Ziya-Writing-13B-v2 / config.json
pskun's picture
9c1be5fd685ec26384b25e879f1ce91702078edc7bf5a6cc193a83f27613fab4
a60a544
raw history blame
No virus
730 Bytes
{
"_name_or_path": "model_ckpt/rrhf_0926",
"architectures": [
"LlamaForCausalLM"
],
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 5120,
"initializer_range": 0.02,
"intermediate_size": 13824,
"layer_norm_eps": 1e-05,
"max_position_embeddings": 8192,
"model_type": "llama",
"num_attention_heads": 40,
"num_hidden_layers": 40,
"num_key_value_heads": 40,
"pad_token_id": 0,
"pretraining_tp": 1,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rotary_emb_base": 10000,
"rotary_pct": 1,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.32.0",
"use_cache": false,
"use_parallel_residual": false,
"vocab_size": 39424
}