Aurum Brain AI v2026.08.25-1e3fbf0

Aurum Brain AI - Custom Fine-Tuned Model

Model AI buatan sendiri, fine-tuned untuk Bahasa Indonesia + coding + web dev.

Spesifikasi

• Base: Qwen/Qwen2.5-1.5B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: 68 percakapan Bahasa Indonesia
• Quantization: Q4_K_M
• Training: 2 epochs, max_len=768

Cara Pakai

PocketPal (HP): Download GGUF → Import → set system prompt dari system_prompt.txt

Ollama:

ollama create aurum-brain -f modelfile
ollama run aurum-brain

llama.cpp:

./main -m aurum-brain-q4_k_m.gguf -p 'Halo, siapa kamu?'

Commit: 1e3fbf0

Aurum Brain AI Starter starter-v2026.08.25 (Qwen2.5-1.5B Q4)

Aurum Brain AI - Starter Model

PENTING: Ini adalah starter model (Qwen2.5-1.5B tanpa fine-tune).
Untuk versi full fine-tuned Aurum Brain AI, tunggu release "v*.." (bukan "starter-*").

Spesifikasi

• Base: Qwen/Qwen2.5-1.5B-Instruct (tanpa fine-tune)
• Quantization: Q4_K_M (~1GB)
• Sudah pintar Bahasa Indonesia & coding dari base model
• Personality AI: set system prompt dari system_prompt.txt

Cara Pakai

1. Download aurum-brain-starter-q4_k_m.gguf (~1GB)
2. Download system_prompt.txt
3. Import ke PocketPal (lihat docs/POCKETPAL.md)
4. Set system prompt di Settings
5. Mulai chat!

Beda dengan Full Version

Aspek	Starter (ini)	Full Version
Base	Qwen2.5-1.5B	Qwen2.5-3B
Fine-tune	Tidak	LoRA r=32
Size	~1GB	~2GB
Bahasa Indonesia	Default multilingual	Fine-tuned khusus
Personality	From system prompt only	Baked in via fine-tune
Speed (mobile)	Cepat (5-10s)	Sedang (10-30s)

Untuk HP lama atau RAM kecil (<4GB), pakai starter ini.
Untuk hasil terbaik, tunggu full version.

Aurum Brain AI daily-2026.08.25-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.24-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.23-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.22-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.21-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.20-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.19-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)

Aurum Brain AI daily-2026.08.18-02

Aurum Brain AI - Build $VER

AI model auto-trained. Makin pintar tiap hari! 🧠

Spesifikasi

• Base: Qwen/Qwen2.5-3B-Instruct
• Fine-tune: LoRA r=16, alpha=32
• Dataset: $DATASET_SIZE percakapan Bahasa Indonesia
• Quantization: Q4_K_M (~1GB)
• Training: 1 epoch, max_len=512 (daily)

Cara Pakai

1. Download GGUF
2. Import ke PocketPal / Ollama / llama.cpp
3. Set system prompt dari system_prompt.txt

Auto-trained: $(date -u)