hf-tiny-v2/tiny-random-DiffLlamaForCausalLM
05
1{2 "architectures": [3 "DiffLlamaForCausalLM"4 ],5 "attention_bias": false,6 "attention_dropout": 0.0,7 "attention_probs_dropout_prob": 0.1,8 "bos_token_id": 128000,9 "dtype": "float32",10 "eos_token_id": 128009,11 "head_dim": 8,12 "hidden_act": "gelu",13 "hidden_dropout_prob": 0.1,14 "hidden_size": 32,15 "initializer_range": 0.02,16 "intermediate_size": 37,17 "is_decoder": true,18 "lambda_std_dev": 0.1,19 "max_position_embeddings": 512,20 "model_type": "diffllama",21 "num_attention_heads": 4,22 "num_hidden_layers": 2,23 "num_key_value_heads": 4,24 "pad_token_id": 128004,25 "rms_norm_eps": 1e-05,26 "rope_parameters": {27 "rope_theta": 10000.0,28 "rope_type": "default"29 },30 "tie_word_embeddings": false,31 "transformers_version": "5.16.0.dev0",32 "type_vocab_size": 16,33 "use_cache": true,34 "vocab_size": 12825635}36 