Instructions to use jor-t/llm-demo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jor-t/llm-demo with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="jor-t/llm-demo")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("jor-t/llm-demo", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use jor-t/llm-demo with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jor-t/llm-demo:Q8_0
# Run inference directly in the terminal:
llama cli -hf jor-t/llm-demo:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jor-t/llm-demo:Q8_0
# Run inference directly in the terminal:
llama cli -hf jor-t/llm-demo:Q8_0

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf jor-t/llm-demo:Q8_0
# Run inference directly in the terminal:
./build/bin/llama cli -hf jor-t/llm-demo:Q8_0

Use Docker

docker model run hf.co/jor-t/llm-demo:Q8_0

Llama new
LM Studio
Jan

vLLM

How to use jor-t/llm-demo with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jor-t/llm-demo"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jor-t/llm-demo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/jor-t/llm-demo:Q8_0

SGLang

How to use jor-t/llm-demo with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "jor-t/llm-demo" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jor-t/llm-demo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "jor-t/llm-demo" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jor-t/llm-demo",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Ollama
How to use jor-t/llm-demo with Ollama:
```
ollama run hf.co/jor-t/llm-demo:Q8_0
```
Unsloth Desktop
Docker Model Runner
How to use jor-t/llm-demo with Docker Model Runner:
```
docker model run hf.co/jor-t/llm-demo:Q8_0
```

Lemonade

How to use jor-t/llm-demo with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull jor-t/llm-demo:Q8_0

Run and chat with the model

lemonade run user.llm-demo-Q8_0

List all available models

lemonade list

Atomic Chat

llm-demo

Tiny from-scratch Llama models and the trainer that makes them (formerly scratch-demo).

training/ - Tiny LLM Trainer v5.2 (modular; see training/modules.json). Paste the cell from training/README.md into Kaggle.
datasets/english/ - twitter_chat_113000.txt and twitter_chat_115000.txt (refreshed from GitHub; one message per line, conversations separated by blank lines → STORY_SEPARATOR = "blank_line").
datasets/english/chat_76m_words.txt - 76,000,008 words / 485,943 conversations of English chat (~416 MB): SODA (allenai/soda, CC-BY-4.0) 56,999,907 words + UltraChat (HuggingFaceH4/ultrachat_200k, MIT) 18,885,191 words + the Twitter chat 114,910 words, shuffled (seed 1234). One turn per line, no speaker labels, conversations separated by 3 blank lines → STORY_SEPARATOR = "blank_line". Note: datasets/english/ now holds 3 files; link a file directly to train on just one.
datasets/karbi/ - Karbi (mjw) text copied from jor-t/datasets karbi/ (~550k unique words, mostly news); sources and licences in datasets/karbi/README.md.
models/tiny5m/ - first demo model (~5.9M params): HF weights, tokenizer, tiny5m-q8_0.gguf and its Ollama Modelfile.
misc/ - llm_bench.py: llama.cpp GPU speed test + prompt tool for Kaggle (see misc/README.md).
bin/llama.cpp/ - shared prebuilt llama.cpp CUDA programs for Kaggle T4 (bench, completion, cli, server, quantize), named <tag>-<cuda>-sm75[-fa] with a .json manifest (see bin/llama.cpp/README.md).

Each folder has its own README: datasets/english/, datasets/karbi/, models/tiny5m/, and training/ (plus one per subfolder).

Downloads last month: 75

GGUF

Model size

5.87M params

Architecture

llama

Hardware compatibility

8-bit