06e48ef34b
- Add infrastructure.md and workflow files to .gitignore to prevent sensitive info leaks - Change blockchain node mempool backend default from memory to database for persistence - Refactor blockchain node logger with StructuredLogFormatter and AuditLogger (consistent with coordinator) - Add structured logging fields: service, module, function, line number - Unify coordinator config with Database
2.4 KiB
2.4 KiB
Job Submission Guide
Submit compute jobs to the AITBC network.
Basic Submission
aitbc client submit --model gpt2 --input data.txt --output results/
Options Reference
| Option | Required | Description |
|---|---|---|
--model |
Yes | Model to run (e.g., gpt2, llama, stable-diffusion) |
--input |
Yes | Input file or data |
--output |
Yes | Output directory |
--gpu |
No | GPU requirements (v100, a100, rtx3090) |
--gpu-count |
No | Number of GPUs (default: 1) |
--timeout |
No | Job timeout in seconds (default: 3600) |
--priority |
No | Job priority (low, normal, high) |
GPU Requirements
Single GPU
aitbc client submit --model gpt2 --input data.txt --gpu v100
Multiple GPUs
aitbc client submit --model llama --input data.txt --gpu a100 --gpu-count 4
Specific GPU Type
aitbc client submit --model stable-diffusion --input data.txt --gpu rtx3090
Input Methods
File Input
aitbc client submit --model gpt2 --input ./data/training_data.txt
Inline Input
aitbc client submit --model gpt2 --input "Hello, world!"
URL Input
aitbc client submit --model gpt2 --input https://example.com/data.txt
Output Options
Local Directory
aitbc client submit --model gpt2 --input data.txt --output ./results
S3 Compatible Storage
aitbc client submit --model gpt2 --input data.txt --output s3://my-bucket/results
Job Priority
| Priority | Speed | Cost |
|---|---|---|
| low | Standard | 1x |
| normal | Fast | 1.5x |
| high | Priority | 2x |
Examples
Training Job
aitbc client submit \
--model llama \
--input ./training_data.csv \
--output ./model_weights \
--gpu a100 \
--gpu-count 4 \
--timeout 7200 \
--priority high
Inference Job
aitbc client submit \
--model gpt2 \
--input ./prompts.txt \
--output ./outputs \
--gpu v100 \
--timeout 600
Batch Jobs
Submit multiple jobs at once:
# Using a job file
aitbc client submit-batch --file jobs.yaml
Example jobs.yaml:
jobs:
- model: gpt2
input: data1.txt
output: results1/
- model: gpt2
input: data2.txt
output: results2/
Next
- 3_job-lifecycle.md — Status, results, history, cancellation
- 5_pricing-billing.md — Cost structure and invoices