9 citations · 9 across the 2 of their papers we have counts for
3 papers · 1 filter
Do Emergent Misalignment Models Share a Circuit?
Muhammad Haseeb, Muhammad Tahir
Fine-tuning language models on narrowly defined harmful tasks can induce emergent misalignment (EM) across unrelated domains. While prior work has identified shared misalignment-as…
BaCP: Backbone Contrastive Pruning for Preserving Representations in Extremely Sparse Neural Networks
Mohammad Haroon Khawaja, Muhammad Haseeb, Mohammad Fatim Shoaib +1
Unstructured pruning at extreme sparsity often suffers from representational collapse, causing sharp drops in accuracy. To address this, we study Backbone Contrastive Pruning (BaCP…
Mechanistic Evidence for Preserved-but-Misaligned Representations in Non-IID FedAvg
Muhammad Haseeb, Salaar Masood, Muhammad Abdullah Sohail +2
Federated Averaging (FedAvg) often degrades under non-IID client data, but it remains unclear whether this degradation reflects the loss of client-learned representations or a fail…