Downloads · 30 days
0
htdy7703/tflite-blockwise-scale-index-crash-poc-20260710
tflite-blockwise-scale-index-crash-poc-20260710 is a machine learning model from htdy7703. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This PoC demonstrates a native crash in the released LiteRT / TFLite runtime when a blockwise-quantized weight tensor uses an out-of-range BlockwiseQuantization.scales tensor index.
Downloads · 30 days
0
Access
Public
Updated Jul 10, 2026
Repo size
—
Likes
0
Public
Click a slice to open those files.
.py6 KB · 70%
From the Hugging Face model README
This PoC demonstrates a native crash in the released LiteRT / TFLite runtime when a blockwise-quantized weight tensor uses an out-of-range BlockwiseQuantization.scales tensor index.
What this PoC shows:
FULLY_CONNECTED .tflite model can be built and executed on ai-edge-litert==2.1.6BlockwiseQuantization.scales to an out-of-range tensor index causes the released runtime to crash with SIGSEGVallocate_tensors()invoke()What this PoC does not show:
Runtime used:
ai-edge-litert==2.1.6Run:
python3 -m pip install --progress-bar off --target /tmp/tflite-hunt \
flatbuffers tflite ai-edge-litert
PYTHONPATH=/tmp/tflite-hunt python3 repro_blockwise_scale_tensor_index.py