yjoonjang commited on
Commit
dd38901
·
verified ·
1 Parent(s): 53e42a4

Update evaluation: PublicHealthQA/Belebele re-evaluated on corrected BM25 pools

Browse files

- PublicHealthQA and BelebeleRetrieval were scored against another query's BM25 candidates (the benchmark's mteb 1.x pools used per-process query ids); both are re-evaluated for every model, Belebele on its kor_Hang-kor_Hang subset.
- Tables, key numbers and the figure updated; KURE-Reranker models drawn as stars.
- Model weights are unchanged.

Files changed (2) hide show
  1. README.md +60 -60
  2. assets/pps_vs_ndcg9.png +2 -2
README.md CHANGED
@@ -25,9 +25,9 @@ tags:
25
 
26
  ## Key Characteristics
27
 
28
- - **Strong Korean reranking:** 88.49 average nDCG@10 across nine Korean benchmarks, the highest average among the evaluated models with fewer than 4B parameters and within 1.1 points of Qwen3-Reranker-4B.
29
  - **Long-document reranking:** 81.63 nDCG@10 on MultiLongDocRetrieval (MLDR), ranking second only to Qwen3-Reranker-8B (82.55) among the evaluated models.
30
- - **Efficient:** 55.1 pairs per second averaged over nine benchmarks on an NVIDIA RTX A6000 48GB, about 2.3× Qwen3-Reranker-4B and 3.6× Qwen3-Reranker-8B under the measurement conditions described below.
31
 
32
  ## Model Overview
33
 
@@ -139,31 +139,31 @@ Evaluation was conducted using the [reranker-simple-benchmark](https://github.co
139
 
140
  <img src="./assets/pps_vs_ndcg9.png" width="700" alt="Reranking performance (mean nDCG@10) vs. throughput (mean PPS)">
141
 
142
- The figure plots mean nDCG@10 against mean PPS across the nine Korean benchmarks, including MLDR. Dot color marks model size. Throughput uses each model's own input limit and its best measured batch size, so it measures neither speed at equal token lengths nor end-to-end retrieval latency. The x-axis is a non-uniform log scale that stretches 20–26 and compresses 26–50 pairs/s so that nearby points stay readable.
143
 
144
  ### Results — MTEB-ko-retrieval (9 subsets)
145
  <!-- **공식 9개 subset 을 모두 평가한 모델**의 9-subset mean NDCG@10, PPS -->
146
  | Model | Params | Mean NDCG@10 | Mean PPS |
147
  |---|---|---|---|
148
- | Qwen/Qwen3-Reranker-8B | 8.2B | 0.9030 | 15.2 |
149
- | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 0.8960 | 25.3 |
150
- | Qwen/Qwen3-Reranker-4B | 4.0B | 0.8957 | 24.3 |
151
- | **nlpai-lab/KURE-Reranker-base** | 1.7B | 0.8849 | 55.1 |
152
- | **nlpai-lab/KURE-Reranker-nano** | 149M | 0.8808 | 473.7 |
153
- | zeroentropy/zerank-2-reranker | 4.0B | 0.8695 | 29.4 |
154
- | lightonai/LightOn-rerank-PW-4B | 4.5B | 0.8664 | 15.0 |
155
- | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 0.8661 | 65.2 |
156
- | BAAI/bge-reranker-v2-m3 | 568M | 0.8586 | 404.1 |
157
- | Qwen/Qwen3-Reranker-0.6B | 596M | 0.8577 | 100.2 |
158
- | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 0.8522 | 127.3 |
159
- | nlpai-lab/LAMAR-600m | 568M | 0.8406 | 408.8 |
160
- | dragonkue/bge-reranker-v2-m3-ko | 568M | 0.8263 | 401.5 |
161
- | BAAI/bge-reranker-v2-gemma | 2.5B | 0.8186 | 58.7 |
162
- | upskyy/ko-reranker-8k | 568M | 0.8085 | 404.0 |
163
- | Dongjin-kr/ko-reranker | 560M | 0.7950 | 482.8 |
164
- | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 0.7806 | 99.6 |
165
-
166
- PPS is measured on a single NVIDIA RTX A6000 48GB per model and averaged over the nine benchmarks. For each benchmark, every model is timed on the same ~550 sampled query–document pairs (whole queries, fixed seed). Inputs are length-sorted and dynamically padded within each batch. Models run in bf16 with flash_attention_2, except `KaLM-Embedding/KaLM-Reranker-V1-Large-R2`, which uses sdpa because transformers does not support flash_attention_2 for T5Gemma2.
167
 
