Magnus Lundstedt PRO
maglun
AI & ML interests
Local AI inference. Most recently built a tool for inspecting HF models: https://tensorlens.dev
Recent Activity
repliedto their post 1 day ago
I work on quantizing models to run efficiently on local hardware, and kept being curious how existing quants spend their "bit budget" during optimization and built a local tool to explore. Many quants apply one setting across all tensors, but some do more interesting things: the model in the screenshot holds attention K at 4.5 bits while Q/V/O get 8.5, and protects layer 0 MLP.
It turned out useful enough that I made it public: https://tensorlens.dev
Explore any HF model in the browser without downloading it. The anatomy map is read from the safetensors header via a range request, and only tensors you click ever stream. Large tensors are sampled rather than streamed in full.
Limitations: safetensors only (no GGUF yet), some exotic variants don't work yet, and gated repos aren't supported yet.
Feedback very welcome, especially models that break it. repliedto their post 3 days ago
I work on quantizing models to run efficiently on local hardware, and kept being curious how existing quants spend their "bit budget" during optimization and built a local tool to explore. Many quants apply one setting across all tensors, but some do more interesting things: the model in the screenshot holds attention K at 4.5 bits while Q/V/O get 8.5, and protects layer 0 MLP.
It turned out useful enough that I made it public: https://tensorlens.dev
Explore any HF model in the browser without downloading it. The anatomy map is read from the safetensors header via a range request, and only tensors you click ever stream. Large tensors are sampled rather than streamed in full.
Limitations: safetensors only (no GGUF yet), some exotic variants don't work yet, and gated repos aren't supported yet.
Feedback very welcome, especially models that break it. posted an update 3 days ago
I work on quantizing models to run efficiently on local hardware, and kept being curious how existing quants spend their "bit budget" during optimization and built a local tool to explore. Many quants apply one setting across all tensors, but some do more interesting things: the model in the screenshot holds attention K at 4.5 bits while Q/V/O get 8.5, and protects layer 0 MLP.
It turned out useful enough that I made it public: https://tensorlens.dev
Explore any HF model in the browser without downloading it. The anatomy map is read from the safetensors header via a range request, and only tensors you click ever stream. Large tensors are sampled rather than streamed in full.
Limitations: safetensors only (no GGUF yet), some exotic variants don't work yet, and gated repos aren't supported yet.
Feedback very welcome, especially models that break it.Organizations
