{ "_attn_implementation_autoset": true, "_name_or_path": "hf-internal-testing/tiny-random-FalconModel", "activation": "gelu", "alibi": false, "architectures": [ "FalconModel" ], "attention_dropout": 0.0, "attention_probs_dropout_prob": 0.1, "bias": false, "bos_token_id": 11, "eos_token_id": 11, "ffn_hidden_size": 128, "hidden_act": "gelu", "hidden_dropout": 0.0, "hidden_dropout_prob": 0.1, "hidden_size": 32, "initializer_range": 0.02, "intermediate_size": 37, "layer_norm_epsilon": 1e-05, "max_position_embeddings": 512, "model_type": "falcon", "multi_query": true, "new_decoder_architecture": true, "num_attention_heads": 4, "num_hidden_layers": 2, "num_kv_heads": 4, "num_ln_in_parallel_attn": 2, "pad_token_id": 1, "parallel_attn": true, "rope_scaling": null, "rope_theta": 10000.0, "transformers_version": "4.48.2", "type_vocab_size": 16, "use_cache": true, "vocab_size": 1024 }