Open book qa — Epoch AI

Open book qa — Epoch AI is run by Epoch AI. It has ranked 70 models, of which the catalogue holds 24, scoring from 22.4 to 88 on Accuracy.

measured by Epoch AI · the board itself

Place	Model	Metric	Score
1stof 70	Phi-3-mini-4k-instruct	Accuracy	88
2ndof 70	Phi-3-small-8k-instruct	Accuracy	88
3rdof 70	Phi-3-medium-128k-instruct	Accuracy	87.4
5thof 70	GPT-3.5 Turbo (1106)	Accuracy	86
9thof 70	Llama 3 8B Instruct	Accuracy	82.6
10thof 70	Mistral-7B-v0.1	Accuracy	79.8
11thof 70	gemma-7b	Accuracy	78.6
12thof 70	phi-2	Accuracy	73.6
17thof 70	falcon-180B	Accuracy	64.2
19thof 70	Llama-2-70b-hf	Accuracy	60.2
24thof 70	Llama-2-7b	Accuracy	58.6
32ndof 70	llama-7b	Accuracy	57.2
34thof 70	Llama-2-13b	Accuracy	57
35thof 70	falcon-40b	Accuracy	56.6
36thof 70	llama-13b	Accuracy	56.4
46thof 70	falcon-7b	Accuracy	51.6
53rdof 70	xgen-7b-8k-base	Accuracy	40.2
55thof 70	opt-1.3b	Accuracy	39.8
58thof 70	gpt-neox-20b	Accuracy	38.8
59thof 70	gpt-j-6b	Accuracy	38.2
60thof 70	phi-1_5	Accuracy	37.2
63rdof 70	stablelm-tuned-alpha-7b	Accuracy	32.4
69thof 70	gpt-neo-2.7B	Accuracy	23.2
70thof 70	gpt2-xl	Accuracy	22.4

Place

Model

Metric

Score

1stof 70

Accuracy

2ndof 70

Accuracy

3rdof 70