168
  `jinaai/jina-reranker-v3` and `jinaai/jina-reranker-v3.5` are listwise rerankers that cannot be compared under the same 8,192-token condition on MLDR, so they are excluded from the 9-subset table and their MLDR cells are left blank below.
169
 
@@ -171,49 +171,49 @@ PPS is measured on a single NVIDIA RTX A6000 48GB per model and averaged over th
171
 
172
  | Model | Params | Ko-StrategyQA | AutoRAGRetrieval | PublicHealthQA | BelebeleRetrieval | MIRACLRetrieval | MrTidyRetrieval | MultiLongDocRetrieval | SQuADKorV1Retrieval | LawIRKo |
173
  |---|---|---|---|---|---|---|---|---|---|---|
174
- | Qwen/Qwen3-Reranker-8B | 8.2B | 0.8720 | 0.9653 | 0.8899 | 0.9908 | 0.8513 | 0.8416 | 0.8255 | 0.9897 | 0.9013 |
175
- | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 0.8796 | 0.9415 | 0.8956 | 0.9936 | 0.8382 | 0.8500 | 0.7970 | 0.9863 | 0.8824 |
176
- | Qwen/Qwen3-Reranker-4B | 4.0B | 0.8733 | 0.9630 | 0.8702 | 0.9904 | 0.8559 | 0.8298 | 0.8161 | 0.9861 | 0.8766 |
177
- | **nlpai-lab/KURE-Reranker-base** | 1.7B | 0.8548 | 0.9762 | 0.8716 | 0.9880 | 0.8371 | 0.7970 | 0.8163 | 0.9891 | 0.8343 |
178
- | **nlpai-lab/KURE-Reranker-nano** | 149M | 0.8582 | 0.9741 | 0.8475 | 0.9830 | 0.8400 | 0.8011 | 0.7966 | 0.9895 | 0.8368 |
179
- | jinaai/jina-reranker-v3.5 | 597M | 0.8539 | 0.9838 | 0.8094 | 0.9733 | 0.8565 | 0.8194 | — | 0.9887 | 0.8545 |
180
- | jinaai/jina-reranker-v3 | 597M | 0.8553 | 0.9773 | 0.7960 | 0.9695 | 0.8449 | 0.8104 | — | 0.9859 | 0.8546 |
181
- | zeroentropy/zerank-2-reranker | 4.0B | 0.8712 | 0.9436 | 0.8646 | 0.9846 | 0.8003 | 0.8027 | 0.7120 | 0.9791 | 0.8669 |
182
- | lightonai/LightOn-rerank-PW-4B | 4.5B | 0.8567 | 0.9321 | 0.8693 | 0.9882 | 0.8072 | 0.8091 | 0.7609 | 0.9803 | 0.7938 |
183
- | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 0.8563 | 0.9531 | 0.8772 | 0.9778 | 0.7939 | 0.8771 | 0.6787 | 0.9681 | 0.8130 |
184
- | BAAI/bge-reranker-v2-m3 | 568M | 0.8487 | 0.9663 | 0.8475 | 0.9853 | 0.8129 | 0.8222 | 0.6690 | 0.9853 | 0.7906 |
185
- | Qwen/Qwen3-Reranker-0.6B | 596M | 0.8336 | 0.9321 | 0.8477 | 0.9779 | 0.8470 | 0.7325 | 0.7819 | 0.9804 | 0.7865 |
186
- | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 0.8536 | 0.9480 | 0.8491 | 0.9883 | 0.8182 | 0.8026 | 0.6719 | 0.9858 | 0.7527 |
187
- | nlpai-lab/LAMAR-600m | 568M | 0.8461 | 0.9591 | 0.8225 | 0.9835 | 0.8214 | 0.7975 | 0.5679 | 0.9850 | 0.7822 |
188
- | dragonkue/bge-reranker-v2-m3-ko | 568M | 0.8232 | 0.9684 | 0.8708 | 0.9769 | 0.7573 | 0.6776 | 0.7061 | 0.9846 | 0.6721 |
189
- | BAAI/bge-reranker-v2-gemma | 2.5B | 0.8614 | 0.9407 | 0.8698 | 0.9857 | 0.8362 | 0.8429 | 0.2881 | 0.9858 | 0.7572 |
190
- | upskyy/ko-reranker-8k | 568M | 0.8143 | 0.9230 | 0.8388 | 0.9291 | 0.7249 | 0.6998 | 0.5975 | 0.9718 | 0.7770 |
191
- | Dongjin-kr/ko-reranker | 560M | 0.8468 | 0.9014 | 0.7675 | 0.9759 | 0.8017 | 0.7772 | 0.3721 | 0.9785 | 0.7343 |
192
- | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 0.8329 | 0.9794 | 0.8534 | 0.9777 | 0.8449 | 0.7650 | 0.1829 | 0.9850 | 0.6042 |
193
 
