File size: 352 Bytes
b7becf2
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
# Integer Overflow in ggml_row_size() Bypassing All GGUF Validation

PoC for integer overflow in llama.cpp tensor size calculation.

## Files
- `poc_tensor_overflow.gguf` - Crafted GGUF model
- `poc_tensor_overflow.py` - Generator script
- `test_tensor_overflow.c` - C test harness

## Disclaimer
For security research and responsible disclosure only.