Upload transformers_config.json with huggingface_hub

transformers_config.json

@@ -9,7 +9,7 @@
   },
   "training_config": {
     "num_train_epochs": 3,
-    "per_device_train_batch_size": 2,
+    "per_device_train_batch_size": 4,
     "gradient_accumulation_steps": 4,
     "learning_rate": 2e-5,
     "lr_scheduler_type": "cosine",
@@ -27,7 +27,8 @@
     "output_dir": "fine_tuned_model",
     "disable_tqdm": false,
     "report_to": ["tensorboard"],
-    "logging_first_step": true
+    "logging_first_step": true,
+    "dataloader_num_workers": 4
   },
   "hardware_config": {
     "fp16": true,
@@ -35,7 +36,10 @@
     "gradient_checkpointing": true,
     "device_map": "auto",
     "attn_implementation": "eager",
-    "use_flash_attention": false
+    "use_flash_attention": false,
+    "memory_optimization": {
+      "expandable_segments": true
+    }
   },
   "quantization_config": {
     "load_in_4bit": true,