Skip to content

ENVIRONMENT.md

Latest commit

 

History

History
102 lines (73 loc) · 2.54 KB

ENVIRONMENT.md

File metadata and controls

102 lines (73 loc) · 2.54 KB

Environment Variables Reference

Required API Keys

Set at least one API key for benchmark testing (Stage 5):

# OpenRouter (multi-model aggregator)
export OPENROUTER_API_KEY=sk-or-v1-...

# OpenAI
export OPENAI_API_KEY=sk-proj-...

# Anthropic (Claude)
export ANTHROPIC_API_KEY=sk-ant-api03-...

# xAI (Grok)
export XAI_API_KEY=xai-...

Or add to .env file:

echo "OPENROUTER_API_KEY=sk-or-v1-..." >> .env
echo "ANTHROPIC_API_KEY=sk-ant-api03-..." >> .env

Optional Variables

Generation Pipeline (Stage 1-4)

For --from-pdf study generation:

# Default provider (optional, default: gemini)
GOOGLE_API_KEY=AIzaSy...

# Or use CLI flags: --generation-provider openai --generation-model gpt-4o

Response Formatter

Controls the LLM used for formatting participant responses:

# Default: OpenRouter + DeepSeek
FORMATTER_PROVIDER=openrouter
FORMATTER_MODEL=deepseek/deepseek-v3.2

# Use Claude for formatting
FORMATTER_PROVIDER=anthropic
FORMATTER_MODEL=claude-3-haiku-20240307

# Use GPT-4o-mini for formatting
FORMATTER_PROVIDER=openai
FORMATTER_MODEL=gpt-4o-mini

Model Auto-detection

You don't need to specify provider - it's auto-detected from model name:

Model Name Provider API Key Required
mistralai/mistral-nemo OpenRouter OPENROUTER_API_KEY
deepseek/deepseek-v3.2 OpenRouter OPENROUTER_API_KEY
qwen/qwen-2.5-72b-instruct OpenRouter OPENROUTER_API_KEY
gpt-4o OpenAI OPENAI_API_KEY
gpt-4o-mini OpenAI OPENAI_API_KEY
claude-3-5-sonnet-20241022 Anthropic ANTHROPIC_API_KEY
claude-3-haiku-20240307 Anthropic ANTHROPIC_API_KEY
grok-2 xAI XAI_API_KEY

Full Example

# .env file
OPENROUTER_API_KEY=sk-or-v1-...
OPENAI_API_KEY=sk-proj-...
ANTHROPIC_API_KEY=sk-ant-api03-...
XAI_API_KEY=xai-...

# Optional: customize formatter (default is deepseek via OpenRouter)
FORMATTER_PROVIDER=openai
FORMATTER_MODEL=gpt-4o-mini

# Optional: for PDF study generation
GOOGLE_API_KEY=AIzaSy...

Then run benchmarks without additional config:

# Uses OPENROUTER_API_KEY (auto-detected)
python scripts/run_baseline_pipeline.py --study-id study_001 --real-llm --model mistralai/mistral-nemo

# Uses ANTHROPIC_API_KEY (auto-detected)
python scripts/run_baseline_pipeline.py --study-id study_001 --real-llm --model claude-3-5-sonnet-20241022

# Uses OPENAI_API_KEY (auto-detected)
python scripts/run_baseline_pipeline.py --study-id study_001 --real-llm --model gpt-4o