| { | |
| "vocab_size": 32768, | |
| "max_seq_len": 4096, | |
| "n_layer": 20, | |
| "n_embd": 256, | |
| "n_head": 4, | |
| "n_kv_head": 2, | |
| "mlp_hidden": 768, | |
| "rope_base": 10000.0, | |
| "rms_eps": 1e-06, | |
| "qk_norm": true, | |
| "z_loss": 0.0, | |
| "model_type": "gouda-ricotta", | |
| "architectures": [ | |
| "RicottaForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_ricotta.RicottaConfig", | |
| "AutoModelForCausalLM": "modeling_ricotta.RicottaForCausalLM" | |
| }, | |
| "name": "Ricotta-2.0", | |
| "torch_dtype": "bfloat16", | |
| "tie_word_embeddings": true | |
| } |