| { |
| "architectures": [ |
| "Model808AIForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "attn_logit_softcapping": null, |
| "bos_token_id": 0, |
| "dtype": "float32", |
| "eos_token_id": 0, |
| "head_dim": 64, |
| "hidden_size": 768, |
| "initializer_range": 0.02, |
| "intermediate_size": 2048, |
| "max_position_embeddings": 512, |
| "model_type": "808ai", |
| "num_attention_heads": 12, |
| "num_hidden_layers": 20, |
| "num_key_value_heads": 4, |
| "pad_token_id": 0, |
| "rms_norm_eps": 1e-06, |
| "rope_theta": 10000.0, |
| "sliding_window": 512, |
| "sliding_window_pattern": 2, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.16.1", |
| "use_cache": true, |
| "use_sliding_window": true, |
| "vocab_size": 50257 |
| } |
|
|
