Add new SentenceTransformer model
Browse files- README.md +22 -22
- config.json +1 -1
- model.safetensors +2 -2
README.md
CHANGED
|
@@ -78,28 +78,28 @@ model-index:
|
|
| 78 |
type: test
|
| 79 |
metrics:
|
| 80 |
- type: cosine_accuracy@1
|
| 81 |
-
value: 0.
|
| 82 |
name: Cosine Accuracy@1
|
| 83 |
- type: cosine_precision@1
|
| 84 |
-
value: 0.
|
| 85 |
name: Cosine Precision@1
|
| 86 |
- type: cosine_recall@1
|
| 87 |
-
value: 0.
|
| 88 |
name: Cosine Recall@1
|
| 89 |
- type: cosine_ndcg@10
|
| 90 |
-
value: 0.
|
| 91 |
name: Cosine Ndcg@10
|
| 92 |
- type: cosine_mrr@1
|
| 93 |
-
value: 0.
|
| 94 |
name: Cosine Mrr@1
|
| 95 |
- type: cosine_map@100
|
| 96 |
-
value: 0.
|
| 97 |
name: Cosine Map@100
|
| 98 |
- type: cosine_auc_precision_cache_hit_ratio
|
| 99 |
-
value: 0.
|
| 100 |
name: Cosine Auc Precision Cache Hit Ratio
|
| 101 |
- type: cosine_auc_similarity_distribution
|
| 102 |
-
value: 0.
|
| 103 |
name: Cosine Auc Similarity Distribution
|
| 104 |
---
|
| 105 |
|
|
@@ -165,9 +165,9 @@ print(embeddings.shape)
|
|
| 165 |
# Get the similarity scores for the embeddings
|
| 166 |
similarities = model.similarity(embeddings, embeddings)
|
| 167 |
print(similarities)
|
| 168 |
-
# tensor([[1.0000, 1.0000, 0.
|
| 169 |
-
# [1.0000, 1.0000, 0.
|
| 170 |
-
# [0.
|
| 171 |
```
|
| 172 |
|
| 173 |
<!--
|
|
@@ -205,13 +205,13 @@ You can finetune this model on your own dataset.
|
|
| 205 |
|
| 206 |
| Metric | Value |
|
| 207 |
|:-------------------------------------|:-----------|
|
| 208 |
-
| cosine_accuracy@1 | 0.
|
| 209 |
-
| cosine_precision@1 | 0.
|
| 210 |
-
| cosine_recall@1 | 0.
|
| 211 |
| **cosine_ndcg@10** | **0.7619** |
|
| 212 |
-
| cosine_mrr@1 | 0.
|
| 213 |
-
| cosine_map@100 | 0.
|
| 214 |
-
| cosine_auc_precision_cache_hit_ratio | 0.
|
| 215 |
| cosine_auc_similarity_distribution | 0.1635 |
|
| 216 |
|
| 217 |
<!--
|
|
@@ -286,8 +286,8 @@ You can finetune this model on your own dataset.
|
|
| 286 |
#### Non-Default Hyperparameters
|
| 287 |
|
| 288 |
- `eval_strategy`: steps
|
| 289 |
-
- `per_device_train_batch_size`:
|
| 290 |
-
- `per_device_eval_batch_size`:
|
| 291 |
- `gradient_accumulation_steps`: 2
|
| 292 |
- `weight_decay`: 0.001
|
| 293 |
- `adam_beta2`: 0.98
|
|
@@ -310,8 +310,8 @@ You can finetune this model on your own dataset.
|
|
| 310 |
- `do_predict`: False
|
| 311 |
- `eval_strategy`: steps
|
| 312 |
- `prediction_loss_only`: True
|
| 313 |
-
- `per_device_train_batch_size`:
|
| 314 |
-
- `per_device_eval_batch_size`:
|
| 315 |
- `per_gpu_train_batch_size`: None
|
| 316 |
- `per_gpu_eval_batch_size`: None
|
| 317 |
- `gradient_accumulation_steps`: 2
|
|
@@ -431,7 +431,7 @@ You can finetune this model on your own dataset.
|
|
| 431 |
### Training Logs
|
| 432 |
| Epoch | Step | Validation Loss | test_cosine_ndcg@10 |
|
| 433 |
|:-----:|:----:|:---------------:|:-------------------:|
|
| 434 |
-
| 0 | 0 | 0.
