trl-internal-testing/tiny-CohereForCausalLM

1{2  "architectures": [3    "CohereForCausalLM"4  ],5  "attention_bias": false,6  "attention_dropout": 0.0,7  "bos_token_id": 5,8  "dtype": "float16",9  "eos_token_id": 255001,10  "hidden_act": "silu",11  "hidden_size": 8,12  "initializer_range": 0.02,13  "intermediate_size": 32,14  "layer_norm_eps": 1e-05,15  "logit_scale": 0.125,16  "max_position_embeddings": 8192,17  "model_type": "cohere",18  "num_attention_heads": 4,19  "num_hidden_layers": 2,20  "num_key_value_heads": 2,21  "pad_token_id": 0,22  "rope_scaling": null,23  "rope_theta": 10000.0,24  "transformers_version": "4.56.2",25  "use_cache": true,26  "use_qk_norm": false,27  "vocab_size": 25600028}29