36 lines
1003 B
JSON
36 lines
1003 B
JSON
{
|
|
"_name_or_path": "/run/media/knut/HD/huggingface models/language models/llama-alternatives/LLongMA-3b/",
|
|
"architectures": [
|
|
"LlamaForCausalLM"
|
|
],
|
|
"auto_map": {
|
|
"AutoConfig": "configuration_llama.LlamaConfig",
|
|
"AutoModel": "modeling_llama.LlamaModel",
|
|
"AutoModelForCausalLM": "modeling_llama.LlamaForCausalLM",
|
|
"AutoModelForSequenceClassification": "modeling_llama.LlamaForSequenceClassification"
|
|
},
|
|
"bos_token_id": 1,
|
|
"eos_token_id": 2,
|
|
"hidden_act": "silu",
|
|
"hidden_size": 3200,
|
|
"initializer_range": 0.02,
|
|
"intermediate_size": 8640,
|
|
"max_position_embeddings": 8192,
|
|
"model_type": "llama",
|
|
"num_attention_heads": 32,
|
|
"num_hidden_layers": 26,
|
|
"num_key_value_heads": 32,
|
|
"pad_token_id": 0,
|
|
"pretraining_tp": 1,
|
|
"rms_norm_eps": 1e-06,
|
|
"rope_scaling": {
|
|
"factor": 4.0,
|
|
"type": "linear"
|
|
},
|
|
"tie_word_embeddings": false,
|
|
"torch_dtype": "float32",
|
|
"transformers_version": "4.32.0.dev0",
|
|
"use_cache": true,
|
|
"vocab_size": 32000
|
|
}
|