Llama 3

Llama 3 is a family of models developed by Meta. They are considered one of the most capable LLMs published till date.

note

Model	Params	Context Length	GQA	Token Count	Knowledge Cutoff
Llama-3 8B Instruct	8B	8k	Yes	15T+	March 2023
Llama-3 70B Instruct	70B	8k	Yes	15T+	December 2023

tip

We recommend this model for complex tool calling scenarios, but users should be aware of the relatively short context length.

Llama-3 8B Instruct

• rubra-ai/Meta-Llama-3-8B-Instruct
• rubra-ai/Meta-Llama-3-8B-Instruct-GGUF

Model	Function Calling	MMLU	GPQA	GSM-8K	MATH	MT-bench	MT-bench Pairwise Comparison
							Win	Loss	Tie	Win Rate	Loss Rate	Adjusted Win Rate
Llama-3 8B Instruct	-	65.69	31.47	77.41	27.58	8.07	41	42	77	0.25625	0.2625	0.496875
Rubra Enhanced Llama-3 8B Instruct	89.28%	64.39	31.70	68.99	23.76	8.03	42	41	77	0.2625	0.25625	0.503125

Llama-3 70B Instruct

• rubra-ai/Meta-Llama-3-70B-Instruct
• rubra-ai/Meta-Llama-3-70B-Instruct-GGUF
• rubra-ai/Meta-Llama-3-70B-Instruct-AWQ

Model	Function Calling	MMLU	GPQA	GSM-8K	MATH	MT-bench	MT-bench Pairwise Comparison
							Win	Loss	Tie	Win Rate	Loss Rate	Adjusted Win Rate
Llama-3 70B Instruct	-	79.90	38.17	90.67	44.24	8.88	58	28	74	0.3625	0.1750	0.59375
Rubra Enhanced Llama-3 70B Instruct	97.85%	75.90	33.93	82.26	34.24	8.36	28	58	74	0.1750	0.3625	0.40625