is cohere 1m dataset changed from last december

[Rohith-Scalers]

i am getting different recall values on the same machine and same setup for search test on hnsw index , i am wondering if there is any update in dataset which is effecting this

[xiaofan-luan]

we didn't change it for quite well. but hnsw has some randomness so if the recall is fluctuate under 0.1% this might be reasonable.