trl-internal-testing/tiny-Qwen3MoeForSequenceClassification

1{2  "architectures": [3    "Qwen3MoeForSequenceClassification"4  ],5  "attention_bias": false,6  "attention_dropout": 0.0,7  "bos_token_id": 151643,8  "decoder_sparse_step": 1,9  "dtype": "bfloat16",10  "eos_token_id": 151645,11  "head_dim": 128,12  "hidden_act": "silu",13  "hidden_size": 16,14  "id2label": {15    "0": "LABEL_0"16  },17  "initializer_range": 0.02,18  "intermediate_size": 32,19  "label2id": {20    "LABEL_0": 021  },22  "max_position_embeddings": 40960,23  "max_window_layers": 48,24  "mlp_only_layers": [],25  "model_type": "qwen3_moe",26  "moe_intermediate_size": 768,27  "norm_topk_prob": true,28  "num_attention_heads": 4,29  "num_experts": 4,30  "num_experts_per_tok": 2,31  "num_hidden_layers": 2,32  "num_key_value_heads": 2,33  "output_router_logits": false,34  "rms_norm_eps": 1e-06,35  "rope_scaling": null,36  "rope_theta": 1000000.0,37  "router_aux_loss_coef": 0.001,38  "sliding_window": null,39  "tie_word_embeddings": false,40  "transformers_version": "4.57.3",41  "use_cache": true,42  "use_sliding_window": false,43  "vocab_size": 15193644}45