Hugging Face
Models
Datasets
Spaces
Community
Docs
Enterprise
Pricing
Log In
Sign Up
unsloth
/
Qwen3-VL-8B-Thinking-FP8
like
1
Follow
Unsloth AI
10.7k
Image-Text-to-Text
Safetensors
qwen3_vl
unsloth
conversational
fp8
arxiv:
4 papers
License:
apache-2.0
Model card
Files
Files and versions
xet
Community
1
New discussion
New pull request
Resources
PR & discussions documentation
Code of Conduct
Hub documentation
All
Discussions
Pull requests
View closed (0)
Sort: Recently created
Why is the pad token of all QWEN VL models in Unsloth "<|vision_pad|>", while QWEN officially uses "pad_token": "<|endoftext|>"
1
#1 opened about 1 month ago by
IlikeLLM