Training in progress, step 348

README.md

@@ -31,25 +31,25 @@ model-index:
       type: fineweb_c_eval
     metrics:
     - type: accuracy
-      value: 0.8837890625
+      value: 0.90234375
       name: Accuracy
     - type: accuracy_threshold
-      value: 0.8104759454727173
+      value: 0.7725241184234619
       name: Accuracy Threshold
     - type: f1
-      value: 0.6024844720496895
+      value: 0.6426229508196721
       name: F1
     - type: f1_threshold
-      value: 0.7146279811859131
+      value: 0.601359486579895
       name: F1 Threshold
     - type: precision
-      value: 0.6258064516129033
+      value: 0.7101449275362319
       name: Precision
     - type: recall
-      value: 0.5808383233532934
+      value: 0.5868263473053892
       name: Recall
     - type: average_precision
-      value: 0.6180374262832308
+      value: 0.696727465964923
       name: Average Precision
 ---
 
@@ -110,7 +110,7 @@ pairs = [
 ]
 scores = model.predict(pairs)
 print(scores)
-# [0.1112 0.9909 0.0237 0.0368 0.3648]
+# [1.0987e-03 9.9933e-01 3.3535e-04 1.2066e-03 3.2731e-03]
 
 # Or rank different texts based on similarity to a single text
 ranks = model.rank(
@@ -159,15 +159,15 @@ You can finetune this model on your own dataset.
 * Dataset: `fineweb_c_eval`
 * Evaluated with [<code>CrossEncoderClassificationEvaluator</code>](https://sbert.net/docs/package_reference/cross_encoder/evaluation.html#sentence_transformers.cross_encoder.evaluation.CrossEncoderClassificationEvaluator)
 
-| Metric                | Value     |
-|:----------------------|:----------|
-| accuracy              | 0.8838    |
-| accuracy_threshold    | 0.8105    |
-| f1                    | 0.6025    |
-| f1_threshold          | 0.7146    |
-| precision             | 0.6258    |
-| recall                | 0.5808    |
-| **average_precision** | **0.618** |
+| Metric                | Value      |
+|:----------------------|:-----------|
+| accuracy              | 0.9023     |
+| accuracy_threshold    | 0.7725     |
+| f1                    | 0.6426     |
+| f1_threshold          | 0.6014     |
+| precision             | 0.7101     |
+| recall                | 0.5868     |
+| **average_precision** | **0.6967** |
 
 <!--
 ## Bias, Risks and Limitations
@@ -412,12 +412,22 @@ You can finetune this model on your own dataset.
 | 0.9758 | 282  | 0.9496        | -               | -                                |
 | 0.9965 | 288  | 0.8691        | -               | -                                |
 | 1.0035 | 290  | -             | 0.7659          | 0.6180                           |
+| 1.0173 | 294  | 0.6524        | -               | -                                |
+| 1.0381 | 300  | 0.7019        | -               | -                                |
+| 1.0588 | 306  | 0.4723        | -               | -                                |
+| 1.0796 | 312  | 0.5579        | -               | -                                |
+| 1.1003 | 318  | 0.5251        | -               | -                                |
+| 1.1211 | 324  | 0.6861        | -               | -                                |
+| 1.1419 | 330  | 0.4643        | -               | -                                |
+| 1.1626 | 336  | 0.5970        | -               | -                                |
+| 1.1834 | 342  | 0.4205        | -               | -                                |
+| 1.2042 | 348  | 0.4822        | 0.9538          | 0.6967                           |
 
 
 ### Training Time
-- **Training**: 4.7 minutes
-- **Evaluation**: 1.7 minutes
-- **Total**: 6.4 minutes
+- **Training**: 5.7 minutes
+- **Evaluation**: 2.1 minutes
+- **Total**: 7.8 minutes
 
 ### Framework Versions
 - Python: 3.12.12

eval/CrossEncoderClassificationEvaluator_fineweb_c_eval_results.csv

@@ -4,3 +4,4 @@ epoch,steps,Accuracy,Accuracy_Threshold,F1,F1_Threshold,Precision,Recall,Average
 0.6020761245674741,174,0.849609375,0.81698465,0.5108433734939759,0.6361271,0.4274193548387097,0.6347305389221557,0.4518233024149694
 0.8027681660899654,232,0.869140625,0.81640637,0.53276955602537,0.5794004,0.4117647058823529,0.7544910179640718,0.5406815506036883
 1.0034602076124568,290,0.8837890625,0.81047595,0.6024844720496895,0.714628,0.6258064516129033,0.5808383233532934,0.6180374262832308
+1.2041522491349481,348,0.90234375,0.7725241,0.6426229508196721,0.6013595,0.7101449275362319,0.5868263473053892,0.696727465964923

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:288034c9bbea2f1fa34be4a1b2b589a29b7d0d208b20816b402946412ab02f8a
+oid sha256:d043136e355945d57bcbf7816fcb8037742c275d5c0f6e572f1d812297d2493d
 size 1230138348