Darewin-7B

Home/Models/mlabonne/Darewin-7B

mlabonne/Darewin-7B

kind: merge
quant: merge
family: misc
modality: text-generation
root model: Intel/neural-chat-7b-v3-3
provider: mlabonne
license: apache-2.0
language: —
context: —
library: —
vRAM (GB): —
size on disk: —
downloads: 65
likes: 3
created: —
last modified: 2024-03-04T15:17:37+00:00 (2024)
huggingface: https://huggingface.co/mlabonne/Darewin-7B
tags: transformerssafetensorsmistraltext-generationmergemergekitlazymergekitbase_model:Intel/neural-chat-7b-v3-3base_model:merge:Intel/neural-chat-7b-v3-3base_model:OpenPipe/mistral-ft-optimized-1227base_model:merge:OpenPipe/mistral-ft-optimized-1227base_model:fblgit/una-cybertron-7b-v2-bf16base_model:merge:fblgit/una-cybertron-7b-v2-bf16base_model:mlabonne/NeuralHermes-2.5-Mistral-7Bbase_model:merge:mlabonne/NeuralHermes-2.5-Mistral-7Bbase_model:openaccess-ai-collective/DPOpenHermes-7B-v2base_model:merge:openaccess-ai-collective/DPOpenHermes-7B-v2base_model:openchat/openchat-3.5-0106base_model:merge:openchat/openchat-3.5-0106license:apache-2.0model-indextext-generation-inferenceendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
HellaSwag	knowledge	86.22	HellaSwag (10-Shot)	medium	`\| HellaSwag (10-Shot) \| 86.22 \|`
WinoGrande	knowledge	79.79	Winogrande (5-shot)	medium	`\| Winogrande (5-shot) \| 79.79 \|`
GSM8K	math	71.04	GSM8k (5-shot)	medium	`\| GSM8k (5-shot) \| 71.04 \|`
MMLU	knowledge	65.21	MMLU (5-Shot)	medium	`\| MMLU (5-Shot) \| 65.21 \|`
TruthfulQA	safety	60.38	TruthfulQA (0-shot)	medium	`\| TruthfulQA (0-shot) \| 60.38 \|`

none scraped.