Views
No views yet
1import os
2import torch
3torch.set_default_dtype(torch.bfloat16)
4from transformers import AutoTokenizer, AutoConfig, Cohere2ForCausalLM, AutoModelForCausalLM
5
6model_id = "CohereLabs/tiny-aya-base"
7config = AutoConfig.from_pretrained(model_id)
8
9config.num_hidden_layers=2
10config.layer_types=[
11 "sliding_attention",
12 "full_attention",
13 ]
14config.num_attention_heads=4
15config.hidden_size=4
16config.intermediate_size=5
17
18model = Cohere2ForCausalLM(config)
19tokenizer = AutoTokenizer.from_pretrained(model_id)
20
21output_dir = "./tiny-random-aya-base/"
22os.makedirs(output_dir, exist_ok=True)
23model.save_pretrained(output_dir, safe_serialization=True)
24tokenizer.save_pretrained(output_dir)