config.json

{
    "architectures": [
        "SmolLM3ForCausalLM"
    ],
    "attention_bias": false,
    "attention_dropout": 0.0,
    "bos_token_id": null,
    "eos_token_id": 128001,
    "hidden_act": "silu",
    "hidden_size": 2048,
    "initializer_range": 0.02,
    "intermediate_size": 11008,
    "max_position_embeddings": 65536,
    "model_type": "smollm3",
    "no_rope_layer_interval": 4,
    "num_attention_heads": 16,
    "num_hidden_layers": 36,
    "num_key_value_heads": 4,
    "pad_token_id": 128004,
    "pretraining_tp": 2,
    "quantization": {
        "group_size": 64,
        "bits": 3
    },
    "quantization_config": {
        "group_size": 64,
        "bits": 3
    },
    "rms_norm_eps": 1e-06,
    "rope_scaling": null,
    "rope_theta": 5000000.0,
    "sliding_window": null,
    "tie_word_embeddings": true,
    "torch_dtype": "bfloat16",
    "transformers.js_config": {
        "use_external_data_format": true,
        "dtype": "q4",
        "kv_cache_dtype": {
            "q4f16": "float16",
            "fp16": "float16"
        }
    },
    "transformers_version": "4.53.0.dev0",
    "use_cache": true,
    "use_sliding_window": false,
    "vocab_size": 128256
}