Model save

README.md

@@ -230,7 +230,7 @@ model-index:
       type: unknown
     metrics:
     - type: cosine_accuracy@1
-      value: 0.96
+      value: 0.97
       name: Cosine Accuracy@1
     - type: cosine_accuracy@5
       value: 1.0
@@ -239,7 +239,7 @@ model-index:
       value: 1.0
       name: Cosine Accuracy@10
     - type: cosine_precision@1
-      value: 0.96
+      value: 0.97
       name: Cosine Precision@1
     - type: cosine_precision@3
       value: 0.3333333333333334
@@ -251,7 +251,7 @@ model-index:
       value: 0.09999999999999998
       name: Cosine Precision@10
     - type: cosine_recall@1
-      value: 0.96
+      value: 0.97
       name: Cosine Recall@1
     - type: cosine_recall@3
       value: 1.0
@@ -263,25 +263,25 @@ model-index:
       value: 1.0
       name: Cosine Recall@10
     - type: cosine_ndcg@1
-      value: 0.96
+      value: 0.97
       name: Cosine Ndcg@1
     - type: cosine_ndcg@5
-      value: 0.9839278926071438
+      value: 0.9889278926071438
       name: Cosine Ndcg@5
     - type: cosine_ndcg@10
-      value: 0.9839278926071438
+      value: 0.9889278926071438
       name: Cosine Ndcg@10
     - type: cosine_mrr@1
-      value: 0.96
+      value: 0.97
       name: Cosine Mrr@1
     - type: cosine_mrr@5
-      value: 0.9783333333333333
+      value: 0.985
       name: Cosine Mrr@5
     - type: cosine_mrr@10
-      value: 0.9783333333333333
+      value: 0.985
       name: Cosine Mrr@10
     - type: cosine_map@100
-      value: 0.9783333333333333
+      value: 0.985
       name: Cosine Map@100
 ---
 
@@ -350,7 +350,7 @@ print(query_embeddings.shape, document_embeddings.shape)
 # Get the similarity scores for the embeddings
 similarities = model.similarity(query_embeddings, document_embeddings)
 print(similarities)
-# tensor([[ 0.8839, -0.1092,  0.1013]])
+# tensor([[ 0.8193, -0.1132,  0.0397]])
 ```
 
 <!--
@@ -387,24 +387,24 @@ You can finetune this model on your own dataset.
 
 | Metric              | Value      |
 |:--------------------|:-----------|
-| cosine_accuracy@1   | 0.96       |
+| cosine_accuracy@1   | 0.97       |
 | cosine_accuracy@5   | 1.0        |
 | cosine_accuracy@10  | 1.0        |
-| cosine_precision@1  | 0.96       |
+| cosine_precision@1  | 0.97       |
 | cosine_precision@3  | 0.3333     |
 | cosine_precision@5  | 0.2        |
 | cosine_precision@10 | 0.1        |
-| cosine_recall@1     | 0.96       |
+| cosine_recall@1     | 0.97       |
 | cosine_recall@3     | 1.0        |
 | cosine_recall@5     | 1.0        |
 | cosine_recall@10    | 1.0        |
-| cosine_ndcg@1       | 0.96       |
-| cosine_ndcg@5       | 0.9839     |
-| **cosine_ndcg@10**  | **0.9839** |
-| cosine_mrr@1        | 0.96       |
-| cosine_mrr@5        | 0.9783     |
-| cosine_mrr@10       | 0.9783     |
-| cosine_map@100      | 0.9783     |
+| cosine_ndcg@1       | 0.97       |
+| cosine_ndcg@5       | 0.9889     |
+| **cosine_ndcg@10**  | **0.9889** |
+| cosine_mrr@1        | 0.97       |
+| cosine_mrr@5        | 0.985      |
+| cosine_mrr@10       | 0.985      |
+| cosine_map@100      | 0.985      |
 
 <!--
 ## Bias, Risks and Limitations
@@ -476,9 +476,8 @@ You can finetune this model on your own dataset.
 #### Non-Default Hyperparameters
 
 - `eval_strategy`: epoch
-- `per_device_train_batch_size`: 4
-- `per_device_eval_batch_size`: 4
-- `gradient_accumulation_steps`: 4
+- `per_device_train_batch_size`: 16
+- `per_device_eval_batch_size`: 16
 - `num_train_epochs`: 1
 - `warmup_ratio`: 0.1
 - `seed`: 2025
@@ -499,11 +498,11 @@ You can finetune this model on your own dataset.
 - `do_predict`: False
 - `eval_strategy`: epoch
 - `prediction_loss_only`: True
-- `per_device_train_batch_size`: 4
-- `per_device_eval_batch_size`: 4
+- `per_device_train_batch_size`: 16
+- `per_device_eval_batch_size`: 16
 - `per_gpu_train_batch_size`: None
 - `per_gpu_eval_batch_size`: None
-- `gradient_accumulation_steps`: 4
+- `gradient_accumulation_steps`: 1
 - `eval_accumulation_steps`: None
 - `torch_empty_cache_steps`: None
 - `learning_rate`: 5e-05
@@ -619,9 +618,8 @@ You can finetune this model on your own dataset.
 ### Training Logs
 | Epoch   | Step   | Validation Loss | cosine_ndcg@10 |
 |:-------:|:------:|:---------------:|:--------------:|
-| 0       | 0      | 0.0042          | 0.9926         |
-| **1.0** | **25** | **0.0014**      | **0.9839**     |
-| -1      | -1     | -               | 0.9839         |
+| 0       | 0      | 0.0191          | 0.9926         |
+| **1.0** | **25** | **0.0061**      | **0.9889**     |
 
 * The bold row denotes the saved checkpoint.