Hugging Face's logo

cometadata
/
jina-reranker-v2-multilingual-affiliations-large

Text Ranking
sentence-transformers
Safetensors
multilingual
cross-encoder
reranker
Generated from Trainer
dataset_size:170000
loss:BinaryCrossEntropyLoss
custom_code
Eval Results (legacy)
Model card Files
xet
 Community
adambuttrick commited on
Commit
f57e868
·
verified ·
1 Parent(s): 239549d

Add new CrossEncoder model

Browse files
Files changed (2) hide show
  1. README.md +18 -15
  2. model.safetensors +1 -1
README.md CHANGED
@@ -27,13 +27,13 @@ model-index:
27
  type: affiliation-val
28
  metrics:
29
  - type: map
30
- value: 0.9753
31
  name: Map
32
  - type: mrr@10
33
- value: 0.9753
34
  name: Mrr@10
35
  - type: ndcg@10
36
- value: 0.9854
37
  name: Ndcg@10
38
  ---
39
 
@@ -142,9 +142,9 @@ You can finetune this model on your own dataset.
142
 
143
  | Metric | Value |
144
  |:------------|:---------------------|
145
- | map | 0.9753 (-0.0247) |
146
- | mrr@10 | 0.9753 (-0.0247) |
147
- | **ndcg@10** | **0.9854 (-0.0146)** |
148
 
149
  <!--
150
  ## Bias, Risks and Limitations
@@ -217,7 +217,6 @@ You can finetune this model on your own dataset.
217
  - `per_device_train_batch_size`: 128
218
  - `per_device_eval_batch_size`: 128
219
  - `learning_rate`: 3e-05
220
- - `num_train_epochs`: 1
221
  - `warmup_ratio`: 0.1
222
  - `bf16`: True
223
  - `load_best_model_at_end`: True
@@ -244,7 +243,7 @@ You can finetune this model on your own dataset.
244
  - `adam_beta2`: 0.999
245
  - `adam_epsilon`: 1e-08
246
  - `max_grad_norm`: 1.0
247
- - `num_train_epochs`: 1
248
  - `max_steps`: -1
249
  - `lr_scheduler_type`: linear
250
  - `lr_scheduler_kwargs`: {}
@@ -349,13 +348,17 @@ You can finetune this model on your own dataset.
349
  </details>
350
 
351
  ### Training Logs
352
- | Epoch | Step | Training Loss | affiliation-val_ndcg@10 |
353
- |:------:|:----:|:-------------:|:-----------------------:|
354
- | -1 | -1 | - | 0.9273 (-0.0727) |
355
- | 0.0075 | 1 | 0.1683 | - |
356
- | 0.7519 | 100 | 0.2187 | - |
357
- | -1 | -1 | - | 0.9854 (-0.0146) |
358
-
 
 
 
 
359
 
360
  ### Framework Versions
361
  - Python: 3.12.12
 
27
  type: affiliation-val
28
  metrics:
29
  - type: map
30
+ value: 0.981
31
  name: Map
32
  - type: mrr@10
33
+ value: 0.981
34
  name: Mrr@10
35
  - type: ndcg@10
36
+ value: 0.9897
37
  name: Ndcg@10
38
  ---
39
 
 
142
 
143
  | Metric | Value |
144
  |:------------|:---------------------|
145
+ | map | 0.9810 (-0.0190) |
146
+ | mrr@10 | 0.9810 (-0.0190) |
147
+ | **ndcg@10** | **0.9897 (-0.0103)** |
148
 
149
  <!--
150
  ## Bias, Risks and Limitations
 
217
  - `per_device_train_batch_size`: 128
218
  - `per_device_eval_batch_size`: 128
219
  - `learning_rate`: 3e-05
 
220
  - `warmup_ratio`: 0.1
221
  - `bf16`: True
222
  - `load_best_model_at_end`: True
 
243
  - `adam_beta2`: 0.999
244
  - `adam_epsilon`: 1e-08
245
  - `max_grad_norm`: 1.0
246
+ - `num_train_epochs`: 3
247
  - `max_steps`: -1
248
  - `lr_scheduler_type`: linear
249
  - `lr_scheduler_kwargs`: {}
 
348
  </details>
349
 
350
  ### Training Logs
351
+ | Epoch | Step | Training Loss | Validation Loss | affiliation-val_ndcg@10 |
352
+ |:----------:|:-------:|:-------------:|:---------------:|:-----------------------:|
353
+ | -1 | -1 | - | - | 0.9273 (-0.0727) |
354
+ | 0.0075 | 1 | 0.1683 | - | - |
355
+ | 0.7519 | 100 | 0.2401 | - | - |
356
+ | 1.5038 | 200 | 0.1296 | - | - |
357
+ | **1.8797** | **250** | **-** | **0.2779** | **0.9897 (-0.0103)** |
358
+ | 2.2556 | 300 | 0.1262 | - | - |
359
+ | -1 | -1 | - | - | 0.9897 (-0.0103) |
360
+
361
+ * The bold row denotes the saved checkpoint.
362
 
363
  ### Framework Versions
364
  - Python: 3.12.12
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9841d63b811274db7b1be86f79bfe8c27d7e88d2360eea3d0060e7438640417a
3
  size 556892306
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a742505689704e9f4f3a53a84798a327d2e20cbbd7364c47994abe067162830
3
  size 556892306