Hugging Face's logo

cuadron11
/
jina-reranker-v2-base-multilingual-binary

Text Ranking
sentence-transformers
Safetensors
cross-encoder
reranker
Generated from Trainer
dataset_size:342614
loss:BinaryCrossEntropyLoss
custom_code
Eval Results (legacy)
Model card Files
xet
 Community
cuadron11 commited on
Commit
47a5133
·
verified ·
1 Parent(s): f731e93

Add new CrossEncoder model

Browse files
Files changed (2) hide show
  1. README.md +76 -19
  2. model.safetensors +1 -1
README.md CHANGED
@@ -4,7 +4,7 @@ tags:
4
  - cross-encoder
5
  - reranker
6
  - generated_from_trainer
7
- - dataset_size:40529
8
  - loss:BinaryCrossEntropyLoss
9
  base_model: jinaai/jina-reranker-v2-base-multilingual
10
  pipeline_tag: text-ranking
@@ -24,13 +24,13 @@ model-index:
24
  type: jina-reranker-v2-base-multilingual-binary
25
  metrics:
26
  - type: map
27
- value: 0.003
28
  name: Map
29
  - type: mrr@10
30
- value: 0.0032
31
  name: Mrr@10
32
  - type: ndcg@10
33
- value: 0.0032
34
  name: Ndcg@10
35
  ---
36
 
@@ -139,9 +139,9 @@ You can finetune this model on your own dataset.
139
 
140
  | Metric | Value |
141
  |:------------|:---------------------|
142
- | map | 0.0030 (+0.0026) |
143
- | mrr@10 | 0.0032 (+0.0029) |
144
- | **ndcg@10** | **0.0032 (+0.0026)** |
145
 
146
  <!--
147
  ## Bias, Risks and Limitations
@@ -161,13 +161,13 @@ You can finetune this model on your own dataset.
161
 
162
  #### Unnamed Dataset
163
 
164
- * Size: 40,529 training samples
165
  * Columns: <code>query</code>, <code>answer</code>, and <code>label</code>
166
  * Approximate statistics based on the first 1000 samples:
167
- | | query | answer | label |
168
- |:--------|:------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------|:------------------------------------------------|
169
- | type | string | string | int |
170
- | details | <ul><li>min: 32 characters</li><li>mean: 98.32 characters</li><li>max: 207 characters</li></ul> | <ul><li>min: 583 characters</li><li>mean: 763.83 characters</li><li>max: 978 characters</li></ul> | <ul><li>0: ~82.90%</li><li>1: ~17.10%</li></ul> |
171
  * Samples:
172
  | query | answer | label |
173
  |:----------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------|
@@ -317,13 +317,70 @@ You can finetune this model on your own dataset.
317
  ### Training Logs
318
  | Epoch | Step | Training Loss | jina-reranker-v2-base-multilingual-binary_ndcg@10 |
319
  |:----------:|:-------:|:-------------:|:-------------------------------------------------:|
320
- | **0.3946** | **500** | **0.0915** | **0.0034 (+0.0028)** |
321
- | 0.7893 | 1000 | 0.0028 | 0.0034 (+0.0028) |
322
- | 1.1839 | 1500 | 0.0123 | 0.0032 (+0.0026) |
323
- | 1.5785 | 2000 | 0.0086 | 0.0032 (+0.0026) |
324
- | 1.9732 | 2500 | 0.0013 | 0.0032 (+0.0026) |
325
- | 2.3678 | 3000 | 0.0121 | 0.0032 (+0.0026) |
326
- | 2.7624 | 3500 | 0.0082 | 0.0032 (+0.0026) |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
327
 
328
  * The bold row denotes the saved checkpoint.
329
 
 
4
  - cross-encoder
5
  - reranker
6
  - generated_from_trainer
7
+ - dataset_size:342614
8
  - loss:BinaryCrossEntropyLoss
9
  base_model: jinaai/jina-reranker-v2-base-multilingual
10
  pipeline_tag: text-ranking
 
