Direction, evidence grade, and study type are AI-generated labels (gpt-5-mini), not human-verified. Syntheses are LLM-written. "Tensions" are machine-detected candidates, not confirmed contradictions. A research-acceleration tool, not peer review.
How this is built →
1Distinct papers
3Unique collaborators
1/1Semantic Scholar citation coverage
Publication span: 2026. Corpus fetch span: 2026.
Identity provenance
Provider IDs
- Semantic Scholar:
2292239822
ORCID evidence
No valid ORCID is stored.
Observed aliases (1)
- Guanghui Wang (semantic scholar, provider refresh)
Topics and outcomes in this view
Assessment themes
- Adoption: 1 paper
- Productivity: 1 paper
Claim outcomes
- Other: 1 paper
- Adoption Rate: 1 paper
- Organizational Efficiency: 1 paper
- Output Quality: 1 paper
Papers in the Semantic Scholar view
Latest stored Semantic Scholar author observations only. Citation counts below are from the same provider and are not combined with other services.
Scroll the table horizontally to see every column.
| Paper | Author evidence | Date | Provider citations |
|---|---|---|---|
| A simple SVD-based factorization of Transformer keys cuts KV cache size by up to 75% with only ~2% perplexity cost and, at 7B scale, saves ~25 GB per user for 128K contexts—enabling roughly 60% more concurrent users; training-from-scratch at r=d/4 matches full-attention perplexity while reducing parameters and speeding training.arxiv | Guanghui Wang provider id |
2026-02-16 | 2 |
Citation observation summary
Semantic Scholar supplied counts for 1 of 1 papers in this view; 0 are missing. The observed paper counts sum to 2 cumulative citations. This is a coverage summary, not an author score or h-index.