Unverified50% confidenceBenchmarkExact time
On Llama-3.2-3B, fine-tuning a brand-new dense model with the 5% of data selected by CACTUS reached 49.44% accuracy on BIG-Bench Hard; selecting data directly with the dense model gave 49.54%; data selected by the ALPS method could only reach 47.22%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedWhen doing 50% pruning on Qwen3-4B, CACTUS reduced the gap between C4 perplexity and the dense model by 45%.64% similarUnverifiedThe locally run Gemma 26B achieved an accuracy of only 49.2% across 453 scored predictions, while its paired always-predict-up baseline had an accuracy of 53.4%.64% similarUnverifiedTaking GPT-3's attention layers as an example, the original weight matrix is approximately 12,288×12,288 (about 150 million parameters), while a LoRA decomposition with rank r=8 requires training only about 200,000 parameters, a compression ratio exceeding 700×.63% similarUnverifiedTEMPEST outperforms the best classical model by 17.9 percentage points and the strongest triplet-loss baseline by 58.4 percentage points.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1022899API
curl https://kongchang.com/api/v1/knowledge/claims/1022899MCP
get_claim(id=1022899)