{ "architectures": [ "SmallmWolofForCausalLM" ], "bos_token_id": 1, "d_ff": 384, "d_model": 320, "dropout": 0.1, "dtype": "float32", "eos_token_id": 2, "head_dim": 40, "load_balance_coef": 0.01, "max_seq_len": 512, "model_type": "smallm_wolof", "moe_top_k": 2, "n_experts": 4, "n_heads": 8, "n_kv_heads": 2, "n_layers": 8, "pad_token_id": 0, "transformers_version": "4.57.6", "use_cache": false, "vocab_size": 8192, "auto_map": { "AutoConfig": "configuration_smallm_wolof.SmallmWolofConfig", "AutoModelForCausalLM": "modeling_smallm_wolof.SmallmWolofForCausalLM" } }