mirror of
https://github.com/harvard-edge/cs249r_book.git
synced 2026-04-29 17:20:21 -05:00
09602445de
- Vol1: chapter updates across backmatter, benchmarking, data, frameworks, etc. - Vol2: content updates, new appendices (assumptions, communication, fleet, reliability) - Quarto: config, styles, formulas, constants - Add SEMINAL_PAPERS_V2.md, learning_objectives_bolding_parallel.sh - VSCode extension: package.json, chapterNavigatorProvider - Landing page and docs updates
93 lines
4.2 KiB
Bash
Executable File
93 lines
4.2 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# Learning Objectives Bolding: Run Gemini CLI in parallel across all chapters
|
|
#
|
|
# Usage: From repo root, run:
|
|
# ./book/tools/scripts/learning_objectives_bolding_parallel.sh
|
|
#
|
|
# Requires: Gemini CLI installed (npm install -g @google/gemini-cli or brew install gemini-cli)
|
|
# Rate limits: Free tier is 60 req/min. Default MAX_PARALLEL=8 to stay under limit.
|
|
# Adjust with: MAX_PARALLEL=4 ./book/tools/scripts/learning_objectives_bolding_parallel.sh
|
|
|
|
set -e
|
|
|
|
# Repo root (script lives at book/tools/scripts/)
|
|
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
|
REPO_ROOT="$(cd "$SCRIPT_DIR/../../.." && pwd)"
|
|
cd "$REPO_ROOT"
|
|
|
|
# Limit concurrent Gemini invocations to avoid rate limits (60/min free tier)
|
|
MAX_PARALLEL="${MAX_PARALLEL:-2}"
|
|
|
|
# All chapters with Learning Objectives (exclude compute_infrastructure_expanded—draft)
|
|
CHAPTERS=(
|
|
"book/quarto/contents/vol1/introduction/introduction.qmd"
|
|
"book/quarto/contents/vol1/ml_systems/ml_systems.qmd"
|
|
"book/quarto/contents/vol1/data_engineering/data_engineering.qmd"
|
|
"book/quarto/contents/vol1/data_selection/data_selection.qmd"
|
|
"book/quarto/contents/vol1/nn_architectures/nn_architectures.qmd"
|
|
"book/quarto/contents/vol1/nn_computation/nn_computation.qmd"
|
|
"book/quarto/contents/vol1/training/training.qmd"
|
|
"book/quarto/contents/vol1/hw_acceleration/hw_acceleration.qmd"
|
|
"book/quarto/contents/vol1/optimizations/model_compression.qmd"
|
|
"book/quarto/contents/vol1/frameworks/frameworks.qmd"
|
|
"book/quarto/contents/vol1/benchmarking/benchmarking.qmd"
|
|
"book/quarto/contents/vol1/ml_ops/ml_ops.qmd"
|
|
"book/quarto/contents/vol1/model_serving/model_serving.qmd"
|
|
"book/quarto/contents/vol1/responsible_engr/responsible_engr.qmd"
|
|
"book/quarto/contents/vol1/ml_workflow/ml_workflow.qmd"
|
|
"book/quarto/contents/vol1/conclusion/conclusion.qmd"
|
|
"book/quarto/contents/vol1/backmatter/appendix_algorithm.qmd"
|
|
"book/quarto/contents/vol1/backmatter/appendix_data.qmd"
|
|
"book/quarto/contents/vol1/backmatter/appendix_machine.qmd"
|
|
"book/quarto/contents/vol1/backmatter/appendix_dam.qmd"
|
|
"book/quarto/contents/vol1/backmatter/appendix_assumptions.qmd"
|
|
"book/quarto/contents/vol2/introduction/introduction.qmd"
|
|
"book/quarto/contents/vol2/compute_infrastructure/compute_infrastructure.qmd"
|
|
"book/quarto/contents/vol2/network_fabrics/network_fabrics.qmd"
|
|
"book/quarto/contents/vol2/data_storage/data_storage.qmd"
|
|
"book/quarto/contents/vol2/distributed_training/distributed_training.qmd"
|
|
"book/quarto/contents/vol2/collective_communication/collective_communication.qmd"
|
|
"book/quarto/contents/vol2/fault_tolerance/fault_tolerance.qmd"
|
|
"book/quarto/contents/vol2/fleet_orchestration/fleet_orchestration.qmd"
|
|
"book/quarto/contents/vol2/inference/inference.qmd"
|
|
"book/quarto/contents/vol2/edge_intelligence/edge_intelligence.qmd"
|
|
"book/quarto/contents/vol2/ops_scale/ops_scale.qmd"
|
|
"book/quarto/contents/vol2/performance_engineering/performance_engineering.qmd"
|
|
"book/quarto/contents/vol2/sustainable_ai/sustainable_ai.qmd"
|
|
"book/quarto/contents/vol2/responsible_ai/responsible_ai.qmd"
|
|
"book/quarto/contents/vol2/robust_ai/robust_ai.qmd"
|
|
"book/quarto/contents/vol2/security_privacy/security_privacy.qmd"
|
|
"book/quarto/contents/vol2/conclusion/conclusion.qmd"
|
|
)
|
|
|
|
run_chapter() {
|
|
local file="$1"
|
|
local prompt="Read the instructions at .claude/docs/learning-objectives-gemini-instructions.md, then edit this chapter file to add bolding to Learning Objectives: $file"
|
|
echo "[$(date +%H:%M:%S)] Starting: $file"
|
|
if gemini -p "$prompt" --include-directories "book/quarto/contents,.claude/docs"; then
|
|
echo "[$(date +%H:%M:%S)] Done: $file"
|
|
else
|
|
echo "[$(date +%H:%M:%S)] FAILED: $file" >&2
|
|
return 1
|
|
fi
|
|
}
|
|
|
|
export -f run_chapter
|
|
export MAX_PARALLEL
|
|
|
|
echo "Learning Objectives Bolding: Running Gemini CLI on ${#CHAPTERS[@]} chapters (max $MAX_PARALLEL parallel)"
|
|
echo "Instructions: .claude/docs/learning-objectives-gemini-instructions.md"
|
|
echo ""
|
|
|
|
# Run in parallel with semaphore
|
|
for ch in "${CHAPTERS[@]}"; do
|
|
while [ "$(jobs -r 2>/dev/null | wc -l | tr -d ' ')" -ge "$MAX_PARALLEL" ]; do
|
|
sleep 2
|
|
done
|
|
run_chapter "$ch" &
|
|
done
|
|
|
|
wait
|
|
echo ""
|
|
echo "All chapters processed. Review changes with: git diff book/quarto/contents/"
|