Total Robots
—
Deployed
—
Provisioned
—
Online (1h)
—
Draft
—
Retired
—
Recent Robots
| Robot ID | Name | Soul | Brain | Status | Actions |
|---|
| Robot ID | Name | Hardware | Soul | Brain | Status | Provisioned | Actions |
|---|
🤖 Identity
✨ Soul
🧠 Brain
⚙️ Hardware
🌐 Network
🔒 Security
When triggered, permanently destroys on-board storage. Only for classified deployments.
Deployed Robots
| Robot ID | Name | Soul | Last Seen | Firmware | Status | Actions |
|---|
📋 Activity Log — All Robots
📡 Push to Single Robot
🌐 Fleet Broadcast
All deployed + provisioned robots
⚠️ This pushes to ALL active robots. Use carefully.
📋 OTA Update History
Loading…
🧬 ATAVUS Genesis-1 Model
Base: Qwen3-8B Q4_K_M | Fine-tuned for ATAVUS humanoid robotics
Model File
/data/models/qwen3-7b/qwen3-8b-q4_k_m.gguf
Size
4.7 GB (Q4_K_M)
Training Samples
26 humanoid robot dialogues
Fine-tune Method
QLoRA — LoRA rank 16, target: attention layers
Adapter Path
/data/models/genesis-1/adapter/
Training Script
/data/models/genesis_finetune.py
Training Data
/data/models/genesis-1/training_data.jsonl
Runtime
llama.cpp server — /tmp/llama.cpp/build/bin/llama-server
🚀 Run Fine-Tuning
Runs QLoRA fine-tuning on R6515 (AMD EPYC, 251GB RAM, CPU-only). Takes ~2-4 hours for 3 epochs on 26 samples.
cd /data/models && \
/www/wwwroot/atavus.ai/backend/.venv/bin/python3 \
genesis_finetune.py \
--model Qwen/Qwen3-8B \
--epochs 3
/www/wwwroot/atavus.ai/backend/.venv/bin/python3 \
genesis_finetune.py \
--model Qwen/Qwen3-8B \
--epochs 3
Run this in a tmux session on R6515 (185.184.192.91). Adapter saved to /data/models/genesis-1/adapter/ when complete.
📦 Include in Disk Bundle
When provisioning a robot, Genesis-1 is automatically set as the recommended model. The soul.json written to disk includes the correct model reference.
soul.json → brain.model:
"atavus/genesis-1"
boot.sh sets:
ATAVUS_BRAIN_MODEL=atavus/genesis-1
"atavus/genesis-1"
boot.sh sets:
ATAVUS_BRAIN_MODEL=atavus/genesis-1
✅ Genesis-1 is the default model for all new robots