| { |
| "_name_or_path": "/work/parsa/checkpoints/Smollm_1B_FT_sorted/checkpoint-30000/", |
| "architectures": [ |
| "NestedLlamaForCausalLM" |
| ], |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "auto_map": { |
| "AutoConfig": "configuration_nestedllama.NestedLlamaConfig", |
| "AutoModel": "modeling_nestedllama.NestedLlamaModel", |
| "AutoModelForCausalLM": "modeling_nestedllama.NestedLlamaForCausalLM" |
| }, |
| "bos_token_id": 0, |
| "eos_token_id": 0, |
| "exit_attention": false, |
| "exit_decoder_layer": false, |
| "exit_layer_indices": [ |
| 4, |
| 8, |
| 12 |
| ], |
| "exit_mlp": false, |
| "head_dim": 64, |
| "hidden_act": "silu", |
| "hidden_size": 2048, |
| "initializer_range": 0.02, |
| "intermediate_size": 8192, |
| "max_position_embeddings": 2048, |
| "mlp_bias": false, |
| "model_type": "nested_llama", |
| "num_attention_heads": 32, |
| "num_hidden_layers": 16, |
| "num_key_value_heads": 32, |
| "output_exit_layer": 12, |
| "output_exit_layer_index": null, |
| "output_exit_layers": [ |
| 4, |
| 8, |
| 12 |
| ], |
| "output_full_model": true, |
| "pretraining_tp": 1, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": null, |
| "rope_theta": 10000.0, |
| "tie_exit_lm_head": true, |
| "tie_word_embeddings": true, |
| "torch_dtype": "float32", |
| "transformers_version": "4.48.0", |
| "use_cache": true, |
| "vocab_size": 49152 |
| } |
|
|
