| { |
| "architectures": [ |
| "OpenThaiWilaiForCausalLM" |
| ], |
| "attention_dropout": 0.0, |
| "bos_token_id": 2, |
| "dtype": "float32", |
| "eos_token_id": 3, |
| "ffn_dropout": 0.0, |
| "hidden_act": "silu", |
| "hidden_dropout_prob": 0.0, |
| "hidden_size": 768, |
| "initializer_range": 0.02, |
| "intermediate_size": 3072, |
| "layer_norm_eps": 1e-05, |
| "logit_scale": 1.0, |
| "max_position_embeddings": 2048, |
| "mixture_of_depths_layers": null, |
| "model_type": "OpenThaiWilai", |
| "moe_aux_loss_coef": 0.01, |
| "num_experts": 0, |
| "num_heads": 12, |
| "num_layers": 12, |
| "pad_token_id": 0, |
| "rezero": false, |
| "rope": true, |
| "rope_scaling": null, |
| "rope_theta": 10000.0, |
| "sliding_window_size": 4096, |
| "stochastic_depth_prob": 0.0, |
| "transformers_version": "4.56.1", |
| "use_alibi": false, |
| "use_flash_attn": true, |
| "use_mixture_of_depths": false, |
| "use_multimodal": false, |
| "use_parallel_residual": false, |
| "use_reasoning_tokens": false, |
| "use_retrieval_augmented": false, |
| "use_rope": true, |
| "use_sliding_window": false, |
| "vocab_size": 48000 |
| } |
|
|