| { | |
| "model_type": "hyper_gpt_depth_communicator", | |
| "architectures": [ | |
| "HyperGPTForCausalLM" | |
| ], | |
| "hyper_config": { | |
| "vocab_size": 4096, | |
| "block_size": 512, | |
| "n_layer": 6, | |
| "n_embd": 256, | |
| "n_head": 4, | |
| "n_kv_head": 2, | |
| "head_dim": 64, | |
| "intermediate_size": 640, | |
| "conn_dim": 64, | |
| "conn_head": 4, | |
| "dropout": 0.0, | |
| "rope_theta": 100000.0 | |
| }, | |
| "tokenizer_name": "AxiomicLabs/GPT-S-5M", | |
| "use_block_residuals": true, | |
| "use_depth_gate": false, | |
| "depth_gate_init": 0.0, | |
| "conn_stat_batch": 8, | |
| "conn_stat_tokens": 16, | |
| "auto_map": { | |
| "AutoConfig": "hyper.HyperGPTConfig", | |
| "AutoModelForCausalLM": "hyper.HyperGPTForCausalLM" | |
| } | |
| } |