fix(ci): train-dqn uses compile-and-train (no pre-built binary dependency)
Was referencing training-pipeline which needs binary-tag for pre-built binaries from GitLab packages. Now references compile-and-train which compiles from source then trains. No stale binary risk. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
@@ -1,11 +1,11 @@
|
||||
# Train DQN: fetch binaries → hyperopt → train → evaluate → upload.
|
||||
# Train DQN: compile → hyperopt → train → evaluate → upload.
|
||||
#
|
||||
# Usage:
|
||||
# argo submit --from=wftmpl/train-dqn -p binary-tag=dev-abc1234
|
||||
# argo submit --from=wftmpl/train-dqn -p binary-tag=latest -p hyperopt-trials=50
|
||||
# argo submit --from=wftmpl/train-dqn
|
||||
# argo submit --from=wftmpl/train-dqn -p commit-ref=my-branch
|
||||
#
|
||||
# Thin wrapper around training-pipeline (templateRef) with DQN-tuned defaults.
|
||||
# DQN has a 28D hyperopt parameter space — more trials recommended.
|
||||
# Thin wrapper around compile-and-train with DQN-tuned defaults.
|
||||
# Compiles from source (no pre-built binary dependency).
|
||||
---
|
||||
apiVersion: argoproj.io/v1alpha1
|
||||
kind: WorkflowTemplate
|
||||
@@ -15,18 +15,14 @@ metadata:
|
||||
labels:
|
||||
app.kubernetes.io/name: train-dqn
|
||||
app.kubernetes.io/part-of: foxhunt
|
||||
app.kubernetes.io/component: gpu-test
|
||||
spec:
|
||||
entrypoint: pipeline
|
||||
serviceAccountName: argo-workflow
|
||||
podMetadata:
|
||||
labels:
|
||||
app.kubernetes.io/part-of: foxhunt
|
||||
app.kubernetes.io/component: gpu-test
|
||||
app.kubernetes.io/component: training
|
||||
app.kubernetes.io/component: compile-and-train
|
||||
securityContext:
|
||||
runAsUser: 1000
|
||||
runAsGroup: 1000
|
||||
fsGroup: 1000
|
||||
ttlStrategy:
|
||||
secondsAfterCompletion: 3600
|
||||
@@ -36,16 +32,16 @@ spec:
|
||||
parameters:
|
||||
- name: model
|
||||
value: dqn
|
||||
- name: binary-tag
|
||||
value: latest
|
||||
- name: commit-ref
|
||||
value: main
|
||||
- name: gpu-pool
|
||||
value: ci-training-h100
|
||||
- name: hyperopt-trials
|
||||
value: "50" # 24D param space needs full exploration
|
||||
value: "50"
|
||||
- name: hyperopt-epochs
|
||||
value: "20" # 20 epochs per trial with early stopping
|
||||
value: "20"
|
||||
- name: train-epochs
|
||||
value: "200" # matches dqn-production.toml
|
||||
value: "200"
|
||||
- name: symbol
|
||||
value: ES.FUT
|
||||
- name: data-dir
|
||||
@@ -67,47 +63,10 @@ spec:
|
||||
- name: cuda-compute-cap
|
||||
value: "90"
|
||||
|
||||
volumes:
|
||||
- name: training-data
|
||||
persistentVolumeClaim:
|
||||
claimName: training-data-pvc
|
||||
readOnly: true
|
||||
|
||||
volumeClaimTemplates:
|
||||
- metadata:
|
||||
name: workspace
|
||||
spec:
|
||||
accessModes: ["ReadWriteOnce"]
|
||||
storageClassName: scw-bssd
|
||||
resources:
|
||||
requests:
|
||||
storage: 5Gi
|
||||
|
||||
templates:
|
||||
- name: pipeline
|
||||
dag:
|
||||
tasks:
|
||||
- name: fetch-binary
|
||||
steps:
|
||||
- - name: compile-and-train
|
||||
templateRef:
|
||||
name: training-pipeline
|
||||
template: fetch-binary
|
||||
- name: hyperopt
|
||||
templateRef:
|
||||
name: training-pipeline
|
||||
template: hyperopt
|
||||
dependencies: [fetch-binary]
|
||||
- name: train-best
|
||||
templateRef:
|
||||
name: training-pipeline
|
||||
template: train-best
|
||||
dependencies: [hyperopt]
|
||||
- name: evaluate
|
||||
templateRef:
|
||||
name: training-pipeline
|
||||
template: evaluate
|
||||
dependencies: [train-best]
|
||||
- name: upload-results
|
||||
templateRef:
|
||||
name: training-pipeline
|
||||
template: upload-results
|
||||
dependencies: [evaluate]
|
||||
name: compile-and-train
|
||||
template: pipeline
|
||||
|
||||
Reference in New Issue
Block a user