423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-10-16 00:28:10 +08:00
Code Issues Packages Projects Releases Wiki Activity

add image num check

Browse Source 
Former-commit-id: 15201113bf16b748c0a758c7a5b363da8272e0e6
This commit is contained in:
hiyouga 2024-09-02 01:31:36 +08:00
parent 806e1061d4
commit bae73e676c
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
src/llamafactory/data/mm_plugin.py
View File

@ -210,12 +210,17 @@ class Qwen2vlPlugin(BasePlugin):
merge_length: int = getattr(image_processor, "merge_size") ** 2 merge_length: int = getattr(image_processor, "merge_size") ** 2
if len(images) > 0: if len(images) > 0:
image_grid_thw = _get_mm_inputs(images, processor)["image_grid_thw"] image_grid_thw = _get_mm_inputs(images, processor)["image_grid_thw"]
else:
image_grid_thw = []
num_images = 0 num_images = 0
messages = deepcopy(messages) messages = deepcopy(messages)
for message in messages: for message in messages:
content = message["content"] content = message["content"]
while IMAGE_PLACEHOLDER in content: while IMAGE_PLACEHOLDER in content:
if num_images >= len(image_grid_thw):
raise ValueError("`len(images)` is less than the number of {} tokens.".format(IMAGE_PLACEHOLDER))
content = content.replace( content = content.replace(
IMAGE_PLACEHOLDER, IMAGE_PLACEHOLDER,
"<|vision_start|>{}<|vision_end|>".format( "<|vision_start|>{}<|vision_end|>".format(