diff --git a/eval_configs/minigptv2_eval_textvqa.yaml b/eval_configs/minigptv2_eval_textvqa.yaml
new file mode 100644
index 0000000..1937a5b
--- /dev/null
+++ b/eval_configs/minigptv2_eval_textvqa.yaml
@@ -0,0 +1,25 @@
+model:
+  arch: minigpt_v2
+  model_type: pretrain
+  max_txt_len: 500
+  end_sym: "</s>"
+  low_resource: True
+  prompt_template: '[INST] {} [/INST]'
+  # ckpt: "./minigpt4/textvqa_outputs/20250112105/checkpoint_9.pth"
+  ckpt: "./ckpt/checkpoint_stage3.pth"
+  lora_r: 64
+  lora_alpha: 16
+
+
+datasets:
+  cc_sbu_align:
+    vis_processor:
+      train:
+        name: "blip2_image_eval"
+        image_size: 448
+    text_processor:
+      train:
+        name: "blip_caption"
+
+run:
+  task: image_text_pretrain