jaesang-lee/A.X-3.1-GGUF

Included files

• split files matching a.x-3.1-bf16-split.gguf-*.gguf: full-precision BF16 GGUF converted from the upstream safetensors release
• a.x-3.1-q4_k_m.gguf: Q4_K_M quantized GGUF for practical local inference
• LICENSE: upstream Apache-2.0 license
• SHA256SUMS: SHA-256 checksums for the published artifacts

Conversion details

• Base model: skt/A.X-3.1
• Toolchain: ggml-org/llama.cpp
• Converter: convert_hf_to_gguf.py
• Quantizer: llama-quantize
• Generated on: 2026-04-06 UTC

Example usage with llama.cpp

Notes

• The non-quantized export is kept in BF16 because the upstream model is published in BF16.
• The BF16 export is split into multiple GGUF parts to stay under the Hugging Face per-file size limit.
• Q4_K_M is included as the default practical release because it offers a strong size/quality tradeoff.