HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks

DiscoLM-70B-GGUF

TheBloke/DiscoLM-70B-GGUF

kind
gguf
quant
gguf (2-bit)
family
misc
modality
text-generation
root model
DiscoResearch/DiscoLM-70b
provider
TheBloke
license
llama2
language
en, de
context
8192,
library
transformers
vRAM (GB)
—
size on disk
—
downloads
317
likes
4
created
—
last modified
2023-12-06T17:11:41+00:00 (2023)
huggingface
https://huggingface.co/TheBloke/DiscoLM-70B-GGUF
tags
transformersggufllamagoliathdeutschllama2discoresearchtext-generationendedataset:Open-Orca/SlimOrca-Dedupdataset:teknium/openhermesdataset:meta-math/MetaMathQAdataset:migtissera/Synthia-v1.3dataset:THUDM/AgentInstructdataset:LeoLM/German_Songsdataset:LeoLM/German_Poemsdataset:LeoLM/OpenSchnabeltierdataset:bjoernp/ultrachat_debase_model:DiscoResearch/DiscoLM-70bbase_model:quantized:DiscoResearch/DiscoLM-70blicense:llama2region:us

Benchmark scores (9)

benchmarkcategoryscorecontextconfevidence
HellaSwagknowledge85.41HellaSwag (10-shot)medium| HellaSwag (10-shot) | 85.41 |
WinoGrandeknowledge83.27Winogrande (5-shot)medium| Winogrande (5-shot) | 83.27 |
GSM8Kmath70.6GSM8Kmedium| GSM8K | 70.6 |
MMLUknowledge68.64MMLU (5-shot)medium| MMLU (5-shot) | 68.64 |
MMLUknowledge64.7MMLUmedium| MMLU | 64.7 |
GSM8Kmath63.68GSM8k (5-shot)medium| GSM8k (5-shot) | 63.68 |
BBHreasoning63.4BBHmedium| BBH | 63.4 |
TruthfulQAsafety57.69TruthfulQA (0-shot)medium| TruthfulQA (0-shot) | 57.69 |
MATHmath17.8Mathmedium| Math | 17.8 |

Discussions (0)

none scraped.

Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.