Edit model card

yujiepan/llama-2-tiny-3layers-random

This model is randomly initialized, using the config from meta-llama/Llama-2-7b-chat-hf but with the following modifications:

{
  "hidden_size": 8,
  "intermediate_size": 32,
  "num_attention_heads": 2,
  "num_hidden_layers": 3,
  "num_key_value_heads": 2,
}
Downloads last month
1,912
Safetensors
Model size
515k params
Tensor type
F32
·
FP16
·

Collection including yujiepan/llama-2-tiny-3layers-random