|
|
--- |
|
|
source_datasets: |
|
|
- grimulkan/physical-reasoning |
|
|
task_categories: |
|
|
- text-generation |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
# grimulkan_physical-reasoning-gemini-2.0-flash-exp-ShareGPT |
|
|
|
|
|
[grimulkan/physical-reasoning](https://huggingface.co/datasets/grimulkan/physical-reasoning) with responses generated with `gemini-2.0-flash-exp`. |
|
|
|
|
|
## Generation Details |
|
|
|
|
|
- If `BlockedPromptException`, `StopCandidateException`, or `InvalidArgument` was returned, the sample was skipped. |
|
|
- If `["candidates"][0]["safety_ratings"] == "SAFETY"` the sample was skipped. |
|
|
- If `["candidates"][0]["finish_reason"] != 1` the sample was skipped. |
|
|
|
|
|
```py |
|
|
model = genai.GenerativeModel( |
|
|
model_name, |
|
|
safety_settings=[ |
|
|
{ |
|
|
"category": "HARM_CATEGORY_HARASSMENT", |
|
|
"threshold": "BLOCK_NONE", |
|
|
}, |
|
|
{ |
|
|
"category": "HARM_CATEGORY_HATE_SPEECH", |
|
|
"threshold": "BLOCK_NONE", |
|
|
}, |
|
|
{ |
|
|
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT", |
|
|
"threshold": "BLOCK_NONE", |
|
|
}, |
|
|
{ |
|
|
"category": "HARM_CATEGORY_DANGEROUS_CONTENT", |
|
|
"threshold": "BLOCK_NONE", |
|
|
}, |
|
|
] |
|
|
) |
|
|
``` |
|
|
|
|
|
## License |
|
|
|
|
|
Whatever the original dataset's license + Gemini responses would be is what this dataset uses. I add no extra restrictions. |
|
|
|
