DCFormer-2.8B / config.json
mqyqlx
upload model and code
29a399e
raw
history blame contribute delete
No virus
751 Bytes
{
"architectures": [
"DCFormer"
],
"auto_map": {
"AutoConfig": "configuration_dcformer.DCFormerConfig",
"AutoModelForCausalLM": "modeling_dcformer.DCFormer"
},
"block_size": 2048,
"bos_token_id": 1,
"dim": 2560,
"eos_token_id": 2,
"head_dim": 80,
"intermediate_size": 6912,
"is_training": false,
"model_type": "dcformer",
"n_head": 32,
"n_layer": 32,
"n_local_heads": 32,
"norm_eps": 1e-06,
"q_chunk_size": 128,
"query_wise": false,
"rope_base": 10000,
"tie_word_embeddings": false,
"torch_dtype": "float16",
"transformers_version": "4.33.2",
"use_dcmha": true,
"use_gradient_checkpointing": false,
"use_qk_norm": true,
"vocab_size": 50257,
"window_size": 256,
"window_type": null
}