messages stringlengths 14.6k 63.3k | src_task stringlengths 25 25 | model stringclasses 1
value | skill_level stringclasses 1
value | category stringclasses 1
value | contam_score float32 0 0 |
|---|---|---|---|---|---|
[{"role": "user", "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the foll... | data_processing_task_3216 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
[{"role": "user", "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the foll... | data_processing_task_3772 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
[{"role": "user", "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the foll... | data_processing_task_2591 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
[{"role": "user", "content": "You are an AI assistant tasked with solving command-line tasks in a Linux environment. You will be given a task description and the output from previously executed commands. Your goal is to solve the task by providing batches of shell commands.\n\nFormat your response as JSON with the foll... | data_processing_task_3939 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_4243 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_3150 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_0236 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_2764 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_2635 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
"[{\"role\": \"user\", \"content\": \"You are an AI assistant tasked with solving command-line tasks(...TRUNCATED) | data_processing_task_2429 | deepseek-ai/DeepSeek-V3.2 | easy | data_processing | 0 |
End of preview. Expand in Data Studio
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
tb2-sft-coldstart-decontam
Decontaminated Terminus-2 SFT cold-start trajectories from nvidia/Nemotron-Terminal-Corpus (DeepSeek-V3.2), with every held-out Terminal-Bench task removed by n-gram phrase containment + the TB canary GUID. messages is a JSON chat array (assistant turns = loss-masked SFT targets). See decontam_report.json for the contamination rate per held-out task.
- Downloads last month
- 38