Unverified50% confidenceFactExact time
The original Whisper research described training on 680,000 hours of multilingual and multitask supervised data and reported strong zero-shot transfer across benchmarks.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe 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.72% similarUnverifiedThe capabilities and behavior of large language models are deeply dependent on training data.69% similarUnverifiedThe pretraining data scale of mainstream large language models often exceeds 1T (trillion) tokens, and lightweight fine-tuning (SFT) also commonly requires billions of tokens.67% similarUnverifiedMiniCPM5-2B open-sourced its training data, including approximately 500,000 agent samples and over 80,000 reinforcement learning samples.67% similarUnverifiedTraining a large language model often requires thousands or even tens of thousands of accelerator cards working in concert.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1072890API
curl https://kongchang.com/api/v1/knowledge/claims/1072890MCP
get_claim(id=1072890)