| { | |
| "architectures": [ | |
| "OxMiniForCausalLM" | |
| ], | |
| "bias": false, | |
| "block_size": 256, | |
| "dropout": 0.1, | |
| "ffn_multiplier": 4, | |
| "hc_streams": 4, | |
| "kda_to_mla_ratio": 3, | |
| "mla_latent_dim": 32, | |
| "model_type": "oxmini", | |
| "n_embd": 128, | |
| "n_head": 4, | |
| "n_layer": 8, | |
| "rms_norm_eps": 1e-05, | |
| "seed": 1337, | |
| "tie_embeddings": true, | |
| "use_mhc": true, | |
| "use_sinkhorn_mhc": false, | |
| "vocab_size": 66 | |
| } | |