| { | |
| "model_type": "jora_mini_hybrid", | |
| "architectures": [ | |
| "JoraMiniForCausalLM" | |
| ], | |
| "version": "v11.2-hybrid", | |
| "hidden_size": 512, | |
| "num_hidden_layers": 16, | |
| "intermediate_size": 1536, | |
| "vocab_size": 128000, | |
| "max_position_embeddings": 2048, | |
| "num_attention_heads": 4, | |
| "num_key_value_heads": 1, | |
| "head_dim": 128, | |
| "rope_theta": 100000.0, | |
| "partial_rotary_factor": 0.25, | |
| "hidden_act": "silu", | |
| "rms_norm_eps": 1e-06, | |
| "tie_word_embeddings": true, | |
| "dtype": "bfloat16", | |
| "eos_token_id": 1, | |
| "total_steps": 114112, | |
| "final_loss": 0.6274442672729492, | |
| "tokens_seen": 15007219712 | |
| } |