| { | |
| "base_model_name_or_path": "meta-llama/Llama-3.1-8B-Instruct", | |
| "compression": { | |
| "disabled_ids": null, | |
| "initial_vocab_size": 128256, | |
| "max_codebook_size": 2048, | |
| "max_subtokens": 4 | |
| }, | |
| "encoder": { | |
| "hidden_size": 4096, | |
| "intermediate_size": null, | |
| "num_heads": 32, | |
| "num_hidden_layers": 2, | |
| "position_encoding": "learnable", | |
| "tie_encoders": true | |
| }, | |
| "encoder_type": "transformer" | |
| } |