24
  type: jina-reranker-v2-base-multilingual-binary
25
  metrics:
26
  - type: map
27
+ value: 0.0034
28
  name: Map
29
  - type: mrr@10
30
+ value: 0.0034
31
  name: Mrr@10
32
  - type: ndcg@10
33
+ value: 0.0034
34
  name: Ndcg@10
35
  ---
36
 
 
139
 
140
  | Metric | Value |
141
  |:------------|:---------------------|
142
+ | map | 0.0034 (+0.0031) |
143
+ | mrr@10 | 0.0034 (+0.0032) |
144
+ | **ndcg@10** | **0.0034 (+0.0028)** |
145
 
146
  <!--
147
  ## Bias, Risks and Limitations
 
161
 
162
  #### Unnamed Dataset
163
 
164
+ * Size: 342,614 training samples
165
  * Columns: <code>query</code>, <code>answer</code>, and <code>label</code>
166
  * Approximate statistics based on the first 1000 samples:
167
+ | | query | answer | label |
168
+ |:--------|:-------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------|:-----------------------------------------------|
169
+ | type | string | string | int |
170
+ | details | <ul><li>min: 69 characters</li><li>mean: 104.56 characters</li><li>max: 207 characters</li></ul> | <ul><li>min: 672 characters</li><li>mean: 760.52 characters</li><li>max: 873 characters</li></ul> | <ul><li>0: ~98.00%</li><li>1: ~2.00%</li></ul> |
171
  * Samples:
172
  | query | answer | label |
173
  |:----------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------|
 
