Views
No views yet
poc_oom_tensor.ggml — 64 bytes, tensor claiming [1M, 1M] float32 = 4TBpoc_many_tensors.ggml — 62KB, 1000 tensor entries causing parsing overheadbenign.ggml — Clean GGML file for comparison1# GGML files parsed by llama.cpp, ggml library
2# OOM on tensor allocation from manipulated dimensions