squeezebert/squeezebert-mnli-headless
0211
1{2 "attention_probs_dropout_prob": 0.1,3 "embedding_size": 768,4 "hidden_act": "gelu",5 "hidden_dropout_prob": 0.1,6 "hidden_size": 768,7 "initializer_range": 0.02,8 "intermediate_size": 3072,9 "max_position_embeddings": 512,10 "model_type": "squeezebert",11 "num_attention_heads": 12,12 "num_hidden_layers": 12,13 "type_vocab_size": 2,14 "vocab_size": 30528,15 "q_groups": 4,16 "k_groups": 4,17 "v_groups": 4,18 "post_attention_groups": 1,19 "intermediate_groups": 4,20 "output_groups": 421}22 