tomtseng's picture
Upload from /robust_llm_data/checkpoints/tom_035_ft_qwen25_large_imdb/tom-035-ft-qwen25-large-imdb-0005/a14e58d465ab654a5558fa506f3774bfef7e20e92b5ba0c651855c9d4d12a5db/epoch_0003/model by /root/robust-llm/scripts/oskar/push_checkpoints_to_hub.py
d488f0e verified
raw
history blame contribute delete
800 Bytes
{
"_name_or_path": "Qwen/Qwen2.5-14B",
"architectures": [
"Qwen2ForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 151643,
"eos_token_id": 151643,
"hidden_act": "silu",
"hidden_size": 5120,
"initializer_range": 0.02,
"intermediate_size": 13824,
"max_position_embeddings": 131072,
"max_window_layers": 48,
"model_type": "qwen2",
"num_attention_heads": 40,
"num_hidden_layers": 48,
"num_key_value_heads": 8,
"pad_token_id": 151643,
"problem_type": "single_label_classification",
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 1000000.0,
"sliding_window": null,
"tie_word_embeddings": false,
"torch_dtype": "float32",
"transformers_version": "4.46.2",
"use_cache": false,
"use_sliding_window": false,
"vocab_size": 152064
}