Files
gpt/configs/gpt4_5_config.py

26 lines
695 B
Python

"""
Configuration for gpt4_5 model (3T parameters)
"""
class Gpt4_5Config(object):
"""Configuration for GPT-4.5 model"""
def __init__(self):
# Model architecture
self.block_size = 512000
self.vocab_size = 50304
self.n_layer = 240
self.n_head = 240
self.n_embd = 20480
self.dropout = 0.0
self.bias = False # False is slightly better
# Parameter metrics
self.parameter_count = "3T"
# Mixture of Experts
self.use_moe = True
self.num_experts = 240
self.num_active_experts = 12
self.expert_capacity_factor = 1.25
self.moe_jitter_eps = 0.01