194
  ### Per-dataset PPS
195
 
196
  | Model | Params | Ko-StrategyQA | AutoRAGRetrieval | PublicHealthQA | BelebeleRetrieval | MIRACLRetrieval | MrTidyRetrieval | MultiLongDocRetrieval | SQuADKorV1Retrieval | LawIRKo |
197
  |---|---|---|---|---|---|---|---|---|---|---|
198
- | Qwen/Qwen3-Reranker-8B | 8.2B | 16.5 | 7.4 | 17.6 | 19.4 | 25.0 | 26.6 | 0.7 | 10.8 | 12.4 |
199
- | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 28.6 | 12.4 | 28.5 | 33.7 | 39.6 | 43.8 | 0.8 | 18.7 | 21.4 |
200
- | Qwen/Qwen3-Reranker-4B | 4.0B | 26.2 | 11.8 | 28.2 | 31.1 | 40.1 | 42.8 | 1.1 | 17.1 | 19.9 |
201
- | **nlpai-lab/KURE-Reranker-base** | 1.7B | 59.2 | 27.3 | 64.6 | 71.1 | 89.1 | 97.1 | 2.6 | 39.1 | 45.7 |
202
- | **nlpai-lab/KURE-Reranker-nano** | 149M | 475.4 | 233.2 | 569.7 | 630.9 | 765.8 | 855.2 | 16.5 | 330.8 | 386.0 |
203
- | jinaai/jina-reranker-v3.5 | 597M | 141.0 | 38.6 | 148.6 | 174.6 | 277.3 | 295.7 | — | 68.4 | 86.7 |
204
- | jinaai/jina-reranker-v3 | 597M | 113.8 | 25.0 | 121.8 | 169.2 | 244.6 | 261.6 | — | 48.7 | 64.0 |
205
- | zeroentropy/zerank-2-reranker | 4.0B | 31.0 | 12.7 | 33.8 | 37.8 | 51.0 | 56.4 | 1.1 | 18.8 | 22.3 |
206
- | lightonai/LightOn-rerank-PW-4B | 4.5B | 16.2 | 7.3 | 18.6 | 19.5 | 23.4 | 26.1 | 0.7 | 10.7 | 12.4 |
207
- | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 70.0 | 32.9 | 76.8 | 84.1 | 104.2 | 113.6 | 3.2 | 47.0 | 54.7 |
208
- | BAAI/bge-reranker-v2-m3 | 568M | 420.7 | 195.6 | 471.3 | 545.3 | 653.2 | 724.8 | 9.3 | 271.6 | 345.5 |
209
- | Qwen/Qwen3-Reranker-0.6B | 596M | 105.9 | 49.4 | 117.9 | 130.3 | 162.6 | 177.4 | 4.3 | 70.8 | 83.6 |
210
- | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 133.7 | 54.4 | 146.9 | 163.0 | 220.0 | 243.1 | 3.9 | 84.9 | 95.9 |
211
- | nlpai-lab/LAMAR-600m | 568M | 418.0 | 196.5 | 480.5 | 557.2 | 656.7 | 742.5 | 9.2 | 275.0 | 343.8 |
212
- | dragonkue/bge-reranker-v2-m3-ko | 568M | 416.9 | 195.1 | 467.2 | 540.7 | 653.4 | 716.7 | 9.2 | 271.9 | 342.7 |
213
- | BAAI/bge-reranker-v2-gemma | 2.5B | 64.7 | 26.1 | 67.5 | 75.8 | 99.8 | 106.8 | 2.5 | 40.0 | 44.7 |
214
- | upskyy/ko-reranker-8k | 568M | 411.8 | 195.6 | 474.3 | 551.4 | 649.3 | 726.0 | 9.2 | 275.1 | 342.8 |
215
- | Dongjin-kr/ko-reranker | 560M | 520.8 | 258.0 | 510.1 | 554.8 | 747.6 | 765.2 | 272.7 | 334.4 | 381.2 |
216
- | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 104.6 | 49.4 | 117.8 | 129.8 | 160.5 | 175.6 | 4.3 | 72.1 | 82.6 |
217
 
