openai chatgpt模型项目
openai chatgpt网站收集整理
随着大量的大型语言模型(LLM)和聊天机器人每周都在发布,通常都对其性能进行了浮夸的宣称,很难筛选出开源社区正在取得的真正进展,以及哪种模型是当前的技术水平 开放式LLM排行榜旨在跟踪、排名和评估LLM和聊天机器人的发布情况。我们在Eleuther AI语言模型评估线束的4个关键基准上评估模型,Eleuther人工智能语言模型评估套件是一个统一的框架,用于在大量不同的评估任务上测试生成语言模型。该排行榜的一个关键优势是,社区中的任何人都可以在 GPU集群,只要它是 轮毂上有重物的变压器模型。我们还支持为非商业许可模型(如LLaMa)评估具有增量权重的模型。
Model | Revision |
---|---|
llama-65b | main |
MetaIX/GPT4-X-Alpasta-30b | main |
digitous/Alpacino30b | main |
Aeala/GPT4-x-AlpacaDente2-30b | main |
TheBloke/dromedary-65b-lora-HF | main |
llama-30b | main |
TheBloke/vicuna-13B-1.1-HF | main |
chavinlo/gpt4-x-alpaca | main |
eachadea/vicuna-13b | main |
stable-vicuna-13b | main |
eachadea/vicuna-7b-1.1 | main |
llama-13b | main |
alpaca-13b | main |
wordcab/llama-natural-instructions-13b | main |
chainyo/alpaca-lora-7b | main |
llama-7b | main |
nomic-ai/gpt4all-j | main |
EleutherAI/gpt-neox-20b | main |
togethercomputer/RedPajama-INCITE-Base-7B-v0.1 | main |
OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5 | main |
databricks/dolly-v2-12b | main |
Pirr/pythia-13b-deduped-green_devil | main |
databricks/dolly-v2-7b | main |
EleutherAI/gpt-j-6b | main |
facebook/opt-13b | main |
KoboldAI/OPT-13B-Nerybus-Mix | main |
togethercomputer/RedPajama-INCITE-Base-3B-v1 | main |
databricks/dolly-v2-3b | main |
HuggingFaceH4/starchat-alpha | main |
Salesforce/codegen-16B-multi | main |
stabilityai/stablelm-tuned-alpha-7b | main |
facebook/opt-1.3b | main |
gpt2-xl | main |
aisquared/dlite-v2-774m | main |
gpt2-large | main |
gpt2-medium | main |
cerebras/Cerebras-GPT-1.3B | main |
facebook/opt-350m | main |
facebook/opt-125m | main |
gpt2 | main |
distilgpt2 | main |
cerebras/Cerebras-GPT-111M | main |
vicgalle/gpt2-alpaca-gpt4 | main |
bigscience/bloomz-3b | main |
lamini/instruct-tuned-3b | main |
hakurei/instruct-12b | main |
stabilityai/stablelm-tuned-alpha-3b | main |
pythainlp/wangchanglm-7.5B-sft-en-sharded | main |
stabilityai/stablelm-base-alpha-3b | main |