MAB (Plan)
SAB1 (Research)
SAB2 (Strategy)
SAB3 (Content)
Campaign Mode
Platform: All
Activity Stream
00:00:00
Agent Info
SAB1
Market Research Analyst
Idle
Provider
Ollama
Model
llama3:8b
Tools
serper_search, website_scraper
Agent Info
SAB2
Strategy Specialist
Idle
Provider
Ollama
Model
llama3:8b
Agent Info
SAB3
Senior Copywriter
Idle
Provider
Ollama
Model
llama3:8b
Settings
Configure agents, pre-instructions, and API keys
System Configuration
Global Engine Settings
Demo Mode Active
When enabled, agents will use real LLM inference and live tools instead of mock demo data.
MAB - Main Agent Brain
Orchestrator & Manager
SAB1 - Research Analyst
Market Research
SAB2 - Strategy Specialist
Content Strategy
SAB3 - Content Writer
Copywriting
Tool Registry
| Name | Status | Assigned to | Actions |
|---|---|---|---|
| serper_search | Installed | SAB1 | |
| website_scraper | Not Installed | SAB1, SAB2 |
Performance Dashboard
Real-time AMD GPU inference metrics and benchmark comparison.
Total Requests
0
All-time requests
Avg Tokens/sec
0
Throughput metric
AMD Requests
0
GPU-accelerated
Avg Latency
0 ms
Response time
Latency Over Time
Benchmark Results
| Provider | Model | Latency (ms) | TPS | GPU |
|---|---|---|---|---|
| No benchmark data yet. Click "Run Benchmark" to start. | ||||
Performance Summary
Last Updated
Never
Demo Mode
Enabled
About NeuroDesk AMD
Powered by AMD Developer Cloud.
🚀 Powered by AMD MI300X
NeuroDesk leverages AMD Developer Cloud with MI300X GPUs for high-performance LLM inference.
AMD Cloud (vLLM)
Open-source models served on AMD GPUs via vLLM
RecommendedHugging Face Compatible
Deploy as Hugging Face Space with public-safe mode
ROCm Ready
Optimized for ROCm stack on AMD GPUs
Supported Model Endpoints
- Qwen2.5 Instruct - Qwen/Qwen2.5-7B-Instruct
- Llama 3.1 Instruct - meta-llama/Llama-3.1-8B-Instruct
- Mistral Instruct - mistralai/Mistral-7B-Instruct
- DeepSeek Instruct - deepseek-ai/DeepSeek-V2.5
Getting Started
- Sign up for AMD Developer Cloud
- Create a VM with AMD MI300X
- Install vLLM or SGLang for OpenAI-compatible endpoint
- Configure your model endpoint and API key in Settings
Session History
View and resume your past campaigns.