reference stringclasses 6
values | prediction stringclasses 6
values | wer float64 0.13 0.27 |
|---|---|---|
mmlu is an immediate means of testing performance claude 2 0 is the latest claude model from anthropic notux 8x7b v1 is a fine tune of mixtral wizardlm 70b is a fine tune of llama 70b | mmlu is a means of testing performance claude 2 0 is the latest claude model from anthropic notixit by 7bv1 is a fine tune of mixtral wizardlm 70b is a fine tune of lama 70b | 0.166667 |
it's actually wizardlm 70b v1 0 spin self play fine tuning that improves llms decilm 7b is a fast model with 7 billion parameters arena elo is a means of measuring performance the fastest openai model is gpt 4 turbo | it's actually wizard lm70b v1 0 spin self play fine tuning that improves llms the clm7b is a fast model with 7 billion parameters arena elo is a means of measuring performance the fastest openai model is gpt 4 terminal | 0.125 |
openchat is a fine tune then basically it's a fine tune of the mistral 7b model tricksy is an approach for fast inference using sparsity microsoft have launched phi 2 mt bench is a metric for performance eval | openchat is a fine tuned then basically it's a fine tuned off the mistral 7b model trixie is an approach for fast inference use xpercity microsoft have launched fi2 mt bench is a metric for performance evaluation | 0.236842 |
mistral medium is a larger mixture of experts claude 1 is an earlier version of claude from anthropic mixtral 8x7b instruct v0 1 is the mixture of experts with 7b models tulu 2 dpo 70b is a fine tune of the 70b model gemini pro is google's best model | mixtral medium is a larger mixture of experts cloud 1 is an earlier version of cloud from anthropic mixtral 8x7b instruct v 0 1 is the mixture of experts with 7b models tulu 2dp0 is its fine tuned 7b model gemini pro is google's best model | 0.265306 |
solar 10 7b is the fastest rather sorry it's the it's a strong it's a version of mistral 7b with extra layers claude 2 1 is the latest model from anthropic mixtral 8x7b is a mixture of experts lightning attention is another version of attention that improves inference speed | solar 10 7b is the fastest rudder sorry it's a version of miestral 7b with extra layers claude 2 1 is the latest model from anthropic miestral 8x7b is a mixture of experts lightning attention is another version of attention that improves inference speed | 0.163265 |
and yi 34b chat is a fine tune of actually i'm not sure what that is but it's i think it's a trained a pre trained model of llama style | and ye34b chat is a fine tune of actually i'm not sure what that is but i think it's a pre trained model of lama | 0.233333 |
README.md exists but content is empty.
- Downloads last month
- 6