218
  Batch size starts at 8 and doubles until an out-of-memory error. Samples are repeated to fill complete batches. After warmup, three full passes are timed with CUDA events. Throughput is the total number of processed pairs divided by the accumulated GPU forward time. The highest-throughput successful batch is reported. Tokenization, data loading, and CPU preprocessing are excluded. The batch-search approach is informed by the [Ettin reranker speed benchmark](https://huggingface.co/blog/ettin-reranker#speed).
219
 
 
25
 
26
  ## Key Characteristics
27
 
28
+ - **Strong Korean reranking:** 88.00 average nDCG@10 across nine Korean benchmarks, the highest average among the evaluated models with fewer than 4B parameters and within 1.2 points of Qwen3-Reranker-4B.
29
  - **Long-document reranking:** 81.63 nDCG@10 on MultiLongDocRetrieval (MLDR), ranking second only to Qwen3-Reranker-8B (82.55) among the evaluated models.
30
+ - **Efficient:** 55.0 pairs per second averaged over nine benchmarks on an NVIDIA RTX A6000 48GB, about 2.3× Qwen3-Reranker-4B and 3.7× Qwen3-Reranker-8B under the measurement conditions described below.
31
 
32
  ## Model Overview
33
 
 
139
 
140
  <img src="./assets/pps_vs_ndcg9.png" width="700" alt="Reranking performance (mean nDCG@10) vs. throughput (mean PPS)">
141
 
142
+ The figure plots mean nDCG@10 against mean PPS across the nine Korean benchmarks, including MLDR. Dot color marks model size; stars mark the KURE-Reranker models. Throughput uses each model's own input limit and its best measured batch size, so it measures neither speed at equal token lengths nor end-to-end retrieval latency. The x-axis is a non-uniform log scale that stretches 20–26 and compresses 26–50 pairs/s so that nearby points stay readable.
143
 
