tingao/synthetic-data-generator

1# =============================================================================2# DOCKER CONFIGURATION ONLY - FULL SETUP (APP + OLLAMA + ARGILLA)3# =============================================================================4 5# Note: Before building:6# 1. Copy this template to the root directory: cp docker/.env.docker.template .env7# 2. Comment/uncomment the sections you want to use (OLLAMA and/or ARGILLA)8# 3. Then build and run with the appropriate docker compose command9 10# Hugging Face token with read/write permissions11HF_TOKEN=your_token_here12 13# -----------------------------------------------------------------------------14# GENERATION SETTINGS15# -----------------------------------------------------------------------------16MAX_NUM_TOKENS=204817MAX_NUM_ROWS=100018DEFAULT_BATCH_SIZE=519 20# -----------------------------------------------------------------------------21# OLLAMA DOCKER CONFIGURATION22# -----------------------------------------------------------------------------23OLLAMA_BASE_URL=http://ollama:1143424OLLAMA_HARDWARE=latest # latest (for CPU/NVIDIA), rocm (for AMD)25 26# LLAMA 3.227MODEL=llama3.2:1b28TOKENIZER_ID=meta-llama/Llama-3.2-1B-Instruct29MAGPIE_PRE_QUERY_TEMPLATE=llama330 31# DEEPSEEK R132#MODEL=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B33#TOKENIZER_ID=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B34#MAGPIE_PRE_QUERY_TEMPLATE= "<｜begin▁of▁sentence｜>User: "35 36# -----------------------------------------------------------------------------37# ARGILLA DOCKER CONFIGURATION (persistent data)38# -----------------------------------------------------------------------------39ARGILLA_API_URL=http://argilla:690040ARGILLA_USERNAME=admin41ARGILLA_PASSWORD=admin123442ARGILLA_API_KEY=admin.123443ARGILLA_REINDEX_DATASET=1