Hugging Face's logo

MohammadKhosravi
/
llama3.1-8b-cpmt-cefr-combined-loss-2layer-head-6k

PEFT
conditional-prefix-memory-tuning
cefr-alignment
combined-loss
topic-aligned
auxiliary-classifier-head
Model card Files
xet
 Community

Combined Loss C-PMT Global Controller (6K Balanced Subset)

This controller is trained using a multi-task Combined Loss Objective ($L_{\text{total}} = L_{\text{token}} + 0.5 \cdot L_{\text{cefr}}$).

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for MohammadKhosravi/llama3.1-8b-cpmt-cefr-combined-loss-2layer-head-6k

Base model

meta-llama/Llama-3.1-8B
Finetuned
meta-llama/Llama-3.1-8B-Instruct
Adapter
(2742)
this model

Collection including MohammadKhosravi/llama3.1-8b-cpmt-cefr-combined-loss-2layer-head-6k