Training run - 2026-05-26 23:13:26
Some checks failed
QMB Training / QMB-Training (push) Failing after 3m23s

This commit is contained in:
J jonny 2026-05-26 23:13:26 -04:00
parent f1794acd81
commit 95aa6b80b7
2 changed files with 71 additions and 6 deletions

View File

@ -29,9 +29,8 @@ export:
use_dynamo: true use_dynamo: true
use_onnxsim: false use_onnxsim: false
model: model:
name: iq_tiny_cnn name: VTCNN1
params: params: {}
hidden_channels: 16
optimization: optimization:
loss: loss:
name: cross_entropy name: cross_entropy
@ -47,7 +46,8 @@ runtime:
amp_enabled: false amp_enabled: false
autocast_dtype: float32 autocast_dtype: float32
checkpoint_every_n_epochs: 1 checkpoint_every_n_epochs: 1
component_modules: [] component_modules:
- /opt/qmb/riahub/model/src/qmb/models/RadioML/VTCNN1.py
device: auto device: auto
epochs: 1 epochs: 1
progress_bar: false progress_bar: false

View File

@ -15,7 +15,7 @@ permissions:
jobs: jobs:
QMB-Training: QMB-Training:
runs-on: "ubuntu-latest-2080" runs-on: "whitehorse-p40-qmb"
env: env:
RIAHUB_BASE_URL: ${{ vars.RIAHUB_BASE_URL || secrets.RIAHUB_BASE_URL || '' }} RIAHUB_BASE_URL: ${{ vars.RIAHUB_BASE_URL || secrets.RIAHUB_BASE_URL || '' }}
QMB_OUTPUT_ROOT: "/opt/qmb/outputs" QMB_OUTPUT_ROOT: "/opt/qmb/outputs"
@ -39,6 +39,71 @@ jobs:
fi fi
- name: Download Selected Model
env:
RIAHUB_USER: ${{ secrets.QMBDEMO_USER }}
RIAHUB_TOKEN: ${{ secrets.QMBDEMO_TOKEN }}
run: |
set -euo pipefail
DEFAULT_BASE_URL="https://riahub.ai"
BASE_URL_SOURCE=${RIAHUB_BASE_URL:-$DEFAULT_BASE_URL}
BASE_URL_SOURCE="${BASE_URL_SOURCE%/}"
build_base_candidates() {
local raw="$1"
if [[ "$raw" =~ ^https?:// ]]; then
echo "$raw"
if [[ "$raw" == http://* ]]; then
echo "https://${raw#http://}"
elif [[ "$raw" == https://* ]]; then
echo "http://${raw#https://}"
fi
return
fi
echo "https://$raw"
echo "http://$raw"
}
REPO_PATH="/qoherent/qmb-mvp.git"
REL_PATH="src/qmb/models/RadioML/VTCNN1.py"
REF="db9dffb7d382993cf46f392c34a47e4cdd59585e"
DEST_PATH="/opt/qmb/riahub/model/src/qmb/models/RadioML/VTCNN1.py"
TMP_DIR=$(mktemp -d)
cleanup() { sudo rm -rf "$TMP_DIR"; }
trap cleanup EXIT
mapfile -t BASE_CANDIDATES < <(build_base_candidates "$BASE_URL_SOURCE")
CLONED=0
for base in "${BASE_CANDIDATES[@]}"; do
base="${base%/}"
REPO_URL="${base}${REPO_PATH}"
AUTHED_URL=$(printf '%s' "$REPO_URL" | sed -E "s#^(https?://)#\\1${RIAHUB_USER}:${RIAHUB_TOKEN}@#")
echo "Cloning model repo from $REPO_URL"
sudo rm -rf "$TMP_DIR"
if sudo git clone --filter=blob:none --no-checkout "$AUTHED_URL" "$TMP_DIR"; then
CLONED=1
break
fi
done
if [[ "$CLONED" -ne 1 ]]; then
echo "Failed to clone model repo using base URL candidates derived from: $BASE_URL_SOURCE" >&2
exit 1
fi
if ! command -v git-lfs >/dev/null 2>&1; then
sudo apt-get update -y
sudo apt-get install -y git-lfs
fi
sudo git -C "$TMP_DIR" lfs install --local || true
sudo git -C "$TMP_DIR" sparse-checkout init --no-cone
sudo git -C "$TMP_DIR" sparse-checkout set --no-cone -- "$REL_PATH"
sudo git -C "$TMP_DIR" fetch --depth=1 origin "$REF"
sudo git -C "$TMP_DIR" -c advice.detachedHead=false checkout FETCH_HEAD
sudo git -C "$TMP_DIR" lfs fetch origin --include="$REL_PATH" --exclude="" || true
sudo git -C "$TMP_DIR" lfs checkout || true
sudo mkdir -p "$(dirname "$DEST_PATH")"
sudo cp -f "$TMP_DIR/$REL_PATH" "$DEST_PATH"
sudo git -C "$TMP_DIR" remote remove origin || true
sudo git -C "$TMP_DIR" config --local --unset-all http.extraheader || true
- name: Checkout Training Dataset - name: Checkout Training Dataset
env: env:
RIAHUB_USER: ${{ secrets.QMBDEMO_USER }} RIAHUB_USER: ${{ secrets.QMBDEMO_USER }}
@ -317,4 +382,4 @@ jobs:
path: ${{ github.workspace }}/.riahub/artifacts/training path: ${{ github.workspace }}/.riahub/artifacts/training
if-no-files-found: warn if-no-files-found: warn
# committed at 2026-05-27T02:54:53.931581+00:00 # committed at 2026-05-27T03:13:26.168613+00:00