Datasets:
Download validation.json from zetomatoz/guidellm-agentic-coding-trajectories: direct link, hf CLI and curl.
- Browser
- Download file 1.45 kB
-
https://huggingface.co/datasets/zetomatoz/guidellm-agentic-coding-trajectories/resolve/main/validation.json
- Command line
-
hf download hf://datasets/zetomatoz/guidellm-agentic-coding-trajectories/validation.json
-
curl -L -o validation.json https://huggingface.co/datasets/zetomatoz/guidellm-agentic-coding-trajectories/resolve/main/validation.json
1.45 kB
| { | |
| "guidellm_commit": "a938aaa620045e3b90202f3b3a87f34f86958144", | |
| "scope": "offline mapper, finalizer, schemas, and request formatter; simulated responses", | |
| "endpoint_tested": false, | |
| "checks": [ | |
| "Default mapper recognizes only conversation_turns, ignoring numeric metadata", | |
| "All 630 rows finalize into valid linear full-history graphs", | |
| "All original per-step fixed text token targets are preserved exactly", | |
| "Tool calls omit ignore_eos, stop, and max_completion_tokens", | |
| "Text continuations keep ignore_eos=true and the original token target", | |
| "Tool results bind to fresh live response IDs and appear once per step", | |
| "Streaming tool argument fragments compile into complete JSON", | |
| "Original sample IDs, source metadata, and non-native prompts are unchanged" | |
| ], | |
| "sets": { | |
| "8turn": { | |
| "standard": 480, | |
| "formatted_requests": 960, | |
| "fixed_text_output_tokens": 98710, | |
| "client_tool_call": 240, | |
| "tool_response_injection": 240, | |
| "rows": 90 | |
| }, | |
| "24turn": { | |
| "standard": 7080, | |
| "formatted_requests": 14520, | |
| "fixed_text_output_tokens": 1889433, | |
| "client_tool_call": 3720, | |
| "tool_response_injection": 3720, | |
| "rows": 450 | |
| }, | |
| "48turn": { | |
| "standard": 2640, | |
| "formatted_requests": 6000, | |
| "fixed_text_output_tokens": 908668, | |
| "client_tool_call": 1680, | |
| "tool_response_injection": 1680, | |
| "rows": 90 | |
| } | |
| } | |
| } | |