Unverified50% confidenceEventExact time
The research team reran the official 50-billion-token mid-training phase of OLMo 2 1B, replacing selected training data chunks with documents from 7,814 matched document pairs, where for each pair a coin flip decided which document was inserted into training.
1
Sources
50%
Confidence
Long-term
Relevance
11/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe training dataset contains 4096 numerical Richards equation trajectories, split into training, validation, and evaluation sets in a 2867/614/615 ratio.73% similarUnverifiedThe OmniGround team built a 580K-sample training dataset by reconstructing temporal question answering from dense audio-visual annotations and organizing general omni-modal instruction data.72% similarUnverifiedMiniCPM5-2B open-sourced its training data, including approximately 500,000 agent samples and over 80,000 reinforcement learning samples.71% similarUnverifiedThe training incorporated behavioral metrics such as merge rate, code review time, and repository size, ultimately generating more than 1 million step-to-step state transition data entries.71% similarUnverifiedThe algorithm outputs the last training iterate rather than a historical average, and the samples from both the step-size selection phase and the formal training phase are counted toward the total sample budget.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1125006API
curl https://kongchang.com/api/v1/knowledge/claims/1125006MCP
get_claim(id=1125006)