Video-Text-to-Text
Transformers
Safetensors
English
qwen3_vl
image-text-to-text
video
multi-hop-reasoning
reinforcement-learning
rlvr
grpo
qwen3-vl
Instructions to use ngqtrung/Video-HopChain-8B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ngqtrung/Video-HopChain-8B with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("ngqtrung/Video-HopChain-8B") model = AutoModelForMultimodalLM.from_pretrained("ngqtrung/Video-HopChain-8B", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Commit History
-
Use the capitalised dataset ID (ngqtrung/Video-HopChain)
Verified signatured7051e5 -
Use the capitalised repository IDs (Video-HopChain-8B)
Verified signaturedeb3a63
-
Card: names and numbers aligned with the paper
Verified signatureef46c29 -
Card: consistent naming and wording
Verified signatured9ee7ff
-
Link the standard RL initialization checkpoint
Verified signature7116c8b -
Use the name Video-HopChain consistently
Verified signature145f792
-
Link the project page
Verified signature312721c
-
arXiv BibTeX entry
Verified signature212913d -
Link arXiv:2609.25773
Verified signaturedd893d1
-
Rename the method to Confidence-Gated Exploration (CGE)
Verified signatureda5c94e
-
Link the collection
Verified signature0bda9e6 -
Add model card
Verified signature1bacb7d -
Declare bfloat16 in config: every tensor is BF16, float32 forced a needless upcast on load
Verified signature3190785 -
Add Video-HopChain 8B (combined run, global_step_10)
Verified signatureaa774bc -
initial commit
Verified signature4ca3fb3