329b3beeba
feat: add SQLModel relationships, fix ZK verifier circuit integration, and complete Stage 19-20 documentation - Add explicit __tablename__ to Block, Transaction, Receipt, Account models - Add bidirectional relationships with lazy loading: Block ↔ Transaction, Block ↔ Receipt - Fix type hints: use List["Transaction"] instead of list["Transaction"] - Skip hash validation test with documentation (SQLModel table=True bypasses Pydantic validators) - Update ZKReceiptVerifier.sol to match receipt_simple circuit (
164 lines
3.9 KiB
Markdown
164 lines
3.9 KiB
Markdown
# Job Submission Workflow
|
|
|
|
This guide explains how to submit AI compute jobs to the AITBC network and track their progress.
|
|
|
|
## Overview
|
|
|
|
The job submission workflow:
|
|
|
|
1. **Prepare** - Choose model and parameters
|
|
2. **Submit** - Send job to Coordinator API
|
|
3. **Queue** - Job enters the processing queue
|
|
4. **Execute** - Miner processes your job
|
|
5. **Complete** - Receive results and receipt
|
|
|
|
## Submission Methods
|
|
|
|
### Web Interface
|
|
|
|
1. Go to [Marketplace](https://aitbc.bubuit.net/marketplace/)
|
|
2. Select a service (e.g., "Text Generation", "Image Generation")
|
|
3. Enter your prompt and configure options
|
|
4. Click **Submit Job**
|
|
5. View job status in your dashboard
|
|
|
|
### CLI
|
|
|
|
```bash
|
|
# Basic submission
|
|
./aitbc-cli.sh submit "Explain machine learning in simple terms"
|
|
|
|
# With model selection
|
|
./aitbc-cli.sh submit "Generate a haiku about coding" --model llama3.2
|
|
|
|
# With parameters
|
|
./aitbc-cli.sh submit "Write a story" --model llama3.2 --max-tokens 500 --temperature 0.7
|
|
|
|
# Check job status
|
|
./aitbc-cli.sh status <job_id>
|
|
|
|
# List your jobs
|
|
./aitbc-cli.sh jobs
|
|
```
|
|
|
|
### Python SDK
|
|
|
|
```python
|
|
from aitbc_sdk import AITBCClient
|
|
|
|
client = AITBCClient(
|
|
api_url="https://aitbc.bubuit.net/api",
|
|
api_key="your-api-key" # Optional for authenticated requests
|
|
)
|
|
|
|
# Submit a text generation job
|
|
job = client.submit_job(
|
|
prompt="What is the capital of France?",
|
|
model="llama3.2",
|
|
params={
|
|
"max_tokens": 100,
|
|
"temperature": 0.5
|
|
}
|
|
)
|
|
|
|
print(f"Job ID: {job.id}")
|
|
print(f"Status: {job.status}")
|
|
|
|
# Wait for completion
|
|
result = client.wait_for_job(job.id, timeout=60)
|
|
print(f"Output: {result.output}")
|
|
```
|
|
|
|
### Direct API
|
|
|
|
```bash
|
|
# Submit job
|
|
curl -X POST https://aitbc.bubuit.net/api/v1/jobs \
|
|
-H "Content-Type: application/json" \
|
|
-d '{
|
|
"prompt": "Hello, world!",
|
|
"model": "llama3.2",
|
|
"params": {"max_tokens": 50}
|
|
}'
|
|
|
|
# Check status
|
|
curl https://aitbc.bubuit.net/api/v1/jobs/<job_id>
|
|
```
|
|
|
|
## Job Parameters
|
|
|
|
| Parameter | Type | Default | Description |
|
|
|-----------|------|---------|-------------|
|
|
| `prompt` | string | required | Input text or instruction |
|
|
| `model` | string | `llama3.2` | AI model to use |
|
|
| `max_tokens` | int | 256 | Maximum output tokens |
|
|
| `temperature` | float | 0.7 | Creativity (0.0-1.0) |
|
|
| `top_p` | float | 0.9 | Nucleus sampling |
|
|
| `stream` | bool | false | Stream output chunks |
|
|
|
|
## Available Models
|
|
|
|
| Model | Type | Use Case |
|
|
|-------|------|----------|
|
|
| `llama3.2` | Text | General chat, Q&A, writing |
|
|
| `llama3.2:1b` | Text | Fast, lightweight tasks |
|
|
| `codellama` | Code | Code generation, debugging |
|
|
| `stable-diffusion` | Image | Image generation |
|
|
|
|
## Job States
|
|
|
|
| State | Description |
|
|
|-------|-------------|
|
|
| `pending` | Job submitted, waiting for miner |
|
|
| `running` | Miner is processing the job |
|
|
| `completed` | Job finished successfully |
|
|
| `failed` | Job failed (see error message) |
|
|
| `cancelled` | Job was cancelled by user |
|
|
|
|
## Tracking Your Jobs
|
|
|
|
### View in Explorer
|
|
|
|
Visit [Explorer](https://aitbc.bubuit.net/explorer/) to see:
|
|
- Recent jobs and their status
|
|
- Your job history (if authenticated)
|
|
- Receipt details and proofs
|
|
|
|
### Programmatic Tracking
|
|
|
|
```python
|
|
# Poll for status
|
|
import time
|
|
|
|
while True:
|
|
job = client.get_job(job_id)
|
|
print(f"Status: {job.status}")
|
|
|
|
if job.status in ["completed", "failed", "cancelled"]:
|
|
break
|
|
|
|
time.sleep(2)
|
|
```
|
|
|
|
## Cancelling Jobs
|
|
|
|
```bash
|
|
# CLI
|
|
./aitbc-cli.sh cancel <job_id>
|
|
|
|
# API
|
|
curl -X POST https://aitbc.bubuit.net/api/v1/jobs/<job_id>/cancel
|
|
```
|
|
|
|
## Best Practices
|
|
|
|
1. **Be specific** - Clear prompts get better results
|
|
2. **Set appropriate limits** - Use `max_tokens` to control costs
|
|
3. **Handle errors** - Always check job status before using output
|
|
4. **Use streaming** - For long outputs, enable streaming for faster feedback
|
|
|
|
## Next Steps
|
|
|
|
- [Payments and Receipts](payments-receipts.md) - Understanding costs and proofs
|
|
- [Troubleshooting](troubleshooting.md) - Common issues and solutions
|