-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy path.env.example
More file actions
60 lines (53 loc) · 2.18 KB
/
Copy path.env.example
File metadata and controls
60 lines (53 loc) · 2.18 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
# Wine in a Million configuration
# Create this file with:
# make init-env
# --- GCP --------------------------------------------------------------------
PROJECT_ID=wine-in-a-million-501121
REGION=us-central1
NAME=wine-in-a-million
AUTO_APPROVE=false
# Optional image tag override. By default make deploy uses backend-<source-hash>.
TAG=
# Local Kaggle CSV path used by make local and make deploy after download.
DATASET=backend/data/winemag-data-130k-v2.csv
# --- Embedding model / Hugging Face ------------------------------------------
# Public models often work without a token, but setting one avoids rate limits
# and is required for gated/private models.
EMBEDDING_MODEL=BAAI/bge-small-en-v1.5
HUGGINGFACE_TOKEN=
# --- Kaggle dataset download -------------------------------------------------
# Create this at https://www.kaggle.com/settings/api
# It is only needed if Kaggle local auth is not already configured.
KAGGLE_API_TOKEN=
# Legacy credentials may work with older Kaggle clients; prefer KAGGLE_API_TOKEN.
KAGGLE_USERNAME=
KAGGLE_KEY=
KAGGLE_DATASET=zynicide/wine-reviews
KAGGLE_OUT_DIR=backend/data
# --- Cloudflare Workers frontend --------------------------------------------
# Use `npx --yes wrangler@4.22.0 login`, or set a token here.
# Token needs Workers Scripts edit/deploy permissions for your account.
CLOUDFLARE_API_TOKEN=
WORKER_NAME=wine-in-a-million
# --- Required streaming LLM recommendations ---------------------------------
# For OpenAI, leave LLM_BASE_URL blank (the SDK defaults to https://api.openai.com/v1).
OPENAI_API_KEY=
LLM_API_KEY_SECRET=wine-llm-api-key
LLM_MODEL=gpt-4o-mini
LLM_BASE_URL=
# --- Cloud Run tuning ---------------------------------------------------------
BACKEND_MIN_INSTANCES=0
BACKEND_MAX_INSTANCES=5
BACKEND_MEMORY=2Gi
BACKEND_CPU=2
# Cloud Run requires linux/amd64 images. Keep this unless deploying elsewhere.
BACKEND_PLATFORM=linux/amd64
# Ingest can take a long time for the full Kaggle dataset.
# Default: run embeddings locally and write directly to the production GCS LanceDB bucket.
# Set INGEST_MODE=cloud to use the Cloud Run Job fallback.
RUN_INGEST=true
INGEST_MODE=local
WAIT_FOR_INGEST=false
FORCE_REEMBED=false
INGEST_TIMEOUT=7200s
BATCH_SIZE=64