair.berkeley.edu/blog/2023/04/03/koala/
MPT-30B-Instruct,5.22,-,0.478,CC-BY-SA 3.0,https://huggingface.co/mosaicml/mpt-30b-instruct
Falcon-40B-Instruct,5.17,-,0.547,Apache 2.0,https://huggingface.co/tiiuae/falcon-40b-instruct
ChatGLM2-6B,4.96,-,0.455,Apache-2.0,https://huggingface.co/THUDM/chatglm2-6b
H2O-Oasst-OpenLLaMA-13B,4.63,-,0.428,Apache 2.0,https://huggingface.co/h2oai/h2ogpt-gm-oasst1-en-2048-open-llama-13b
Alpaca-13B,4.53,920,0.481,Non-commercial,https://crfm.stanford.edu/2023/03/13/alpaca.html
ChatGLM-6B,4.50,899,0.361,Non-commercial,https://huggingface.co/THUDM/chatglm-6b
OpenAssistant-Pythia-12B,4.32,914,0.270,Apache 2.0,https://huggingface.co/OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5
RWKV-4-Raven-14B,3.98,939,0.256,Apache 2.0,https://huggingface.co/BlinkDL/rwkv-4-raven
Dolly-V2-12B,3.28,840,0.257,MIT,https://huggingface.co/databricks/dolly-v2-12b
FastChat-T5-3B,3.04,890,0.477,Apache 2.0,https://huggingface.co/lmsys/fastchat-t5-3b-v1.0
StableLM-Tuned-Alpha-7B,2.75,861,0.244,CC-BY-NC-SA-4.0,https://huggingface.co/stabilityai/stablelm-tuned-alpha-7b
LLaMA-13B,2.61,815,0.470,Non-commercial,https://arxiv.org/abs/2302.13971 |