Get all 3 datasets at 25% discount. 924 trajectories total.

WHAT'S INSIDE:

• dataset_pass.jsonl — 551 verified PASS trajectories ($100 value)

• dataset_reasoning.jsonl — 196 CoT reasoning trajectories ($75 value)

• dataset_premium.jsonl — 177 premium PASS + CoT ($150 value)

Total value: $325. Bundle price: $250. You save $75.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

ABOUT THE DATASETS:

All trajectories were generated by an autonomous code/data agent running in a sandboxed Python environment (E2B), and verified by an LLM-as-judge pipeline.

Common properties:

• Tool calls: run_python, read_file, write_file, list_files

• Format: JSONL, ATIF-v0.2 schema

• Avg 4 steps per trajectory

• Categories: algorithms, data_processing, debugging, math_and_stats, file_operations, regex_and_text, testing, utilities

• Commercial license (use in your business, don't resell the file)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

WHO IT'S FOR:

• ML engineers fine-tuning code agents

• Research teams working on agentic reasoning

• Companies building code assistant products

• Anyone who needs high-quality, verified agent trajectories at scale

Perfect for SFT (supervised fine-tuning), CoT distillation, tool-use research, and agent evaluation benchmarks.