| # okto_version: "1.2"
|
| PROJECT "StabilityTrainingExample"
|
| DESCRIPTION "Demonstrates STABILITY block for safe training"
|
|
|
| ENV {
|
| accelerator: "gpu"
|
| min_memory: "8GB"
|
| precision: "fp16"
|
| }
|
|
|
| DATASET {
|
| train: "examples/datasets/demo_train.jsonl"
|
| validation: "examples/datasets/demo_train.jsonl"
|
| format: "jsonl"
|
| type: "chat"
|
| }
|
|
|
| MODEL {
|
| name: "stable-model"
|
| base: "oktoseek/base-mini"
|
| device: "cuda"
|
| }
|
|
|
| TRAIN {
|
| epochs: 20
|
| batch_size: 32
|
| learning_rate: 0.0001
|
| device: "cuda"
|
| }
|
|
|
| STABILITY {
|
| stop_if_nan: true
|
| stop_if_diverges: true
|
| min_improvement: 0.001
|
| }
|
|
|
| CONTROL {
|
| on_nan {
|
| STOP_TRAINING
|
| LOG "NaN detected, stopping training"
|
| }
|
|
|
| IF loss > 10.0 {
|
| STOP_TRAINING
|
| LOG "Loss diverged, stopping training"
|
| }
|
| }
|
|
|
| MONITOR {
|
| metrics: ["loss", "val_loss"]
|
| notify_if {
|
| loss > 5.0
|
| }
|
| }
|
|
|
| EXPORT {
|
| format: ["okm"]
|
| path: "export/"
|
| }
|
|
|
|
|
