| { | |
| "_name_or_path": "/home/mkurman/gitlab/ai/model-trainer/results/one_2/starter_140", | |
| "architectures": [ | |
| "OneForCausalLM" | |
| ], | |
| "hidden_size": 1024, | |
| "initializer_factor": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_norm_eps": 1e-05, | |
| "max_model_length": 131072, | |
| "max_position_embeddings": 131072, | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 6, | |
| "pad_token_id": 49152, | |
| "torch_dtype": "float32", | |
| "transformer_dim": 512, | |
| "transformers_version": "4.49.0", | |
| "type_vocab_size": 2, | |
| "use_flash_attention": false, | |
| "use_single_token": false, | |
| "vocab_size": 49153 | |
| } | |