synthetic-data-generator / .env.local.template
daqc's picture
Refactor environment templates for improved configuration flexibility
9cfb192
raw
history blame
1.82 kB
# =============================================================================
# LOCAL/API CONFIGURATION
# =============================================================================
# -----------------------------------------------------------------------------
# REQUIRED CONFIGURATION
# -----------------------------------------------------------------------------
# Hugging Face token (required for all setups)
HF_TOKEN=hf_...
# Generation Settings
MAX_NUM_TOKENS=2048
MAX_NUM_ROWS=1000
DEFAULT_BATCH_SIZE=5
# Required for chat data generation with Llama or Qwen models
# Options: "llama3", "qwen2", or custom template string
MAGPIE_PRE_QUERY_TEMPLATE=llama3
# -----------------------------------------------------------------------------
# A. CLOUD API SERVICES
# -----------------------------------------------------------------------------
# 1. HUGGING FACE INFERENCE API (Default, Recommended)
MODEL=meta-llama/Llama-3.1-8B-Instruct
# MODEL=Qwen/Qwen2.5-1.5B-Instruct
# 2. OPENAI API
# OPENAI_BASE_URL=https://api.openai.com/v1/
# MODEL=gpt-4
# API_KEY=sk-...
# 3. HUGGING FACE SPACE FOR ARGILLA (optional)
# ARGILLA_API_URL=https://your-space.hf.space/
# ARGILLA_API_KEY=your_key
# -----------------------------------------------------------------------------
# B. LOCAL SERVICES (Requires Installation)
# -----------------------------------------------------------------------------
# 1. LOCAL OLLAMA
# OLLAMA_BASE_URL=http://127.0.0.1:11434/
# MODEL=llama3.2:1b
# TOKENIZER_ID=meta-llama/Llama-3.2-1B-Instruct
# 2. LOCAL VLLM
# VLLM_BASE_URL=http://127.0.0.1:8000/
# MODEL=Qwen/Qwen2.5-1.5B-Instruct
# TOKENIZER_ID=Qwen/Qwen2.5-1.5B-Instruct
# 3. LOCAL TGI
# HUGGINGFACE_BASE_URL=http://127.0.0.1:3000/
# MODEL=meta-llama/Llama-3.1-8B-Instruct
# TOKENIZER_ID=meta-llama/Llama-3.1-8B-Instruct