Open Models Almanac
Methodsdata curationdata filtering

specific method · filed under data curation

CBRN safety filtering of pre-training data

Pre-training data is screened for harmful content, especially hazardous biosecurity knowledge, using CBRN filters.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See data filtering for the methods filed beside it.

Also called CBRN safety pre-training data filtering.

sources
2
model
1
lab adopt it
1
strongest
used

Documented in

Evidence

2 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
used 2

gpt-oss-120b

OpenAIadopted2 sources

we filtered the data for harmful content in pre-training, especially around hazardous biosecurity knowledge, by reusing the CBRN pre-training filters from GPT-4o

useddata curationin gpt-oss-120b & gpt-oss-20bOpenAI
1 more from this model

we filtered the data for harmful content in pre-training, especially around hazardous biosecurity knowledge, by reusing the CBRN pre-training filters from GPT-4o.

useddata curationin gpt-oss-120b & gpt-oss-20bOpenAI
↑