tiny-random/smollm3
0623
1{2 "architectures": [3 "SmolLM3ForCausalLM"4 ],5 "attention_bias": false,6 "attention_dropout": 0.0,7 "bos_token_id": 128000,8 "eos_token_id": 128012,9 "hidden_act": "silu",10 "hidden_size": 64,11 "initializer_range": 0.02,12 "intermediate_size": 128,13 "layer_types": [14 "full_attention",15 "sliding_attention"16 ],17 "max_position_embeddings": 65536,18 "max_window_layers": 28,19 "mlp_bias": false,20 "model_type": "smollm3",21 "no_rope_layer_interval": 2,22 "no_rope_layers": [23 1,24 025 ],26 "num_attention_heads": 2,27 "num_hidden_layers": 2,28 "num_key_value_heads": 1,29 "pad_token_id": 128004,30 "pretraining_tp": 2,31 "rms_norm_eps": 1e-06,32 "rope_scaling": null,33 "rope_theta": 5000000.0,34 "sliding_window": 128,35 "torch_dtype": "bfloat16",36 "transformers_version": "4.54.0.dev0",37 "use_cache": true,38 "use_sliding_window": true,39 "vocab_size": 12825640}41 