Delete configuration_olmo.py
Browse files- configuration_olmo.py +0 -52
configuration_olmo.py
DELETED
|
@@ -1,52 +0,0 @@
|
|
| 1 |
-
"""
|
| 2 |
-
OLMo configuration
|
| 3 |
-
"""
|
| 4 |
-
|
| 5 |
-
from transformers import AutoConfig, PretrainedConfig
|
| 6 |
-
from transformers.utils import logging
|
| 7 |
-
|
| 8 |
-
from .config import ModelConfig
|
| 9 |
-
from .aliases import PathOrStr
|
| 10 |
-
from .beam_search import Sampler
|
| 11 |
-
from .exceptions import OLMoError
|
| 12 |
-
from .initialization import ModuleType
|
| 13 |
-
from .optim import Optimizer
|
| 14 |
-
from .util import StrEnum
|
| 15 |
-
from .safetensors_util import STKey
|
| 16 |
-
from .torch_util import seed_all
|
| 17 |
-
|
| 18 |
-
logger = logging.get_logger(__name__)
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
class OLMoConfig(PretrainedConfig):
|
| 22 |
-
model_type = "olmo"
|
| 23 |
-
keys_to_ignore_at_inference = ["past_key_values"] # TODO: confirm
|
| 24 |
-
|
| 25 |
-
def __init__(self, use_cache: bool = False, **kwargs):
|
| 26 |
-
model_config = ModelConfig()
|
| 27 |
-
all_kwargs = model_config.asdict()
|
| 28 |
-
all_kwargs.update(kwargs)
|
| 29 |
-
all_kwargs.update({"use_cache": use_cache})
|
| 30 |
-
all_kwargs.update(
|
| 31 |
-
{
|
| 32 |
-
"architectures": all_kwargs.get("architectures", ["OLMoModelForCausalLM"])
|
| 33 |
-
or ["OLMoModelForCausalLM"]
|
| 34 |
-
}
|
| 35 |
-
)
|
| 36 |
-
super().__init__(**all_kwargs)
|
| 37 |
-
|
| 38 |
-
@property
|
| 39 |
-
def num_attention_heads(self):
|
| 40 |
-
return self.n_heads
|
| 41 |
-
|
| 42 |
-
@property
|
| 43 |
-
def num_hidden_layers(self):
|
| 44 |
-
return self.n_layers
|
| 45 |
-
|
| 46 |
-
@property
|
| 47 |
-
def hidden_size(self):
|
| 48 |
-
return self.d_model
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
# Register the config class so that it is available for transformer pipelines, auto-loading etc.
|
| 52 |
-
# AutoConfig.register("olmo", OLMoConfig)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|