144
  ### Results — MTEB-ko-retrieval (9 subsets)
145
  <!-- **공식 9개 subset 을 모두 평가한 모델**의 9-subset mean NDCG@10, PPS -->
146
  | Model | Params | Mean NDCG@10 | Mean PPS |
147
  |---|---|---|---|
148
+ | Qwen/Qwen3-Reranker-8B | 8.2B | 0.8990 | 15.0 |
149
+ | Qwen/Qwen3-Reranker-4B | 4.0B | 0.8916 | 24.3 |
150
+ | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 0.8910 | 25.3 |
151
+ | **nlpai-lab/KURE-Reranker-base** | 1.7B | 0.8800 | 55.0 |
152
+ | **nlpai-lab/KURE-Reranker-nano** | 149M | 0.8757 | 482.4 |
153
+ | zeroentropy/zerank-2-reranker | 4.0B | 0.8652 | 29.3 |
154
+ | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 0.8607 | 64.8 |
155
+ | lightonai/LightOn-rerank-PW-4B | 4.5B | 0.8605 | 15.1 |
156
+ | Qwen/Qwen3-Reranker-0.6B | 596M | 0.8528 | 101.0 |
157
+ | BAAI/bge-reranker-v2-m3 | 568M | 0.8518 | 412.6 |
158
+ | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 0.8468 | 128.2 |
159
+ | nlpai-lab/LAMAR-600m | 568M | 0.8375 | 415.0 |
160
+ | dragonkue/bge-reranker-v2-m3-ko | 568M | 0.8199 | 409.1 |
161
+ | BAAI/bge-reranker-v2-gemma | 2.5B | 0.8099 | 58.8 |
162
+ | upskyy/ko-reranker-8k | 568M | 0.8054 | 409.8 |
163
+ | Dongjin-kr/ko-reranker | 560M | 0.7861 | 488.0 |
164
+ | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 0.7760 | 99.7 |
165
+
166
+ PPS is measured on a single NVIDIA RTX A6000 48GB per model and averaged over the nine benchmarks. For each benchmark, every model is timed on the same ~550 sampled query–document pairs (whole queries, fixed seed). Inputs are length-sorted and dynamically padded within each batch. Models run in bf16 with flash_attention_2, except `KaLM-Embedding/KaLM-Reranker-V1-Large-R2`, which uses sdpa because transformers does not support flash_attention_2 for T5Gemma2. Each query is reranked over its gold documents plus, as negatives, the BM25 top-50 minus the golds; BelebeleRetrieval uses its kor_Hang-kor_Hang subset.
167
 
168
  `jinaai/jina-reranker-v3` and `jinaai/jina-reranker-v3.5` are listwise rerankers that cannot be compared under the same 8,192-token condition on MLDR, so they are excluded from the 9-subset table and their MLDR cells are left blank below.
169
 
