46 lines
1.2 KiB
JSON
46 lines
1.2 KiB
JSON
|
{
|
||
|
|
"_name_or_path": "/data/Telechat/TeleChat2/TeleChat2-7B",
|
||
|
|
"apply_residual_connection_post_layernorm": false,
|
||
|
|
"architectures": [
|
||
|
|
"TeleChat2ForCausalLM"
|
||
|
|
],
|
||
|
|
"attention_dropout": 0.0,
|
||
|
|
"attention_softmax_in_fp32": true,
|
||
|
|
"auto_map": {
|
||
|
|
"AutoConfig": "configuration_telechat2.Telechat2Config",
|
||
|
|
"AutoModelForCausalLM": "modeling_telechat2.Telechat2ForCausalLM"
|
||
|
|
},
|
||
|
|
"base_seqlen": 8192,
|
||
|
|
"bias_dropout_fusion": true,
|
||
|
|
"bos_token_id": 1,
|
||
|
|
"embed_layernorm": false,
|
||
|
|
"eos_token_id": 2,
|
||
|
|
"ffn_hidden_size": 12288,
|
||
|
|
"flash_attn": false,
|
||
|
|
"hidden_dropout": 0.0,
|
||
|
|
"hidden_size": 4096,
|
||
|
|
"initializer_range": 0.02,
|
||
|
|
"layer_norm_epsilon": 1e-05,
|
||
|
|
"logn": true,
|
||
|
|
"masked_softmax_fusion": true,
|
||
|
|
"model_type": "telechat",
|
||
|
|
"n_head": 32,
|
||
|
|
"n_inner": null,
|
||
|
|
"n_layer": 30,
|
||
|
|
"num_key_value_heads": 32,
|
||
|
|
"offset_alibi": 100,
|
||
|
|
"pad_token_id": 3,
|
||
|
|
"pretraining_tp": 1,
|
||
|
|
"seq_length": 32768,
|
||
|
|
"skip_bias_add": true,
|
||
|
|
"skip_bias_add_qkv": false,
|
||
|
|
"slow_but_exact": false,
|
||
|
|
"tie_word_embeddings": false,
|
||
|
|
"torch_dtype": "float16",
|
||
|
|
"training_seqlen": 8192,
|
||
|
|
"transformers_version": "4.47.0",
|
||
|
|
"unk_token_id": 0,
|
||
|
|
"use_cache": true,
|
||
|
|
"vocab_size": 131072
|
||
|
|
}
|