Downloads · 30 days
0
llama3.1-8b-cpmt-cefr-wce-global-controller
MohammadKhosravi/llama3.1-8b-cpmt-cefr-wce-global-controller
llama3.1-8b-cpmt-cefr-wce-global-controller is a machine learning model from MohammadKhosravi. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft. The card lists the license as mit.
This controller repository holds the optimized parameters generated across the full 60K imbalanced EFCAMDAT corpus splits.
Access
Public
Updated Jul 19, 2026
Repo size
67.4 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt67.4 MB · 100%
At a glance
- Library
- peft
- License
- mit
- Access
- Public
- Created
- Jul 19, 2026
- Updated
- Jul 19, 2026
- SHA
- 0eafc34f
Base models
Model card
From the Hugging Face model README
Balanced C-PMT Global Controller (Weighted Cross Entropy Edition)
This controller repository holds the optimized parameters generated across the full 60K imbalanced EFCAMDAT corpus splits.
Loss Formulation Details
To mitigate the data starvation bottleneck impacting minority classifications (C1/C2), optimizations utilize sequence-level Weighted Cross Entropy (WCE). Gradients from sparse classes are mathematically prioritized during backpropagation according to inverse frequency distributions:
- A1 Weight: 0.5989
- A2 Weight: 0.6809
- B1 Weight: 0.8089
- B2 Weight: 0.8303
- C1 Weight: 3.1740
- C2 Weight: 9.4571
This forces the network blocks to balance hidden-state updates evenly across structural complexity parameters while protecting base linguistic fluency.
- Library
- peft
- License
- mit
- Created
- Jul 19, 2026
- Updated
- Jul 19, 2026