| { | |
| "_name_or_path": "outputs/tinyllama-lckv-w10-100b", | |
| "architectures": [ | |
| "LCKVLlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_lckv.LCKVLlamaConfig", | |
| "AutoModelForCausalLM": "modeling_lckv.LCKVLlamaForCausalLM" | |
| }, | |
| "backward_passes": 2, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "force_nodiag": true, | |
| "forward_passes": 7, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 5632, | |
| "layer_types": "0_1_2_3_4_16_16_16_16_16_16_16_16_16_16_16_16_17_18_19_20_21", | |
| "max_position_embeddings": 2048, | |
| "mlp_bias": false, | |
| "model_type": "lckv-llama", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 22, | |
| "num_key_value_heads": 4, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "sliding_window": 4096, | |
| "tie_word_embeddings": false, | |
| "tokenizer_class": "LlamaTokenizer", | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.45.2", | |
| "use_cache": true, | |
| "use_sequential": false, | |
| "vocab_size": 32000 | |
| } | |