YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Adaptive-MSC v10 (Best Model)
This repository contains the best performing model from the v10 training run.
Training Conditions
- Architecture: 6-Layer Interaction Head (based on AdaptiveMSCModelV7)
- Loss Function: Cross Entropy (BCE) + KL Divergence (Ratio 1:1)
- Base Model: DeBERTa-v3-base (with LoRA)
- Epochs: 10
- Best Step: 7420 (Epoch 9)
- Loss at Best Step: 0.701159 (BCE: 0.567136, KL: 0.134024)
Files
- : The best model weights.
- : Full training statistics across all steps.
- : Visualization of the training progress.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support