 
171
 
172
  | Model | Params | Ko-StrategyQA | AutoRAGRetrieval | PublicHealthQA | BelebeleRetrieval | MIRACLRetrieval | MrTidyRetrieval | MultiLongDocRetrieval | SQuADKorV1Retrieval | LawIRKo |
173
  |---|---|---|---|---|---|---|---|---|---|---|
174
+ | Qwen/Qwen3-Reranker-8B | 8.2B | 0.8720 | 0.9653 | 0.8654 | 0.9785 | 0.8513 | 0.8416 | 0.8255 | 0.9897 | 0.9013 |
175
+ | Qwen/Qwen3-Reranker-4B | 4.0B | 0.8733 | 0.9630 | 0.8442 | 0.9794 | 0.8559 | 0.8298 | 0.8161 | 0.9861 | 0.8766 |
176
+ | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 0.8796 | 0.9415 | 0.8633 | 0.9809 | 0.8382 | 0.8500 | 0.7970 | 0.9863 | 0.8824 |
177
+ | **nlpai-lab/KURE-Reranker-base** | 1.7B | 0.8548 | 0.9762 | 0.8381 | 0.9767 | 0.8371 | 0.7970 | 0.8163 | 0.9891 | 0.8343 |
178
+ | **nlpai-lab/KURE-Reranker-nano** | 149M | 0.8582 | 0.9741 | 0.8078 | 0.9768 | 0.8400 | 0.8011 | 0.7966 | 0.9895 | 0.8368 |
179
+ | jinaai/jina-reranker-v3.5 | 597M | 0.8539 | 0.9838 | 0.8337 | 0.9659 | 0.8565 | 0.8194 | — | 0.9887 | 0.8545 |
180
+ | jinaai/jina-reranker-v3 | 597M | 0.8553 | 0.9773 | 0.7762 | 0.9606 | 0.8449 | 0.8104 | — | 0.9859 | 0.8546 |
181
+ | zeroentropy/zerank-2-reranker | 4.0B | 0.8712 | 0.9436 | 0.8377 | 0.9731 | 0.8003 | 0.8027 | 0.7120 | 0.9791 | 0.8669 |
182
+ | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 0.8563 | 0.9531 | 0.8407 | 0.9659 | 0.7939 | 0.8771 | 0.6787 | 0.9681 | 0.8130 |
183
+ | lightonai/LightOn-rerank-PW-4B | 4.5B | 0.8567 | 0.9321 | 0.8312 | 0.9735 | 0.8072 | 0.8091 | 0.7609 | 0.9803 | 0.7938 |
184
+ | Qwen/Qwen3-Reranker-0.6B | 596M | 0.8336 | 0.9321 | 0.8092 | 0.9721 | 0.8470 | 0.7325 | 0.7819 | 0.9804 | 0.7865 |
185
+ | BAAI/bge-reranker-v2-m3 | 568M | 0.8487 | 0.9663 | 0.7982 | 0.9725 | 0.8129 | 0.8222 | 0.6690 | 0.9853 | 0.7906 |
186
+ | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 0.8536 | 0.9480 | 0.8130 | 0.9755 | 0.8182 | 0.8026 | 0.6719 | 0.9858 | 0.7527 |
187
+ | nlpai-lab/LAMAR-600m | 568M | 0.8461 | 0.9591 | 0.8044 | 0.9740 | 0.8214 | 0.7975 | 0.5679 | 0.9850 | 0.7822 |
188
+ | dragonkue/bge-reranker-v2-m3-ko | 568M | 0.8232 | 0.9684 | 0.8155 | 0.9744 | 0.7573 | 0.6776 | 0.7061 | 0.9846 | 0.6721 |
189
+ | BAAI/bge-reranker-v2-gemma | 2.5B | 0.8614 | 0.9407 | 0.8069 | 0.9696 | 0.8362 | 0.8429 | 0.2881 | 0.9858 | 0.7572 |
190
+ | upskyy/ko-reranker-8k | 568M | 0.8143 | 0.9230 | 0.7792 | 0.9609 | 0.7249 | 0.6998 | 0.5975 | 0.9718 | 0.7770 |
191
+ | Dongjin-kr/ko-reranker | 560M | 0.8468 | 0.9014 | 0.6976 | 0.9655 | 0.8017 | 0.7772 | 0.3721 | 0.9785 | 0.7343 |
192
+ | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 0.8329 | 0.9794 | 0.8211 | 0.9689 | 0.8449 | 0.7650 | 0.1829 | 0.9850 | 0.6042 |
193
 
