language, license, library_name, pipeline_tag, base_model, tags

language	license	library_name	pipeline_tag	base_model	tags
id	apache-2.0	gguf	text-generation	AksaraLLM/Kiel-Pro-0.5B-v3-chat	gguf llama.cpp ollama indonesian aksarallm qwen2

Kiel-Pro-0.5B-v3-chat-GGUF

GGUF quantizations of AksaraLLM/Kiel-Pro-0.5B-v3-chat for inference with llama.cpp, Ollama, LM Studio, and other GGUF runtimes.

Files

File	Quant	Size	Recommended use
Kiel-Pro-0.5B-v3-chat.f16.gguf	F16	0.99 GB	lossless from safetensors
Kiel-Pro-0.5B-v3-chat.q8_0.gguf	Q8_0	0.53 GB	near-lossless, ~2× smaller
Kiel-Pro-0.5B-v3-chat.q6_k.gguf	Q6_K	0.51 GB	high quality, ~2.5× smaller
Kiel-Pro-0.5B-v3-chat.q5_k_m.gguf	Q5_K_M	0.42 GB	good quality, ~3× smaller
Kiel-Pro-0.5B-v3-chat.q4_k_m.gguf	Q4_K_M	0.40 GB	recommended default, ~4× smaller

CPU benchmark (AMD EPYC 7763, 2 threads, AVX2)

Quant	Prompt eval (32 tok)	Generation (16 tok)
q4_k_m	36.8 tok/s	21.6 tok/s

So a 494M model at q4_k_m runs comfortably on a CPU laptop. Larger quants (q5_k_m, q6_k, q8_0) trade a bit of speed for better quality.

Quick start — llama.cpp

huggingface-cli download AksaraLLM/Kiel-Pro-0.5B-v3-chat-GGUF Kiel-Pro-0.5B-v3-chat.q4_k_m.gguf --local-dir .
./llama-cli -m Kiel-Pro-0.5B-v3-chat.q4_k_m.gguf -p "Indonesia adalah" -n 64

Quick start — Ollama

huggingface-cli download AksaraLLM/Kiel-Pro-0.5B-v3-chat-GGUF Kiel-Pro-0.5B-v3-chat.q4_k_m.gguf Modelfile --local-dir .
ollama create aksara-kiel-pro-0.5b-v3-chat -f Modelfile
ollama run aksara-kiel-pro-0.5b-v3-chat "Siapa kamu?"

Source model

See AksaraLLM/Kiel-Pro-0.5B-v3-chat for architecture, training data, eval results, and limitations.

Conversion provenance

• Converted with convert_hf_to_gguf.py from llama.cpp
• Quantized with llama-quantize from the same build
• Architecture detected as qwen2
• All files listed above are reproducible from the source HF safetensors