simplified_phi2 / config.json
BucketOfFish's picture
Fixed weight loading from original Phi2 model
10aca20
raw
history blame
890 Bytes
{
"_name_or_path": "BucketOfFish/simplified_phi2",
"architectures": [
"Phi2Model",
"Phi2ModelForCausalLM"
],
"auto_map": {
"AutoConfig": "phi2_configuration.Phi2Config",
"AutoModel": "phi2_model.Phi2Model",
"AutoModelForCausalLM": "phi2_model.Phi2ModelForCausalLM"
},
"model_type": "phi2",
"torch_dtype": "float16",
"transformers_version": "4.29.0",
"vocab_size": 51200,
"vocab_chunk_for_gpu_efficiency": 64,
"initial_cos_sin_cache_len": 2048,
"d_embedding": 2560,
"n_blocks": 32,
"n_heads": 32,
"use_flash_attn": false,
"use_flash_rotary": false,
"use_fused_dense": false,
"attn_pdrop": 0.0,
"embd_pdrop": 0.0,
"resid_pdrop": 0.1,
"layer_norm_epsilon": 1e-05,
"weight_initialization_range": 0.02,
"tie_word_embeddings": false,
"checkpointing": false
}