mech-interp suite for Granite4 models that use Mamba-2 architecture
-
Updated
Feb 21, 2026 - Python
#
mamba-2
Here are 3 public repositories matching this topic...
Early baby steps towards a long-term vision regarding Mamba-2's state interpretability.
python machine-learning research deep-learning transformers pytorch mamba state-space-models ai-research latent-space alsi non-linear-dynamics mechanistic-interpretability semantic-control representation-engineering llm-steering mamba-2 latent-steering phi-projector inference-time-intervention
-
Updated
Feb 4, 2026 - Python
Systematic study of LoRA fine-tuning strategies for IBM Granite 4.0-H-Micro (Mamba-2 + Transformer hybrid). Demonstrates the impact of architecture-aware target selection and SSM core parameter co-training, including analysis of PEFT serialization behavior. Reports up to 37% relative improvement over LoRA-only baselines.
lora granite mamba state-space-model peft llama-cpp gguf parameter-efficient-fine-tuning llm-finetuning hybrid-architecture ibm-granite mamba-2 granite4
-
Updated
Mar 1, 2026 - Python
Improve this page
Add a description, image, and links to the mamba-2 topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the mamba-2 topic, visit your repo's landing page and select "manage topics."