| { | |
| "_name_or_path": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse", | |
| "architectures": [ | |
| "NestedLlamaForCausalLM" | |
| ], | |
| "attention_bias": true, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--configuration_celerity.CelerityConfig", | |
| "AutoModel": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--modeling_celerity.CelerityModel", | |
| "AutoModelForCausalLM": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--modeling_celerity.CelerityForCausalLM", | |
| "AutoModelForQuestionAnswering": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--modeling_celerity.CelerityForQuestionAnswering", | |
| "AutoModelForSequenceClassification": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--modeling_celerity.CelerityForSequenceClassification", | |
| "AutoModelForTokenClassification": "melhoushi/slimpj_h1152_d23_gbs76_tpp20.0_lp0.2_linear_linear_reverse--modeling_celerity.CelerityForTokenClassification" | |
| }, | |
| "bos_token_id": 1, | |
| "embeddings_scale": 1.0, | |
| "eos_token_id": 2, | |
| "exit_attention": false, | |
| "exit_decoder_layer": true, | |
| "exit_layer_indices": [ | |
| 3, | |
| 6, | |
| 12, | |
| 17 | |
| ], | |
| "exit_mlp": false, | |
| "head_dim": 128, | |
| "hidden_act": "relu2", | |
| "hidden_size": 1152, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 9216, | |
| "max_position_embeddings": 8192, | |
| "mhsa_residual_scale": 0.037584782608695654, | |
| "mlp_bias": true, | |
| "mlp_residual_scale": 0.021978260869565214, | |
| "model_type": "nested_llama", | |
| "norm_type": "layernorm", | |
| "num_attention_heads": 9, | |
| "num_hidden_layers": 23, | |
| "num_key_value_heads": 9, | |
| "output_exit_layers": [ | |
| 3, | |
| 6, | |
| 12, | |
| 17 | |
| ], | |
| "output_full_model": true, | |
| "output_logits_scale": 0.0008680555555555555, | |
| "position_embedding_type": "alibi", | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "scale_qk_dot_by_d": true, | |
| "tie_exit_lm_head": true, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.48.0", | |
| "tuple_outputs": true, | |
| "use_cache": false, | |
| "use_mlp_gating": false, | |
| "vocab_size": 128256 | |
| } | |