Unverified50% confidenceBenchmarkExact time
In the full Spider-ACL test, schemagate answered 1.8% of the forbidden questions, but the records of unauthorized data reads were 0/8,128.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedBy re-tallying the results files, readers found a row of data that the authors had skipped: the maintainer file lists 9 of 212 tests judged WEAK (4%), while the agent file lists 79 of 591 (13%).60% similarUnverifiedOn a test set of 600 synthetic invoices, the QLoRA fine-tuned model achieved a full-invoice exact rate of 83.0%, an auto-approval rate of 86.8%, and 0 out of 461 validation-coverage field errors in approved invoices (upper bound 0.8%).59% similarUnverifiedWhen the open-source small model gemma2:2b was used as the checker, it accepted 8 of the 28 wrong answers (6 wrong party and 2 wrong action), while rejecting 10 of the 14 correct answers.59% similarUnverifiedWithout validation, 12.8% of the fine-tuned model's invoices have validation-coverage field errors.58% similarUnverifiedIn that blind review experiment, Grok gave 40/40 OK on the first pass but all rationale fields were empty, and it failed to detect the objective error where the source text wrote 1 but the output multiplicity was 0, so that review was marked invalid.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1053851API
curl https://kongchang.com/api/v1/knowledge/claims/1053851MCP
get_claim(id=1053851)