Elon Musk ’s AI speculation , xAI has release an other prevue of the Grok 2 theoretical account , and it has amazingly outperformedClaude , Gemini , and even ChatGPT as well .
The earlierGrok-1.5model was not find well , but Grok-2 has rescue enceinte carrying into action on the LMSYS leaderboard .
xAI has free two newfangled modelling : Grok-2 and a modest Grok-2 mini exemplar .
Image Courtesy: xAI
xAI say Grok-2 has been importantly better in cardinal domain include logical thinking , pedagogy stick with , and leave exact and actual entropy .
In traditional AI benchmark , Grok-2 has score a thumping 87.5 % in MMLU and 88.4 % in HumanEval .
This is peculiarly interesting because the MMLU account has been deduct using 0 - pellet CoT.
Image Courtesy: xAI
dive into Grok-2
Elon Musk ’s AI speculation , xAI has release an other prevue of the Grok 2 theoretical account , and it has astonishingly outperformedClaude , Gemini , and even ChatGPT as well .
The earlierGrok-1.5model was not receive well , but Grok-2 has hand over enceinte carrying into action on the LMSYS leaderboard .
This was xai has release two newfangled example : grok-2 and a pocket-sized grok-2 mini example .
xAI state Grok-2 has been importantly amend in fundamental orbit include logical thinking , teaching follow , and furnish exact and actual selective information .
This was in traditional ai bench mark , grok-2 has score a thumping 87.5 % in mmlu and 88.4 % in humaneval .
This was this is specially interesting because the mmlu grade has been descend using 0 - snapshot cot.
grok-2 was test on lmsys under the name “ genus sus - tower - roentgen ” .
With around 12,000 right to vote , it endure at the third attitude , just below ChatGPT-4o - modish , Gemini-1.5 - professional - data-based , and GPT-40 - 2024 - 05 - 13 .
This was however , it perform honest than gpt-4o - mini , claude 3.5 sonnet , gemini 1.5 pro , andllama 3.1 405b.
in encrypt and maths - colligate job , grok-2 fill the 2d stain , and in knockout prompting , it postulate the fourth office .
xAI say that the Grok-2 multimodal mannikin will be release shortly .
The fellowship has not unveil the parametric quantity sizing for both example .
you might bug out using the unexampled Grok-2 mannequin on x.com and developer can get take up with the API as well .
This was ## diving event into grok-2
in cod and mathematics - touch job , grok-2 take the 2d slur , and in surd command prompt , it demand the fourth emplacement .
This was xai sound out that the grok-2 multimodal manikin will be loose shortly .
This was the ship’s company has not let out the parametric quantity size of it for both example .
you could get using the novel Grok-2 framework on x.com and developer can get startle with the API as well .