194
  ### Per-dataset PPS
195
 
196
  | Model | Params | Ko-StrategyQA | AutoRAGRetrieval | PublicHealthQA | BelebeleRetrieval | MIRACLRetrieval | MrTidyRetrieval | MultiLongDocRetrieval | SQuADKorV1Retrieval | LawIRKo |
197
  |---|---|---|---|---|---|---|---|---|---|---|
198
+ | Qwen/Qwen3-Reranker-8B | 8.2B | 16.5 | 7.4 | 15.3 | 20.5 | 25.0 | 26.6 | 0.7 | 10.8 | 12.4 |
199
+ | Qwen/Qwen3-Reranker-4B | 4.0B | 26.2 | 11.8 | 25.6 | 34.5 | 40.1 | 42.8 | 1.1 | 17.1 | 19.9 |
200
+ | KaLM-Embedding/KaLM-Reranker-V1-Large-R2 | 7.5B | 28.6 | 12.4 | 25.4 | 36.9 | 39.6 | 43.8 | 0.8 | 18.7 | 21.4 |
201
+ | **nlpai-lab/KURE-Reranker-base** | 1.7B | 59.2 | 27.3 | 58.0 | 77.0 | 89.1 | 97.1 | 2.6 | 39.1 | 45.7 |
202
+ | **nlpai-lab/KURE-Reranker-nano** | 149M | 475.4 | 233.2 | 529.0 | 750.0 | 765.8 | 855.2 | 16.5 | 330.8 | 386.0 |
203
+ | jinaai/jina-reranker-v3.5 | 597M | 141.0 | 38.6 | 123.2 | 197.5 | 277.3 | 295.7 | — | 68.4 | 86.7 |
204
+ | jinaai/jina-reranker-v3 | 597M | 113.8 | 25.0 | 95.7 | 166.3 | 244.6 | 261.6 | — | 48.7 | 64.0 |
205
+ | zeroentropy/zerank-2-reranker | 4.0B | 31.0 | 12.7 | 29.5 | 41.4 | 51.0 | 56.4 | 1.1 | 18.8 | 22.3 |
206
+ | mixedbread-ai/mxbai-rerank-large-v2 | 1.5B | 70.0 | 32.9 | 68.0 | 89.8 | 104.2 | 113.6 | 3.2 | 47.0 | 54.7 |
207
+ | lightonai/LightOn-rerank-PW-4B | 4.5B | 16.2 | 7.3 | 17.0 | 22.2 | 23.4 | 26.1 | 0.7 | 10.7 | 12.4 |
208
+ | Qwen/Qwen3-Reranker-0.6B | 596M | 105.9 | 49.4 | 109.5 | 145.8 | 162.6 | 177.4 | 4.3 | 70.8 | 83.6 |
209
+ | BAAI/bge-reranker-v2-m3 | 568M | 420.7 | 195.6 | 440.9 | 651.8 | 653.2 | 724.8 | 9.3 | 271.6 | 345.5 |
210
+ | nvidia/llama-nemotron-rerank-1b-v2 | 1.2B | 133.7 | 54.4 | 131.6 | 186.3 | 220.0 | 243.1 | 3.9 | 84.9 | 95.9 |
211
+ | nlpai-lab/LAMAR-600m | 568M | 418.0 | 196.5 | 442.5 | 651.1 | 656.7 | 742.5 | 9.2 | 275.0 | 343.8 |
212
+ | dragonkue/bge-reranker-v2-m3-ko | 568M | 416.9 | 195.1 | 434.4 | 641.3 | 653.4 | 716.7 | 9.2 | 271.9 | 342.7 |
213
+ | BAAI/bge-reranker-v2-gemma | 2.5B | 64.7 | 26.1 | 60.5 | 83.8 | 99.8 | 106.8 | 2.5 | 40.0 | 44.7 |
214
+ | upskyy/ko-reranker-8k | 568M | 411.8 | 195.6 | 436.2 | 641.7 | 649.3 | 726.0 | 9.2 | 275.1 | 342.8 |
215
+ | Dongjin-kr/ko-reranker | 560M | 520.8 | 258.0 | 468.3 | 644.1 | 747.6 | 765.2 | 272.7 | 334.4 | 381.2 |
216
+ | telepix/PIXIE-Spell-Reranker-Preview-0.6B | 596M | 104.6 | 49.4 | 106.3 | 141.5 | 160.5 | 175.6 | 4.3 | 72.1 | 82.6 |
217
 
218
  Batch size starts at 8 and doubles until an out-of-memory error. Samples are repeated to fill complete batches. After warmup, three full passes are timed with CUDA events. Throughput is the total number of processed pairs divided by the accumulated GPU forward time. The highest-throughput successful batch is reported. Tokenization, data loading, and CPU preprocessing are excluded. The batch-search approach is informed by the [Ettin reranker speed benchmark](https://huggingface.co/blog/ettin-reranker#speed).
219
 
assets/pps_vs_ndcg9.png CHANGED

Git LFS Details

  • SHA256: 8ecd1b0fb6e76331af76672d4ccf542f3395f0e57ef8ce85aae3abb36af76b47
  • Pointer size: 131 Bytes
  • Size of remote file: 144 kB

Git LFS Details

  • SHA256: a6444fdadc21793f08038938e0117ba2c724b86aa128197a50196fb36abcde6b
  • Pointer size: 131 Bytes
  • Size of remote file: 156 kB