File size: 3,700 Bytes
8dc2fc9 |
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 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 |
""" HelpingAI model configuration"""
from transformers import PretrainedConfig
from transformers.utils import logging
logger = logging.get_logger(__name__)
class HelpingAIConfig(PretrainedConfig):
model_type = "HelpingAI"
keys_to_ignore_at_inference = ["past_key_values"]
def __init__(
self,
vocab_size=50281,
hidden_size=2560,
num_hidden_layers=32,
num_attention_heads=32,
head_dim=256,
num_local_experts=8,
num_experts_per_tok=2,
intermediate_size=6912,
hidden_act="silu",
hidden_dropout=0.0,
attention_dropout=0.0,
classifier_dropout=0.1,
max_position_embeddings=4096,
initializer_range=0.02,
rms_norm_eps=1e-6,
layer_norm_eps=1e-5,
use_cache=False,
bos_token_id=50278,
eos_token_id=50279,
pad_token_id=50279,
tie_word_embeddings=False,
rope_pct=0.25,
rope_theta=10000,
partial_rotary_factor=0.25,
use_qkv_bias=False,
output_router_logits=False,
router_aux_loss_coef=0.02,
**kwargs,
):
super().__init__(bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)
self.vocab_size = vocab_size
self.max_position_embeddings = max_position_embeddings
self.hidden_size = hidden_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.head_dim = head_dim
self.num_local_experts = num_local_experts
self.num_experts_per_tok = num_experts_per_tok
self.intermediate_size = intermediate_size
self.hidden_act = hidden_act
self.hidden_dropout = hidden_dropout
self.attention_dropout = attention_dropout
self.classifier_dropout = classifier_dropout
self.initializer_range = initializer_range
self.rms_norm_eps = rms_norm_eps
self.layer_norm_eps = layer_norm_eps
self.use_cache = use_cache
self.tie_word_embeddings = tie_word_embeddings
self.rope_pct = rope_pct
self.rope_theta = rope_theta
self.partial_rotary_factor = partial_rotary_factor
self.use_qkv_bias = use_qkv_bias
self.output_router_logits = output_router_logits
self.router_aux_loss_coef = router_aux_loss_coef
if self.hidden_size % self.num_attention_heads != 0:
raise ValueError(
"The hidden size is not divisble by the number of attention heads! Make sure to update them!"
)
# Copied from transformers.models.llama.configuration_llama.LlamaConfig._rope_scaling_validation
def _rope_scaling_validation(self):
"""
Validate the `rope_scaling` configuration.
"""
if self.rope_scaling is None:
return
if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
raise ValueError(
"`rope_scaling` must be a dictionary with two fields, `type` and `factor`, " f"got {self.rope_scaling}"
)
rope_scaling_type = self.rope_scaling.get("type", None)
rope_scaling_factor = self.rope_scaling.get("factor", None)
if rope_scaling_type is None or rope_scaling_type not in ["linear", "dynamic"]:
raise ValueError(
f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
)
if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor <= 1.0:
raise ValueError(f"`rope_scaling`'s factor field must be a float > 1, got {rope_scaling_factor}") |