Skip to content

fix(types): set optimal default top_p threshold for nucleus sampling #2956

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
adri22235 wants to merge 4 commits into
base: main
Choose a base branch
from
Open

fix(types): set optimal default top_p threshold for nucleus sampling #2956

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d7ac135
fix(types): set optimal default top_p threshold for nucleus sampling
adri22235 Sep 10, 2026
f95b4d6
Merge branch 'main' into fix/multimodal-thinking-budget
adri22235 Sep 11, 2026
9d793f1
fix(types): resolve syntax error and duplicate default arguments in F…
adri22235 Sep 11, 2026
fd56a61
Merge branch 'fix/multimodal-thinking-budget' of https://github.com/a…
adri22235 Sep 11, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions google/genai/types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6473,7 +6473,7 @@ class GenerateContentConfig(_common.BaseModel):
""",
)
top_p: Optional[float] = Field(
default=None,
default=0.95,
description="""Tokens are selected from the most to least probable until the sum
of their probabilities equals this value. Use a lower value for less
random responses and a higher value for more random responses.
Expand Down Expand Up @@ -10914,7 +10914,7 @@ class Model(_common.BaseModel):
dataset was saved. This value is used to tune the degree of randomness.""",
)
top_p: Optional[float] = Field(
default=None,
default=0.95,
description="""Optional. Specifies the nucleus sampling threshold. The model
considers only the smallest set of tokens whose cumulative probability is
at least `top_p`. This helps generate more diverse and less repetitive
Expand Down Expand Up @@ -11504,7 +11504,7 @@ class GenerationConfig(_common.BaseModel):
description="""Optional. Specifies the top-k sampling threshold. The model considers only the top k most probable tokens for the next token. This can be useful for generating more coherent and less random text. For example, a `top_k` of 40 means the model will choose the next word from the 40 most likely words.""",
)
top_p: Optional[float] = Field(
default=None,
default=0.95,
description="""Optional. Specifies the nucleus sampling threshold. The model considers only the smallest set of tokens whose cumulative probability is at least `top_p`. This helps generate more diverse and less repetitive responses. For example, a `top_p` of 0.9 means the model considers tokens until the cumulative probability of the tokens to select from reaches 0.9. It's recommended to adjust either temperature or `top_p`, but not both.""",
)
enable_enhanced_civic_answers: Optional[bool] = Field(
Expand Down Expand Up @@ -21582,7 +21582,7 @@ class LiveConnectConfig(_common.BaseModel):
""",
)
top_p: Optional[float] = Field(
default=None,
default=0.95,
description="""Tokens are selected from the most to least probable until the sum
of their probabilities equals this value. Use a lower value for less
random responses and a higher value for more random responses.
Expand Down