These are the results for models other than `bert-base-uncased`. We run experiments for 

- datasets: `agnews_business` and `amazon_agri`

- active learning trategies: `entropy`

- pool filtering strategies: `anchoral`, `randomsubset`, `seals`

- models: `albert-base-v2`, `deberta-v3-base`, `t5-base`, `gpt2-base`, `bert-tiny (google/bert_uncased_L-2_H-128_A-2)`