How to use from
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 nabin2004/nebium
# Run inference directly in the terminal:
llama cli -hf nabin2004/nebium
Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf nabin2004/nebium
# Run inference directly in the terminal:
llama cli -hf nabin2004/nebium
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 nabin2004/nebium
# Run inference directly in the terminal:
./llama-cli -hf nabin2004/nebium
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 nabin2004/nebium
# Run inference directly in the terminal:
./build/bin/llama-cli -hf nabin2004/nebium
Use Docker
docker model run hf.co/nabin2004/nebium
Quick Links

Nebium

Causal Transformer for self-supervised next-move prediction on chess games.

Architecture

  • d_model: 512
  • n_heads: 8
  • n_layers: 6
  • vocab_size: 2018
  • max_seq_len: 512
  • positional_encoding: rope
  • activation: swiglu
  • norm: rmsnorm

Artifacts & Formats

  • model.pt: Raw PyTorch state_dict
  • model_config.json: Architecture configuration
  • tokenizer.json: BPE Chess tokenizer
  • nebium.gguf: GGUF format model for local inference and quantization

Last validation metrics

  • val/loss: 2.376061725547775
  • val/accuracy: 0.36386593746665474
  • val/top5_accuracy: 0.705297920239525
  • val/perplexity: 10.762433873185277
  • val/legal_move_rate: 0.9833333333333333
Downloads last month
18
GGUF
Model size
21M params
Architecture
nebium
Hardware compatibility
Log In to add your hardware

We're not able to determine the quantization variants.

Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Space using nabin2004/nebium 1

Collection including nabin2004/nebium