deve

app.py

@@ -38,8 +38,10 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
                 preds = preds[0]
             # Replace -100s used for padding as we can't decode them
             preds = np.where(preds != -100, preds, tokenizer.pad_token_id)
+            preds = np.array(preds)
             decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
             labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
+            labels = np.array(labels)
             decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
       
             result = metric.compute(predictions=decoded_preds, references=decoded_labels, use_stemmer=True)
@@ -47,7 +49,7 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
             prediction_lens = [np.count_nonzero(pred != tokenizer.pad_token_id) for pred in preds]
             result["gen_len"] = np.mean(prediction_lens)
             accuracy = accuracy_score(decoded_labels, decoded_preds)
-            result["accuracy"] = round(accuracy * 100, 4)            
+            result["eval_accuracy"] = round(accuracy * 100, 4)            
             return result
         
         login(api_key.strip())
@@ -70,9 +72,9 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
             weight_decay=0.01,
             #gradient_accumulation_steps=int(grad),
             #max_grad_norm = 3.0, 
-            #load_best_model_at_end=True,
-            #metric_for_best_model="accuracy",
-            #greater_is_better=True,
+            load_best_model_at_end=True,
+            metric_for_best_model="accuracy",
+            greater_is_better=True,
             logging_dir='/data/logs',
             logging_steps=200,
             #push_to_hub=True,
@@ -207,8 +209,8 @@ def fine_tune_model(model, dataset_name, hub_id, api_key, num_epochs, batch_size
             #return 'RUN AGAIN TO LOAD REST OF DATA'
         dataset = load_dataset(dataset_name.strip())
         #dataset['train'] = dataset['train'].select(range(8000))
-        dataset['train'] = dataset['train'].select(range(1000))
-        dataset['validation'] = dataset['validation'].select(range(100))
+        dataset['train'] = dataset['train'].select(range(4000))
+        dataset['validation'] = dataset['validation'].select(range(200))
         train_set = dataset.map(tokenize_function, batched=True)
         #valid_set = dataset['validation'].map(tokenize_function, batched=True)