Fine-tuned roberta-base on SQuAD - Best F1: 91.8861

README.md

@@ -22,7 +22,7 @@ model-index:
     - type: exact_match
       value: N/A
     - type: f1
-      value: 91.82889892857621
+      value: 91.88608512645385
 ---
 
 # roberta-base fine-tuned on SQuAD
@@ -49,7 +49,7 @@ This model is a fine-tuned version of [roberta-base](https://huggingface.co/robe
 
 - **Patience**: 6
 - **Metric**: f1
-- **Best Epoch**: 3
+- **Best Epoch**: 5
 
 ## Usage
 
@@ -78,8 +78,8 @@ print(f"Answer: {answer}")
 
 The model achieved the following results on the evaluation set:
 
-- **Exact Match**: 85.1845
-- **F1 Score**: 91.7171
+- **Exact Match**: 85.3359
+- **F1 Score**: 91.8861
 
 ## Training Configuration Hash
 

eval_results.json

@@ -1,4 +1,4 @@
 {
-  "exact_match": 85.47776726584674,
-  "f1": 91.82889892857621
+  "exact_match": 85.33585619678335,
+  "f1": 91.88608512645385
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3e29993e1ccad6ea0612d65adb25fee74c44db9fa225b258f064c875ad9f928
+oid sha256:916969e644fde665b1962dacfe31b405b82a1688c32205a1ba8deea2a57eeb82
 size 496250232