google/gemma-2-2b
Text Generation • 3B • Updated • 266k • • 686
Yash Bishnoi PRO
bishnoiyash
·
AI & ML interests
▎ Reproducing LLMs from a blank file. Single-file PyTorch Qwen3-0.6B and SmolLM2-135M, checked against the official weights: max|Δlogits| = 0.000e+00 on CPU/fp32.
Recent Activity
published an article 2 days ago
Decoder crowding predicts SAE steering side effects, and it is not just firing frequency updated a model 4 days ago
bishnoiyash/SmolLM2-135M-tinystories published a model 4 days ago
bishnoiyash/SmolLM2-135M-tinystoriesOrganizations
None yet
liked a
model 29 days ago