From 779d0fc5d10f79cf5ad5623c63ea28e4d4d2848b Mon Sep 17 00:00:00 2001
From: Fanyi Pu <FPU001@e.ntu.edu.sg>
Date: Thu, 30 Jan 2025 22:08:23 +0800
Subject: [PATCH] update Qwen2_5_VL model parameters and add .gitignore file

---
 .gitignore                     | 45 ++++++++++++++++++++++++++++++++++
 lmms_eval/models/qwen2_5_vl.py |  7 +++---
 2 files changed, 48 insertions(+), 4 deletions(-)
 create mode 100644 .gitignore

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..cbcbed6
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,45 @@
+**/.cache
+env
+*.pyc
+output/
+data/
+lm_cache
+.idea
+build
+dist
+*.egg-info
+venv
+.vscode/
+temp
+__pycache__
+.ipynb_checkpoints
+temp
+.DS_STORE
+# IPython
+profile_default/
+ipython_config.py
+logs/
+scripts/
+wandb/
+SimSun.ttf
+submissions/
+lmms_eval/tasks/hallusion_bench/hallusion_output_vs_model.json
+lmms_eval/tasks/hallusion_bench/hallusion_output_vd_model.json
+zk.log
+cache_dir
+ckpt
+pretrained/
+LLaVA/
+*logs
+*.isorted
+temp/
+InternVL/
+logs/
+data/
+llava-video/
+Video-MME/
+VATEX/
+lmms_eval/tasks/vatex/__pycache__/utils.cpython-310.pyc
+lmms_eval/tasks/mlvu/__pycache__/utils.cpython-310.pyc
+
+scripts/
diff --git a/lmms_eval/models/qwen2_5_vl.py b/lmms_eval/models/qwen2_5_vl.py
index 6b683c4..ef5de1d 100644
--- a/lmms_eval/models/qwen2_5_vl.py
+++ b/lmms_eval/models/qwen2_5_vl.py
@@ -37,9 +37,9 @@ def __init__(
         device_map: Optional[str] = "auto",
         batch_size: Optional[Union[int, str]] = 1,
         use_cache=True,
-        use_flash_attention_2: Optional[bool] =  False,
-        min_pixels: int = 256 * 28 * 28,
-        max_pixels: int = 256 * 28 * 28,
+        use_flash_attention_2: Optional[bool] = True,
+        min_pixels: int = 12845056,
+        max_pixels: int = 3136,
         max_num_frames: int = 32,
         use_custom_video_loader: Optional[bool] = False,
         fps: Optional[float] = None,  # Only applicable if use_custom_video_loader is True
@@ -82,7 +82,6 @@ def __init__(
         self.max_pixels = max_pixels
         self.min_pixels = min_pixels
         self.max_num_frames = max_num_frames
-        self.processor = AutoProcessor.from_pretrained(pretrained, max_pixels=max_pixels, min_pixels=min_pixels)
         self._tokenizer = AutoTokenizer.from_pretrained(pretrained)
 
         self._config = self.model.config