perSLIMmon-8b-base / config.json
pszemraj's picture
Upload PersimmonForCausalLM
c4a534b
{
"_name_or_path": "/content/perSLIMmon-8b",
"architectures": [
"PersimmonForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 71013,
"eos_token_id": 71013,
"hidden_act": "relu2",
"hidden_dropout": 0.6,
"hidden_size": 4096,
"initializer_range": 0.02,
"intermediate_size": 16384,
"layer_norm_eps": 1e-05,
"max_position_embeddings": 16384,
"model_type": "persimmon",
"num_attention_heads": 64,
"num_hidden_layers": 36,
"num_key_value_heads": 64,
"partial_rotary_factor": 0.5,
"pretraining_tp": 1,
"qk_layernorm": true,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 25000.0,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.36.0.dev0",
"use_cache": true,
"vocab_size": 192144
}