Spaces:
Sleeping
Sleeping
| from state import Model | |
| GEMMA3_270M = Model( | |
| vocab_size=256000, num_layers=9, hidden_dim=1152, intermediate_size=4608, | |
| weight_tied_embeddings=True, active_experts=2, total_experts=8, is_moe=False | |
| ) | |
| GEMMA3_1B = Model( | |
| vocab_size=262208, num_layers=26, hidden_dim=2304, intermediate_size=9216, | |
| weight_tied_embeddings=True, active_experts=2, total_experts=8, is_moe=False | |
| ) | |
| GEMMA3_4B = Model( | |
| vocab_size=262208, num_layers=28, hidden_dim=3072, intermediate_size=12288, | |
| weight_tied_embeddings=True, active_experts=2, total_experts=8, is_moe=False | |
| ) | |
| GEMMA3_12B = Model( | |
| vocab_size=262208, num_layers=42, hidden_dim=4608, intermediate_size=18432, | |
| weight_tied_embeddings=True, active_experts=2, total_experts=8, is_moe=False | |
| ) | |
| GEMMA3_27B = Model( | |
| vocab_size=262208, num_layers=46, hidden_dim=6144, intermediate_size=24576, | |
| weight_tied_embeddings=True, active_experts=2, total_experts=8, is_moe=False | |
| ) | |
| DEFAULTS = { | |
| "Gemma3 270M": GEMMA3_270M, | |
| "Gemma3 1B": GEMMA3_1B, | |
| "Gemma3 4B": GEMMA3_4B, | |
| "Gemma3 12B": GEMMA3_12B, | |
| "Gemma3 27B": GEMMA3_27B, | |
| } | |