| { | |
| "architectures": [ | |
| "TestGeniyForCausalLM" | |
| ], | |
| "model_type": "testgeniy", | |
| "vocab_size": 0, | |
| "hidden_size": 1280, | |
| "num_hidden_layers": 24, | |
| "num_attention_heads": 10, | |
| "num_key_value_heads": 2, | |
| "intermediate_size": 3584, | |
| "head_dim": 128, | |
| "max_position_embeddings": 4096, | |
| "rope_theta": 500000.0, | |
| "local_block_size": 1024, | |
| "layer_types": [ | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention" | |
| ], | |
| "base_checkpoint": "logic_small_scope_step080", | |
| "weights_updated": false | |
| } |