Unverified50% confidenceFactExact time
FuseAMC's two sub-models are a residual CNN with about 198k parameters and a gradient boosting tree (GBT) cumulant classifier.
1
Sources
50%
Confidence
Long-term
Relevance
11/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedColibri supports model parameter scales ranging from 744B to 2.8T.65% similarUnverifiedVKORE-46M-NPU has a parameter scale of 46M, about one-third of GPT-2 Small (about 124 million parameters), and its significance lies in validating that the NPU training path is feasible, rather than in the quality of the model itself.65% similarUnverifiedUsing the Qwen3-27B model on the wikitext-2 dataset with fp32 reference logits as a baseline for KL divergence measurement, the median KLD dropped from 0.0023 to 0.000001, an approximately 2300-fold improvement in precision.62% similarUnverified2.8T usually refers to the total parameter scale, while 104B is more likely an indicator of activated parameters or effective compute.62% similarUnverifiedThe fine-tuned 406M model achieved 36.33 ROUGE-1 on the test set, while the authors' own 1.2B system achieved 35.41 ROUGE-1, and the two cannot be statistically distinguished.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1138037API
curl https://kongchang.com/api/v1/knowledge/claims/1138037MCP
get_claim(id=1138037)