|
{ |
|
"activation_function": "silu", |
|
"add_bias_linear": false, |
|
"attn_pdrop": 0.0, |
|
"architectures":[ |
|
"TeleChatForCausalLM" |
|
], |
|
"auto_map": { |
|
"AutoConfig": "configuration_telechat.TELECHATConfig", |
|
"AutoModel": "modeling_telechat.TELECHAT", |
|
"AutoModelForCausalLM": "modeling_telechat.TELECHAT" |
|
}, |
|
"bos_token_id": 1, |
|
"embd_pdrop": 0.0, |
|
"enable_flash_attn": true, |
|
"eos_token_id": 2, |
|
"initializer_range": 0.02, |
|
"input_mult": 1.0, |
|
"layer_norm_epsilon": 1e-05, |
|
"model_type": "telechat", |
|
"mup_base_width": 256, |
|
"mup_scale_factor": 32.0, |
|
"n_embd": 8192, |
|
"n_head": 64, |
|
"n_inner": 21824, |
|
"n_layer": 64, |
|
"n_positions": 4096, |
|
"output_mult": 1.0, |
|
"pad_token_id": 3, |
|
"relative_encoding": "rotary", |
|
"reorder_and_upcast_attn": true, |
|
"resid_pdrop": 0.0, |
|
"rotary_theta": 10000, |
|
"rotary_use_xpos": false, |
|
"rotary_xpos_scale_base": 512, |
|
"scale_attn_by_inverse_layer_idx": true, |
|
"scale_attn_weights": true, |
|
"summary_activation": null, |
|
"summary_first_dropout": 0.1, |
|
"summary_proj_to_labels": true, |
|
"summary_type": "cls_index", |
|
"summary_use_proj": true, |
|
"tie_word_embeddings": false, |
|
"tokenizer_class": "TELECHATTokenizer", |
|
"torch_dtype": "bfloat16", |
|
"transformers_version": "4.34.1", |
|
"unk_token_id": 0, |
|
"use_RMSNorm": true, |
|
"use_cache": true, |
|
"use_mup": true, |
|
"vocab_size": 80896 |
|
} |
|
|