Download runs.jsonl from LayerLens/gmail-corrected-invoice-test: direct link, hf CLI and curl.
- Browser
- Download file 3.99 kB
-
https://huggingface.co/datasets/LayerLens/gmail-corrected-invoice-test/resolve/main/runs.jsonl
- Command line
-
hf download hf://datasets/LayerLens/gmail-corrected-invoice-test/runs.jsonl
-
curl -L -o runs.jsonl https://huggingface.co/datasets/LayerLens/gmail-corrected-invoice-test/resolve/main/runs.jsonl
3.99 kB
| {"run_key": "corrected-invoice-gpt-5-4-mini", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "GPT-5.4 Mini", "agent_id": "openai/gpt-5.4-mini-2026-03-17", "answer": null, "answer_contains_10625": true, "answer_contains_10475": false, "passed": true, "verdict": "pass", "tool_calls": 3, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-gemini-3-1-pro-preview", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "Gemini 3.1 Pro Preview", "agent_id": "google/gemini-3.1-pro-preview", "answer": null, "answer_contains_10625": true, "answer_contains_10475": false, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-claude-haiku-4-5", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "Claude Haiku 4.5", "agent_id": "anthropic/claude-haiku-4.5", "answer": null, "answer_contains_10625": true, "answer_contains_10475": true, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-claude-sonnet-5", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "Claude Sonnet 5", "agent_id": "anthropic/claude-sonnet-5", "answer": "The corrected amount for INV-2025-0003 is **10,625 USD** (up from the original 10,475 USD, due to a freight line that had been omitted). The due date remained 2025-08-26.", "answer_contains_10625": true, "answer_contains_10475": true, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages", "args": {"q": "INV-2025-0003"}, "result": "5 message IDs, all in 1 thread", "resultSizeEstimate": 5}, {"tool": "get_thread", "args": {"id": "zzz0000000000015"}, "result": "5 messages: original 10,475 USD (1), correction to 10,625 USD (3), buyer confirmation 10,625 USD (4)"}], "trace_events": 9, "input_tokens": 7940, "output_tokens": 176, "cost_usd": 0.01764, "stop_reason": "finished", "grading_evidence_text": "looked for \"10625\" in the agent's answer", "copy_matches_source": true} | |
| {"run_key": "corrected-invoice-gpt-5-5", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "GPT-5.5", "agent_id": "openai/gpt-5.5-20260423", "answer": null, "answer_contains_10625": true, "answer_contains_10475": false, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-gemini-3-8-flash", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "Gemini 3.8 Flash", "agent_id": "google/gemini-3.8-flash", "answer": null, "answer_contains_10625": true, "answer_contains_10475": true, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-claude-opus-5-5", "trace_record": "showcase copy of the original run trace", "task_id": "corrected-invoice-amount", "model": "Claude Opus 5.5", "agent_id": "anthropic/claude-opus-5.5", "answer": null, "answer_contains_10625": true, "answer_contains_10475": true, "passed": true, "verdict": "pass", "tool_calls": 2, "key_tool_calls": [{"tool": "search_messages"}, {"tool": "get_thread"}]} | |
| {"run_key": "corrected-invoice-claude-sonnet-5-5", "trace_record": "task drawer only (8th run added after the 7 showcase copies); trace not read", "task_id": "corrected-invoice-amount", "model": "Claude Sonnet 5.5", "agent_id": null, "answer": null, "answer_contains_10625": null, "answer_contains_10475": true, "passed": true, "verdict": "pass", "tool_calls": null, "key_tool_calls": null} | |