Training run v20250727_173457 - F1: 88.7676, EM: 80.4541

README.md

@@ -22,7 +22,7 @@ model-index:
     - type: exact_match
       value: N/A
     - type: f1
-      value: 89.56708898636393
+      value: 89.93540108105752
 ---
 
 # albert-base-v2 fine-tuned on SQuAD
@@ -37,12 +37,12 @@ This model is a fine-tuned version of [albert-base-v2](https://huggingface.co/al
 - **Dataset**: SQuAD
 - **Optimizer**: adamw
 - **Learning Rate Scheduler**: cosine_with_restarts
-- **Learning Rate**: 8e-05
-- **Batch Size**: 24 per device
-- **Total Batch Size**: 192
+- **Learning Rate**: 6e-05
+- **Batch Size**: 28 per device
+- **Total Batch Size**: 224
 - **Epochs**: 6 (with early stopping)
 - **Weight Decay**: 0.005
-- **Warmup Ratio**: 0.03
+- **Warmup Ratio**: 0.08
 - **Max Gradient Norm**: 0.5
 
 ### Early Stopping
@@ -78,11 +78,11 @@ print(f"Answer: {answer}")
 
 The model achieved the following results on the evaluation set:
 
-- **Exact Match**: 79.2999
-- **F1 Score**: 88.0907
+- **Exact Match**: 80.4541
+- **F1 Score**: 88.7676
 
 ## Training Configuration Hash
 
-Config Hash: d92d5758
+Config Hash: a8d23824
 
 This hash can be used to reproduce the exact training configuration.

eval_results.json

@@ -1,4 +1,4 @@
 {
-  "exact_match": 82.05298013245033,
-  "f1": 89.56708898636393
+  "exact_match": 82.69631031220435,
+  "f1": 89.93540108105752
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1c1b8a698d2b09adb89ae35f8dae34da1657e13cd438f720d72951743d65b517
+oid sha256:191e91d64faf0b47d869e6b4936c080d5182f28224dd5f1615880bfef5bd6fc7
 size 44381360

training_config.json

@@ -10,11 +10,11 @@
   "context_dropout": 0.05,
   "question_paraphrasing": true,
   "negative_sampling": true,
-  "batch_size": 24,
+  "batch_size": 28,
   "num_epochs": 6,
-  "learning_rate": 8e-05,
+  "learning_rate": 6e-05,
   "weight_decay": 0.005,
-  "warmup_ratio": 0.03,
+  "warmup_ratio": 0.08,
   "gradient_accumulation_steps": 2,
   "max_grad_norm": 0.5,
   "optimizer_type": "adamw",