Llama-3.1-Tango-70b / config.json
tatakof's picture
Create config.json
d3776c1 verified
raw
history blame contribute delete
650 Bytes
{
"_name_or_path": "sandbox-ai/Tango-70b",
"architectures": [
"LlamaForCausalLM"
],
"model_type": "llama",
"attention_bias": false,
"attention_dropout": 0.0,
"hidden_act": "silu",
"hidden_size": 8192,
"initializer_range": 0.02,
"intermediate_size": 28672,
"max_position_embeddings": 131072,
"mlp_bias": false,
"num_attention_heads": 64,
"num_hidden_layers": 80,
"num_key_value_heads": 8,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_theta": 500000.0,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"use_cache": true,
"vocab_size": 128256
}