hf-transformers-bot's picture
Update tiny models for Qwen2MoeForTokenClassification
0f472e9 verified
Raw
History Blame Contribute Delete
1.09 kB
{
"architectures": [
"Qwen2MoeForTokenClassification"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"decoder_sparse_step": 1,
"dtype": "float32",
"eos_token_id": 151643,
"hidden_act": "gelu",
"hidden_size": 32,
"initializer_range": 0.02,
"intermediate_size": 32,
"layer_types": [
"full_attention",
"full_attention"
],
"max_position_embeddings": 512,
"max_window_layers": 28,
"mlp_only_layers": [],
"model_type": "qwen2_moe",
"moe_intermediate_size": 16,
"norm_topk_prob": false,
"num_attention_heads": 2,
"num_experts": 8,
"num_experts_per_tok": 2,
"num_hidden_layers": 2,
"num_key_value_heads": 2,
"output_router_logits": false,
"pad_token_id": 151643,
"qkv_bias": true,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 10000.0,
"rope_type": "default"
},
"router_aux_loss_coef": 0.001,
"shared_expert_intermediate_size": 36,
"sliding_window": 0,
"tie_word_embeddings": false,
"transformers_version": "5.16.0.dev0",
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 151646
}