Open Models Almanac
Methodsdata curationdata mixture & curriculum

specific method · filed under data curation

Curated higher-quality mid-training data

A curated mid-training data stage that emphasizes higher-quality examples, including agentic traces and tool calls.

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

Also called curated, higher-quality data, agentic traces and tool calls.

sources
3
model
1
lab adopt it
1
strongest
used

Documented in

Liquid AI

Evidence

4 quotes from 3 sources, one card per model, each opening with its strongest.

How sources treat it
used 4

LFM3-1.7B

Liquid AIadopted3 sources

The final 3.5T tokens are a mid-training stage on curated data with a larger share of agentic traces and tool calls

useddata curationin LFM3-1.7B-BaseLiquid AI
3 more from this model

with a larger share of agentic traces and diverse tool calls to prepare the model for agentic post-training.

useddata curationin LFM3-1.7B-BaseLiquid AI

the last 3.5T tokens use curated, higher-quality data with a larger share of agentic traces and tool calls

useddata curationin LFM3-1.7B-BaseLiquid AI

the last 3.5T tokens use curated, higher-quality data with a larger share of agentic traces and tool calls

useddata curationin LFM3-1.7B-BaseLiquid AI
↑