smollm3 / config.json
yujiepan's picture
Upload folder using huggingface_hub
bcb0ea2 verified
raw
history blame contribute delete
892 Bytes
{
"architectures": [
"SmolLM3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 128000,
"eos_token_id": 128012,
"hidden_act": "silu",
"hidden_size": 64,
"initializer_range": 0.02,
"intermediate_size": 128,
"layer_types": [
"full_attention",
"sliding_attention"
],
"max_position_embeddings": 65536,
"max_window_layers": 28,
"mlp_bias": false,
"model_type": "smollm3",
"no_rope_layer_interval": 2,
"no_rope_layers": [
1,
0
],
"num_attention_heads": 2,
"num_hidden_layers": 2,
"num_key_value_heads": 1,
"pad_token_id": 128004,
"pretraining_tp": 2,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 5000000.0,
"sliding_window": 128,
"torch_dtype": "bfloat16",
"transformers_version": "4.54.0.dev0",
"use_cache": true,
"use_sliding_window": true,
"vocab_size": 128256
}