Skip to content

LMActionConfig

dendron.configs.lm_action_config.LMActionConfig dataclass

Configuration for LM action nodes (generate_action, loglikelihood_action, and loglikelihood_rolling_action).

Controls LM generation and output processes as well as blackboard interactions.

Args:

input_key (Optional[str]):
    The blackboard key used to read the input text that the model will process.
    Defaults to "in".

output_key (Optional[str]):
    The blackboard key where the model's output will be written.
    Defaults to "out".

max_new_tokens (Optional[int]):
    Maximum number of tokens to generate in one forward pass.
    Defaults to 16.

temperature (Optional[float]):
    Sampling temperature for generation. Higher values (e.g., 0.8) make the output
    more random, while lower values (e.g., 0.2) make it more deterministic.
    Defaults to 0.0 (deterministic).

truncation (Optional[bool]):
    Whether to truncate input sequences that exceed the model's maximum length.
    Defaults to False.

max_length (Optional[int]):
    Maximum allowed length for the combined input and output sequence.
    If None, will use the model's default max length. Defaults to None.

prefix_token_id (Optional[int]):
    Token ID to be prepended to all inputs. If specified, this token will be
    added before the input sequence. Defaults to None.

batch_size (Optional[Union[int, str]]):
    Batch size for processing multiple sequences. Use -1 for automatic batching.
    Defaults to -1.

max_batch_size (Optional[int]):
    Maximum allowed batch size when using automatic batching.
    Defaults to 1024.
Source code in src/dendron/configs/lm_action_config.py 
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
@dataclass
class LMActionConfig:
    """
    Configuration for LM action nodes (generate_action, loglikelihood_action, and 
    loglikelihood_rolling_action).

    Controls LM generation and output processes as well as blackboard interactions.

    Args:

        input_key (Optional[str]):
            The blackboard key used to read the input text that the model will process.
            Defaults to "in".

        output_key (Optional[str]):
            The blackboard key where the model's output will be written.
            Defaults to "out".

        max_new_tokens (Optional[int]):
            Maximum number of tokens to generate in one forward pass.
            Defaults to 16.

        temperature (Optional[float]):
            Sampling temperature for generation. Higher values (e.g., 0.8) make the output
            more random, while lower values (e.g., 0.2) make it more deterministic.
            Defaults to 0.0 (deterministic).

        truncation (Optional[bool]):
            Whether to truncate input sequences that exceed the model's maximum length.
            Defaults to False.

        max_length (Optional[int]):
            Maximum allowed length for the combined input and output sequence.
            If None, will use the model's default max length. Defaults to None.

        prefix_token_id (Optional[int]):
            Token ID to be prepended to all inputs. If specified, this token will be
            added before the input sequence. Defaults to None.

        batch_size (Optional[Union[int, str]]):
            Batch size for processing multiple sequences. Use -1 for automatic batching.
            Defaults to -1.

        max_batch_size (Optional[int]):
            Maximum allowed batch size when using automatic batching.
            Defaults to 1024.
    """
    _node_name: str
    _input_key: Optional[str] = field(default="in")
    _output_key: Optional[str] = field(default="out")
    _max_new_tokens: Optional[int] = field(default=16)
    _completions_key: Optional[str] = field(default="completions")
    _temperature: Optional[float] = field(default=0.0)
    _truncation: Optional[bool] = field(default=False)
    _max_length: Optional[int] = field(default=None)
    _prefix_token_id: Optional[int] = field(default=None)
    _batch_size: Optional[Union[int, str]] = field(default=-1)
    _max_batch_size: Optional[int] = field(default=1024)

    def __init__(
        self,
        node_name: str,
        input_key: Optional[str] = "in",
        output_key: Optional[str] = "out",
        completions_key: Optional[str] = "completions",
        max_new_tokens: Optional[int] = 16,
        temperature: Optional[float] = 0.0,
        truncation: Optional[bool] = False,
        max_length: Optional[int] = None,
        prefix_token_id: Optional[int] = None,
        batch_size: Optional[Union[int, str]] = -1,
        max_batch_size: Optional[int] = 1024,
    ):
        self._node_name = node_name
        self._input_key = input_key
        self._output_key = output_key
        self._completions_key = completions_key
        self._max_new_tokens = max_new_tokens
        self._temperature = temperature
        self._truncation = truncation
        self._max_length = max_length
        self._prefix_token_id = prefix_token_id
        self._batch_size = batch_size
        self._max_batch_size = max_batch_size

    @property
    def node_name(self):
        return self._node_name

    @property
    def input_key(self):
        return self._input_key

    @property
    def output_key(self):
        return self._output_key

    @property
    def completions_key(self):
        return self._completions_key

    @property
    def max_new_tokens(self):
        return self._max_new_tokens

    @property
    def temperature(self):
        return self._temperature

    @property
    def truncation(self):
        return self._truncation

    @property
    def max_length(self):
        return self._max_length

    @property
    def prefix_token_id(self):
        return self._prefix_token_id

    @property
    def batch_size(self):
        return self._batch_size

    @property
    def max_batch_size(self):
        return self._max_batch_size

    def to_dict(self) -> dict:
        return {
            "node_name": self.node_name,
            "input_key": self.input_key,
            "output_key": self.output_key,
            "completions_key": self.completions_key,
            "max_new_tokens": self.max_new_tokens,
            "temperature": self.temperature,
            "truncation": self.truncation,
            "max_length": self.max_length,
            "prefix_token_id": self.prefix_token_id,
            "batch_size": self.batch_size,
            "max_batch_size": self.max_batch_size,
        }

node_name property

input_key property

output_key property

max_new_tokens property

temperature property

truncation property

max_length property

prefix_token_id property

batch_size property

max_batch_size property

to_dict()

Source code in src/dendron/configs/lm_action_config.py 
136
137
138
139
140
141
142
143
144
145
146
147
148
149
def to_dict(self) -> dict:
    return {
        "node_name": self.node_name,
        "input_key": self.input_key,
        "output_key": self.output_key,
        "completions_key": self.completions_key,
        "max_new_tokens": self.max_new_tokens,
        "temperature": self.temperature,
        "truncation": self.truncation,
        "max_length": self.max_length,
        "prefix_token_id": self.prefix_token_id,
        "batch_size": self.batch_size,
        "max_batch_size": self.max_batch_size,
    }