Instructions to use thawndev/llama-cpp-systemic-eog-oob-poc with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thawndev/llama-cpp-systemic-eog-oob-poc with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="thawndev/llama-cpp-systemic-eog-oob-poc",
	filename="poc_bert_defaults_oob.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use thawndev/llama-cpp-systemic-eog-oob-poc with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf thawndev/llama-cpp-systemic-eog-oob-poc
# Run inference directly in the terminal:
llama-cli -hf thawndev/llama-cpp-systemic-eog-oob-poc

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf thawndev/llama-cpp-systemic-eog-oob-poc
# Run inference directly in the terminal:
llama-cli -hf thawndev/llama-cpp-systemic-eog-oob-poc

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf thawndev/llama-cpp-systemic-eog-oob-poc
# Run inference directly in the terminal:
./llama-cli -hf thawndev/llama-cpp-systemic-eog-oob-poc

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf thawndev/llama-cpp-systemic-eog-oob-poc
# Run inference directly in the terminal:
./build/bin/llama-cli -hf thawndev/llama-cpp-systemic-eog-oob-poc

Use Docker

docker model run hf.co/thawndev/llama-cpp-systemic-eog-oob-poc

LM Studio
Jan
Ollama
How to use thawndev/llama-cpp-systemic-eog-oob-poc with Ollama:
```
ollama run hf.co/thawndev/llama-cpp-systemic-eog-oob-poc
```

Unsloth Studio

How to use thawndev/llama-cpp-systemic-eog-oob-poc with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for thawndev/llama-cpp-systemic-eog-oob-poc to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for thawndev/llama-cpp-systemic-eog-oob-poc to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for thawndev/llama-cpp-systemic-eog-oob-poc to start chatting

Docker Model Runner
How to use thawndev/llama-cpp-systemic-eog-oob-poc with Docker Model Runner:
```
docker model run hf.co/thawndev/llama-cpp-systemic-eog-oob-poc
```

Lemonade

How to use thawndev/llama-cpp-systemic-eog-oob-poc with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull thawndev/llama-cpp-systemic-eog-oob-poc

Run and chat with the model

lemonade run user.llama-cpp-systemic-eog-oob-poc-{{QUANT_TAG}}

List all available models

lemonade list

llama-cpp-systemic-eog-oob-poc / README.md

thawndev

Upload README.md with huggingface_hub

a91f5d2 verified 22 days ago

preview code

raw

history blame contribute delete

678 Bytes

	# Systemic OOB memory corruption via hardcoded default special token IDs across multiple tokenizer models in llama.cpp

	Affected software: ggerganov/llama.cpp
	Bug class: memory corruption (out-of-bounds read/write via operator[])
	Impact: Out-of-bounds memory access in the EOG (end-of-generation) token processing loop at llama-vocab.cpp:2646, affecting SPM/llama, T5, and GPT-2 tokenizer models. In release builds without STL assertions, operator[] silently reads/writes adjacent heap memory.
	Format: gguf

	This repository contains a proof-of-concept file for a vulnerability report. Access is gated — request via HuggingFace and grant to protectai-bot only.

	# Systemic OOB memory corruption via hardcoded default special token IDs across multiple tokenizer models in llama.cpp

	Affected software: ggerganov/llama.cpp
	Bug class: memory corruption (out-of-bounds read/write via operator[])
	Impact: Out-of-bounds memory access in the EOG (end-of-generation) token processing loop at llama-vocab.cpp:2646, affecting SPM/llama, T5, and GPT-2 tokenizer models. In release builds without STL assertions, operator[] silently reads/writes adjacent heap memory.
	Format: gguf

	This repository contains a proof-of-concept file for a vulnerability report. Access is gated — request via HuggingFace and grant to protectai-bot only.