Instructions to use ZibinDong/ActionCodec2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ZibinDong/ActionCodec2 with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("ZibinDong/ActionCodec2", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 1,507 Bytes
1bfc3f0 | 1 2 3 4 5 6 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 | """Runtime opt-in controls for lossless fixed-vocabulary Set-BPE search."""
from __future__ import annotations
from dataclasses import dataclass
@dataclass(frozen=True)
class EncodingSearch:
'Independent post-encoding stages; defaults retain ranked merge replay.'
spatial_search: bool = False
temporal_search: bool = False
search_window: int = 32
search_beam: int = 4
def __post_init__(self):
for name in ("spatial_search", "temporal_search"):
if not isinstance(getattr(self, name), bool):
raise TypeError(f"{name} must be a bool")
for name in ("search_window", "search_beam"):
value = getattr(self, name)
if isinstance(value, bool) or not isinstance(value, int) or value < 1:
raise ValueError(f"{name} must be a positive integer")
if self.search_beam > 64:
raise ValueError("search_beam must be at most 64")
@property
def enabled(self):
return self.spatial_search or self.temporal_search
def validate_codec(self, scheme, bins):
if not self.enabled:
return
if scheme != "set_bpe":
raise ValueError("packing search requires set_bpe")
if len(bins) > 8 or max(bins) > 255:
raise ValueError("packing search supports at most eight slots and 255 bins per slot")
if self.search_window * len(bins) > 256:
raise ValueError("search_window times profile width must be at most 256")
|