devJy/GXpert-Example-Model1

1{2    "architectures": [3        "Gemma3ForCausalLM"4    ],5    "attention_bias": false,6    "attention_dropout": 0.0,7    "attn_logit_softcapping": null,8    "bos_token_id": 2,9    "cache_implementation": "hybrid",10    "eos_token_id": 106,11    "final_logit_softcapping": null,12    "head_dim": 256,13    "hidden_activation": "gelu_pytorch_tanh",14    "hidden_size": 1152,15    "initializer_range": 0.02,16    "intermediate_size": 6912,17    "max_position_embeddings": 32768,18    "model_type": "gemma3_text",19    "num_attention_heads": 4,20    "num_hidden_layers": 26,21    "num_key_value_heads": 1,22    "pad_token_id": 0,23    "query_pre_attn_scalar": 256,24    "rms_norm_eps": 1e-06,25    "rope_local_base_freq": 10000,26    "rope_scaling": null,27    "rope_theta": 1000000,28    "sliding_window": 512,29    "sliding_window_pattern": 6,30    "torch_dtype": "bfloat16",31    "transformers_version": "4.50.0.dev0",32    "unsloth_fixed": true,33    "unsloth_version": "2025.3.19",34    "use_cache": true,35    "vocab_size": 26214436}