Skip to content

TESTING_GUIDELINES.md

Latest commit

 

History

History
38 lines (33 loc) · 1.48 KB

TESTING_GUIDELINES.md

File metadata and controls

38 lines (33 loc) · 1.48 KB
Raw

How to Test Tool Calling Models on Featherless

1. Prerequisites

  • FEATHERLESS_API_KEY in .env or environment
  • Access to models (some are gated)
  • Project setup: pnpm install

2. Basic Connectivity Test

First verify the model is deployed and reachable:

curl -s -X POST "https://api.featherless.ai/v1/chat/completions" \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"MODEL_ID","messages":[{"role":"user","content":"Hello"}],"max_tokens":20}' | head -50
  • 200 OK with response → model exists
  • 500/404 → not deployed or no access

3. Tool Calling Format Detection

Test with a simple tool call request:

curl -s -X POST "https://api.featherless.ai/v1/chat/completions" \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MODEL_ID",
    "messages": [{"role":"user","content":"Run: echo hello"}],
    "tools": [{"type":"function","function":{"name":"bash","description":"Run bash","parameters":{"type":"object","properties":{"command":{"type":"string"}}},"required":["command"]}}],
    "max_tokens": 150,
    "temperature": 0.7
  }' | python3 -c "import json,sys; d=json.load(sys.stdin); msg=d['choices'][0]['message']; print('tool_calls:', msg.get('tool_calls')); print('content:', msg.get('content','')[:200])"

Check response:

  • tool_calls array present → native format ✅
  • Content contains <function> → add to MODELS_NEED_TOOL_CALL_PARSING
  • Content contains `