trl-internal-testing/tiny-RemoteForCausalLM
0310k
1{2 "architectures": [3 "RemoteForCausalLM"4 ],5 "attention_bias": false,6 "attention_dropout": 0.0,7 "auto_map": {8 "AutoConfig": "configuration_remote.RemoteConfig",9 "AutoModel": "modeling_remote.RemoteModel",10 "AutoModelForCausalLM": "modeling_remote.RemoteForCausalLM",11 "AutoModelForSequenceClassification": "modeling_remote.RemoteForSequenceClassification"12 },13 "bos_token_id": 128000,14 "dtype": "bfloat16",15 "eos_token_id": [16 128001,17 128008,18 12800919 ],20 "head_dim": 2,21 "hidden_act": "silu",22 "hidden_size": 8,23 "initializer_range": 0.02,24 "intermediate_size": 32,25 "max_position_embeddings": 131072,26 "mlp_bias": false,27 "model_type": "remote",28 "num_attention_heads": 4,29 "num_hidden_layers": 2,30 "num_key_value_heads": 2,31 "pretraining_tp": 1,32 "rms_norm_eps": 1e-05,33 "rope_scaling": {34 "factor": 32.0,35 "high_freq_factor": 4.0,36 "low_freq_factor": 1.0,37 "original_max_position_embeddings": 8192,38 "rope_type": "llama3"39 },40 "rope_theta": 500000.0,41 "tie_word_embeddings": true,42 "transformers_version": "4.56.2",43 "use_cache": true,44 "vocab_size": 12825645}46 