317
  ### Training Logs
318
  | Epoch | Step | Training Loss | jina-reranker-v2-base-multilingual-binary_ndcg@10 |
319
  |:----------:|:-------:|:-------------:|:-------------------------------------------------:|
320
+ | **0.0467** | **500** | **0.1385** | **0.0034 (+0.0028)** |
321
+ | 0.0934 | 1000 | 0.0578 | 0.0034 (+0.0028) |
322
+ | 0.1401 | 1500 | 0.0305 | 0.0034 (+0.0028) |
323
+ | 0.1868 | 2000 | 0.0273 | 0.0034 (+0.0028) |
324
+ | 0.2335 | 2500 | 0.0291 | 0.0034 (+0.0028) |
325
+ | 0.2802 | 3000 | 0.0115 | 0.0034 (+0.0028) |
326
+ | 0.3269 | 3500 | 0.0065 | 0.0034 (+0.0028) |
327
+ | 0.3736 | 4000 | 0.0044 | 0.0034 (+0.0028) |
328
+ | 0.4203 | 4500 | 0.0049 | 0.0034 (+0.0028) |
329
+ | 0.4670 | 5000 | 0.0047 | 0.0034 (+0.0028) |
330
+ | 0.5137 | 5500 | 0.0048 | 0.0034 (+0.0028) |
331
+ | 0.5604 | 6000 | 0.0055 | 0.0034 (+0.0028) |
332
+ | 0.6071 | 6500 | 0.0047 | 0.0034 (+0.0028) |
333
+ | 0.6538 | 7000 | 0.0053 | 0.0034 (+0.0028) |
334
+ | 0.7005 | 7500 | 0.0042 | 0.0034 (+0.0028) |
335
+ | 0.7472 | 8000 | 0.0049 | 0.0034 (+0.0028) |
336
+ | 0.7939 | 8500 | 0.0056 | 0.0034 (+0.0028) |
337
+ | 0.8406 | 9000 | 0.0034 | 0.0034 (+0.0028) |
338
+ | 0.8873 | 9500 | 0.0026 | 0.0034 (+0.0028) |
339
+ | 0.9340 | 10000 | 0.0055 | 0.0034 (+0.0028) |
340
+ | 0.9807 | 10500 | 0.0039 | 0.0034 (+0.0028) |
341
+ | 1.0274 | 11000 | 0.0043 | 0.0034 (+0.0028) |
342
+ | 1.0741 | 11500 | 0.0038 | 0.0034 (+0.0028) |
343
+ | 1.1208 | 12000 | 0.0056 | 0.0034 (+0.0028) |
344
+ | 1.1675 | 12500 | 0.0041 | 0.0034 (+0.0028) |
345
+ | 1.2142 | 13000 | 0.0026 | 0.0034 (+0.0028) |
346
+ | 1.2609 | 13500 | 0.0042 | 0.0034 (+0.0028) |
347
+ | 1.3076 | 14000 | 0.0028 | 0.0034 (+0.0028) |
348
+ | 1.3543 | 14500 | 0.0049 | 0.0034 (+0.0028) |
349
+ | 1.4010 | 15000 | 0.004 | 0.0034 (+0.0028) |
350
+ | 1.4477 | 15500 | 0.0037 | 0.0034 (+0.0028) |
351
+ | 1.4943 | 16000 | 0.0029 | 0.0034 (+0.0028) |
352
+ | 1.5410 | 16500 | 0.0038 | 0.0034 (+0.0028) |
353
+ | 1.5877 | 17000 | 0.0049 | 0.0034 (+0.0028) |
354
+ | 1.6344 | 17500 | 0.0041 | 0.0034 (+0.0028) |
355
+ | 1.6811 | 18000 | 0.0045 | 0.0034 (+0.0028) |
356
+ | 1.7278 | 18500 | 0.0038 | 0.0034 (+0.0028) |
357
+ | 1.7745 | 19000 | 0.0037 | 0.0034 (+0.0028) |
358
+ | 1.8212 | 19500 | 0.0033 | 0.0034 (+0.0028) |
359
+ | 1.8679 | 20000 | 0.0036 | 0.0034 (+0.0028) |
360
+ | 1.9146 | 20500 | 0.0035 | 0.0034 (+0.0028) |
361
+ | 1.9613 | 21000 | 0.0048 | 0.0034 (+0.0028) |
362
+ | 2.0080 | 21500 | 0.0059 | 0.0034 (+0.0028) |
363
+ | 2.0547 | 22000 | 0.004 | 0.0034 (+0.0028) |
364
+ | 2.1014 | 22500 | 0.0037 | 0.0034 (+0.0028) |
365
+ | 2.1481 | 23000 | 0.0039 | 0.0034 (+0.0028) |
366
+ | 2.1948 | 23500 | 0.0048 | 0.0034 (+0.0028) |
367
+ | 2.2415 | 24000 | 0.0034 | 0.0034 (+0.0028) |
368
+ | 2.2882 | 24500 | 0.0036 | 0.0034 (+0.0028) |
369
+ | 2.3349 | 25000 | 0.0039 | 0.0034 (+0.0028) |
370
+ | 2.3816 | 25500 | 0.0043 | 0.0034 (+0.0028) |
371
+ | 2.4283 | 26000 | 0.0039 | 0.0034 (+0.0028) |
372
+ | 2.4750 | 26500 | 0.0052 | 0.0034 (+0.0028) |
373
+ | 2.5217 | 27000 | 0.0054 | 0.0034 (+0.0028) |
374
+ | 2.5684 | 27500 | 0.0031 | 0.0034 (+0.0028) |
375
+ | 2.6151 | 28000 | 0.0036 | 0.0034 (+0.0028) |
376
+ | 2.6618 | 28500 | 0.0046 | 0.0034 (+0.0028) |
377
+ | 2.7085 | 29000 | 0.0031 | 0.0034 (+0.0028) |
378
+ | 2.7552 | 29500 | 0.0034 | 0.0034 (+0.0028) |
379
+ | 2.8019 | 30000 | 0.0033 | 0.0034 (+0.0028) |
380
+ | 2.8486 | 30500 | 0.004 | 0.0034 (+0.0028) |
381
+ | 2.8953 | 31000 | 0.0042 | 0.0034 (+0.0028) |
382
+ | 2.9420 | 31500 | 0.0035 | 0.0034 (+0.0028) |
383
+ | 2.9887 | 32000 | 0.0044 | 0.0034 (+0.0028) |
384
 
385
  * The bold row denotes the saved checkpoint.
386
 
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bd2d20a96b6cb844459372b772d35767702b969009f0c0f608f4f866ae7d3519
3
  size 556892306
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef670b1f14e055a87f0d599f39ce8fb29f190929295205bd9698dcd9627fa522
3
  size 556892306