karasu-aya-dareties2 is a merge of the following models using
LazyMergekit:
1slices:
2- sources:
3 - layer_range: [0, 22]
4 model: yuiseki/karasu-aya-ja-v0.1
5 parameters:
6 density: [1, 0.7, 0.1]
7 weight: 1.0
8 - layer_range: [0, 22]
9 model: aipib/karasu-slerp1
10 parameters:
11 density: 0.33
12 weight:
13 - filter: mlp
14 value: 0.5
15 - value: 0
16merge_method: dare_ties
17base_model: aipib/karasu-slerp1
18parameters:
19 normalize: true
20 int8_mask: true
21dtype: bfloat16
1!pip install -qU transformers accelerate
2
3from transformers import AutoTokenizer
4import transformers
5import torch
6
7model = "aipib/karasu-aya-dareties2"
8messages = [{"role": "user", "content": "What is a large language model?"}]
9
10tokenizer = AutoTokenizer.from_pretrained(model)
11prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
12pipeline = transformers.pipeline(
13 "text-generation",
14 model=model,
15 torch_dtype=torch.float16,
16 device_map="auto",
17)
18
19outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
20print(outputs[0]["generated_text"])