Unverified50% confidenceBenchmarkExact time
On the Belebele dataset, Saidzoda-CA-EmbeddingGemma-300M improved recall@1 for the four languages by 14.4 to 26.0 percentage points, reaching 75.2% to 77.7%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/10/2026
First Seen
Valid until: 2/8/2027
Sources
Related Entities
Related Claims
UnverifiedUsing only 64 paired segments, raw Recall@1 can be raised to 88.4%–90.3%, and token cosine similarity to about 0.82.70% similarUnverifiedOn the Tajik legal retrieval task, the unified model achieved a recall@10 of 92.2, essentially on par with the starting model's 92.4, but recall@1 dropped by 4.4 percentage points.68% similarUnverifiedThe unified model's recall@1 on Russian and English dropped by 3.0 and 4.2 percentage points respectively, a common cost of multilingual fine-tuning.67% similarUnverifiedIn the composite score for memory retrieval, similarity has the largest weight, recency accounts for a small proportion, and the result is multiplied by an importance factor with a range of approximately 0.8 to 1.265% similarUnverifiedAfter introducing an explicit resolver to handle information supersession relationships, stale-memory failures dropped to zero and overall accuracy jumped from 33% to 78%.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1153280API
curl https://kongchang.com/api/v1/knowledge/claims/1153280MCP
get_claim(id=1153280)