Acme Learning/

Dashboard

Practice Lab

Tracks A–Gdemo

Real scenarios graded by AI — write prompts, build pipelines, debug agents. The AI grader scores every attempt.

live grading
Active scenarios
12
+3
Across 6 AI domains
Avg AI grade
81%
+4
Last 30 days
Completed today
3
done
Daily streak: 14 days
Grader model
sixfactors-grader-v3
live
Auto-rubric · multi-turn aware
Domain
AFoundations
Graded
Claude vs OpenAI prompt comparison
Correctness · cost · latency · structured output adherence
Turns4 / 4
grader
Claude wins on JSON-schema fidelity; GPT-4o wins on cost. Pick by task profile.
AI grade92
BAgents
In progress
Research agent — Claude tool use
Tool selection · planning · recovery · cost / answer
Turns6 / 12
grader
Plan loop is solid, but you skipped the verification step. Add a re-check before final answer.
Not yet graded
CRAG
Graded
RAG retrieval bench — noisy corpus
Recall@5 · MRR · faithfulness · citation accuracy
Turns8 / 8
grader
Recall@5 = 0.81 (target 0.80). Latency p95 over budget — try smaller chunks.
AI grade87
DVoice + Text
Available
Voice agent — appointment booking
Turn-taking · barge-in recovery · slot accuracy · latency
Turns0 / 14
grader
Real-time grading — speak naturally, the AI scores latency and clarity.
Not yet graded
EData + Tools
In progress
CRM agent via MCP — scoped access
Tool ladder · scope · audit · graceful failure
Turns3 / 8
grader
You read the right contact but skipped the audit log. Wire it before going further.
Not yet graded
FEvals
Needs rework
Capstone eval suite authoring
Coverage · golden quality · regression sensitivity
Turns5 / 10
grader
Goldens skew positive — add 4 adversarial examples and re-run.
AI grade42
GFine-Tune
Available
OpenAI fine-tune uplift drill
Dataset quality · training cost · uplift vs strong prompt
Turns0 / 6
grader
Bring a small task-specific dataset — model trains during morning lab, eval runs in the afternoon.
Not yet graded