|
| 435 |
|
| 436 |
|
| 437 |
### Framework Versions
|
|
|
|
| 78 |
type: test
|
| 79 |
metrics:
|
| 80 |
- type: cosine_accuracy@1
|
| 81 |
+
value: 0.5763286334056399
|
| 82 |
name: Cosine Accuracy@1
|
| 83 |
- type: cosine_precision@1
|
| 84 |
+
value: 0.5763286334056399
|
| 85 |
name: Cosine Precision@1
|
| 86 |
- type: cosine_recall@1
|
| 87 |
+
value: 0.5589816867630893
|
| 88 |
name: Cosine Recall@1
|
| 89 |
- type: cosine_ndcg@10
|
| 90 |
+
value: 0.7619419081029518
|
| 91 |
name: Cosine Ndcg@10
|
| 92 |
- type: cosine_mrr@1
|
| 93 |
+
value: 0.5763286334056399
|
| 94 |
name: Cosine Mrr@1
|
| 95 |
- type: cosine_map@100
|
| 96 |
+
value: 0.7107794631883741
|
| 97 |
name: Cosine Map@100
|
| 98 |
- type: cosine_auc_precision_cache_hit_ratio
|
| 99 |
+
value: 0.3488530268041688
|
| 100 |
name: Cosine Auc Precision Cache Hit Ratio
|
| 101 |
- type: cosine_auc_similarity_distribution
|
| 102 |
+
value: 0.1634818016054941
|
| 103 |
name: Cosine Auc Similarity Distribution
|
| 104 |
---
|
| 105 |
|
|
|
|
| 165 |
# Get the similarity scores for the embeddings
|
| 166 |
similarities = model.similarity(embeddings, embeddings)
|
| 167 |
print(similarities)
|
| 168 |
+
# tensor([[1.0000, 1.0000, 0.3428],
|
| 169 |
+
# [1.0000, 1.0000, 0.3428],
|
| 170 |
+
# [0.3428, 0.3428, 1.0000]])
|
| 171 |
```
|
| 172 |
|
| 173 |
<!--
|
|
|
|
| 205 |
|
| 206 |
| Metric | Value |
|
| 207 |
|:-------------------------------------|:-----------|
|
| 208 |
+
| cosine_accuracy@1 | 0.5763 |
|
| 209 |
+
| cosine_precision@1 | 0.5763 |
|
| 210 |
+
| cosine_recall@1 | 0.559 |
|
| 211 |
| **cosine_ndcg@10** | **0.7619** |
|
| 212 |
+
| cosine_mrr@1 | 0.5763 |
|
| 213 |
+
| cosine_map@100 | 0.7108 |
|
| 214 |
+
| cosine_auc_precision_cache_hit_ratio | 0.3489 |
|
| 215 |
| cosine_auc_similarity_distribution | 0.1635 |
|
| 216 |
|
| 217 |
<!--
|
|
|
|
| 286 |
#### Non-Default Hyperparameters
|
| 287 |
|
| 288 |
- `eval_strategy`: steps
|
| 289 |
+
- `per_device_train_batch_size`: 1152
|
| 290 |
+
- `per_device_eval_batch_size`: 1152
|
| 291 |
- `gradient_accumulation_steps`: 2
|
| 292 |
- `weight_decay`: 0.001
|
| 293 |
- `adam_beta2`: 0.98
|
|
|
|
| 310 |
- `do_predict`: False
|
| 311 |
- `eval_strategy`: steps
|
| 312 |
- `prediction_loss_only`: True
|
| 313 |
+
- `per_device_train_batch_size`: 1152
|
| 314 |
+
- `per_device_eval_batch_size`: 1152
|
| 315 |
- `per_gpu_train_batch_size`: None
|
| 316 |
- `per_gpu_eval_batch_size`: None
|
| 317 |
- `gradient_accumulation_steps`: 2
|
|
|
|
| 431 |
### Training Logs
|
| 432 |
| Epoch | Step | Validation Loss | test_cosine_ndcg@10 |
|
| 433 |
|:-----:|:----:|:---------------:|:-------------------:|
|
| 434 |
+
| 0 | 0 | 0.6981 | 0.7619 |
|
| 435 |
|
| 436 |
|
| 437 |
### Framework Versions
|
config.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
],
|
| 5 |
"attention_probs_dropout_prob": 0.1,
|
| 6 |
"classifier_dropout": null,
|
| 7 |
-
"dtype": "
|
| 8 |
"gradient_checkpointing": false,
|
| 9 |
"hidden_act": "gelu",
|
| 10 |
"hidden_dropout_prob": 0.1,
|
|
|
|
| 4 |
],
|
| 5 |
"attention_probs_dropout_prob": 0.1,
|
| 6 |
"classifier_dropout": null,
|
| 7 |
+
"dtype": "float32",
|
| 8 |
"gradient_checkpointing": false,
|
| 9 |
"hidden_act": "gelu",
|
| 10 |
"hidden_dropout_prob": 0.1,
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4652c7e874d7264659b94a56fddfeb099ad39cc4909b1947b1d06df95e701f72
|
| 3 |
+
size 90864192
|