vdos's picture
Training in progress, step 25
149b3a2 verified
raw
history blame contribute delete
817 Bytes
{
"_attn_implementation_autoset": true,
"_name_or_path": "Xenova/tiny-random-Phi3ForCausalLM",
"architectures": [
"Phi3ForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"embd_pdrop": 0.0,
"eos_token_id": 32000,
"hidden_act": "silu",
"hidden_size": 32,
"initializer_range": 0.02,
"intermediate_size": 64,
"max_position_embeddings": 4096,
"model_type": "phi3",
"num_attention_heads": 4,
"num_hidden_layers": 2,
"num_key_value_heads": 4,
"original_max_position_embeddings": 4096,
"pad_token_id": 32000,
"resid_pdrop": 0.0,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 10000.0,
"sliding_window": 2047,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.46.3",
"use_cache": false,
"vocab_size": 32064
}