Unverified50% confidenceFactExact time
The final adaptation stage updates only 88.66 million parameters, and the training data consists of 8,192 curated views.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/4/2026
First Seen
Valid until: 2/2/2027
Sources
Related Entities
Related Claims
UnverifiedThe total of 2167 multiple-choice questions and 999 experimental records is still insufficient for training a large-scale generative model, and is better suited as an evaluation benchmark or fine-tuning seed.70% similarUnverifiedSample efficiency refers to the number of training samples a model needs to reach a certain performance level; a geometric-mean 1.8x advantage means that a model using Assur features needs about 1/1.8 as many samples as the baseline to reach the same prediction error.67% similarUnverifiedWhen sampling eight candidates without constraints from the model, 10.2% of inputs cannot produce any valid output under limited training data, and this proportion drops to 0.4% when using the full training data67% similarUnverifiedThe learning curve shows that vertical writing layouts cannot be solved by adding data: 8.7% with zero training samples, 84.7% with 800 sheets, and 89.3% for both 2400 and 3000 sheets.66% similarUnverifiedIn a retrospective evaluation covering three deployment cycles and a total of 1,122 recorded coaching dialogues, conditional strategy packages raised the correctness rate from 57.9% to 82.0% (compared with atomic facts).65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1134090API
curl https://kongchang.com/api/v1/knowledge/claims/1134090MCP
get_claim(id=1134090)