初始化项目,由ModelHub XC社区提供模型

Model: emese-tech/csermely
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-07-24 15:59:09 +08:00
commit acb8b97699
8 changed files with 200 additions and 0 deletions

26
config.json Normal file
View File

@@ -0,0 +1,26 @@
{
"architectures": ["LlamaForCausalLM"],
"model_type": "llama",
"model_version": "0.2",
"vocab_size": 32000,
"hidden_size": 896,
"intermediate_size": 2560,
"num_hidden_layers": 16,
"num_attention_heads": 14,
"num_key_value_heads": 14,
"hidden_act": "silu",
"max_position_embeddings": 4096,
"rope_theta": 10000.0,
"rope_scaling": {
"type": "yarn",
"factor": 4.0,
"original_max_position_embeddings": 1024
},
"rms_norm_eps": 1e-5,
"tie_word_embeddings": true,
"torch_dtype": "float16",
"bos_token_id": 2,
"eos_token_id": 3,
"pad_token_id": 1,
"transformers_version": "4.45.0"
}